A Method for Catenary Fault Location and Adaptive Map Display Based on Fault Text Location Parsing and Spatial Constraints
By using a method based on fault text location parsing and spatial constraints, unstructured fault location text is transformed into a structured model, solving the problem of inaccurate contact network fault location and realizing automated fault map display and rapid location.
Patent Information
- Application Number
- CN202611620621.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-20
- Publication Date
- 2026-11-17
AI Technical Summary
In existing technologies, it is difficult to automatically convert the text of the fault location of the overhead contact line into structured spatial conditions, resulting in inaccurate fault location and long handling cycle. It is also difficult to integrate line operation semantics and power supply operation and maintenance semantics in the same model, and there is a lack of unified spatial expression and object linkage.
By using a method based on fault text location parsing and spatial constraints, unstructured fault location text is transformed into a structured model through knowledge extraction and processing. Conflict resolution analysis is then performed to generate a fault map containing map spatial objects and surrounding objects, enabling automated location and display.
It improves the accuracy and efficiency of fault location, avoids the need for manual text interpretation and map searching, and significantly shortens the fault handling cycle.
Smart Images

Figure CN123115441A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of knowledge extraction technology, specifically to a method for catenary fault location and adaptive map display based on fault text location parsing and spatial constraints. Background Technology
[0002] The overhead contact line is a critical component of the railway traction power supply system, and is prone to faults such as insulator flashover, line breakage, and foreign object intrusion during operation. Fault locations are typically entered in business text format, such as "Uplink XX Station K23+400" or "XX Substation 213 Feeder 23.400," which contains a mixture of elements including line name, track type, station or section, kilometer marker, substation, and feeder number. Separators, abbreviations, and kilometer marker styles vary depending on personnel and system. With increasing demands for timeliness in fault handling during power supply maintenance, it is necessary to automatically convert these faults into structured spatial conditions that can be used for map positioning.
[0003] However, related technologies mostly rely on text storage and manual interpretation to utilize fault location text, requiring experience to search and zoom in on the map for location, and easily mistaking feeder numbers and other numbers for kilometer markers; location elements also have two expression calibers, namely line operation semantics and power supply maintenance semantics, which are difficult to integrate and verify in the same model, making it difficult to form a unified spatial expression, and the spatial linkage between surrounding objects such as pillars, key equipment, and work areas and the fault location requires manual layer-by-layer screening, resulting in incomplete fault judgment context and long handling cycle. Summary of the Invention
[0004] This application provides a method for catenary fault location and adaptive map display based on fault text location parsing and spatial constraints. It can automatically convert unstructured fault location text into calculable spatial conditions, improve the accuracy of fault location, and achieve adaptive focused display with fault location as the core.
[0005] To achieve the above objectives, the technical solution adopted in this application is as follows: In a first aspect, this application provides a method for catenary fault location and adaptive map display based on fault text location parsing and spatial constraints. The method includes: in response to receiving a fault event, acquiring the original fault location text and basic context information; performing knowledge extraction on the original fault location text to determine fault location elements; performing knowledge processing based on the fault location elements to determine a fault location model, wherein the fault location model is a structured model obtained by classifying and filling the fault location elements into fields of line, power supply, mileage, spatial representation, and credibility; performing conflict resolution analysis on the fault location elements based on the fault location model to determine target fault location information containing single points or interval mileage; determining a fault map corresponding to the fault event based on the target fault location information, wherein the fault map includes at least map spatial objects and surrounding objects; and displaying the fault map.
[0006] Based on the above technical means, unstructured fault location text is transformed into a five-dimensional structured model through knowledge extraction and processing. Conflict resolution analysis is then performed based on the model to obtain unique and reliable target fault location information. Based on this, a fault map containing map spatial objects and surrounding objects is generated and displayed. This achieves full automation from fault text to map location display, avoiding manual text interpretation and map searching, and significantly improving the efficiency of fault location and judgment.
[0007] One possible implementation method for determining fault location elements by performing knowledge extraction on the original fault location text can be specifically implemented as follows: performing text standardization preprocessing on the original fault location text to obtain standardized text; loading a line operation semantic template and a power supply operation and maintenance semantic template; the line operation semantic template includes combination rules for line name, line type or track, station or section, and kilometer marker; the power supply operation and maintenance semantic template includes combination rules for substation, feeder number, and kilometer marker; based on the standardized text and the basic context information, performing structured parsing on the standardized text using the line operation semantic template and the power supply operation and maintenance semantic template to extract fault location elements.
[0008] Based on the above technical means, the differences between separators and kilometer marker formats are eliminated through text standardization preprocessing, and the system is compatible with both line operation and power supply maintenance methods through parallel matching of dual semantic templates. This avoids the problem that a single-caliber system cannot recognize the other type of text and expands the coverage of fault texts that can be automatically located.
[0009] One possible implementation, whereby the knowledge processing based on the fault location elements is used to determine the fault location model, can be specifically implemented as follows: Initialize an empty model carrier and create a line dimension container, a power supply dimension container, a mileage dimension container, a spatial representation container, and a credibility container; perform semantic standardization processing on the fault location elements, mapping the original representations to standard names, and binding auxiliary metadata to each fault location element; the auxiliary metadata includes at least source, conflict marker, and comprehensive score; fill the processed fault location elements into their respective containers according to their corresponding dimensions; perform cross-dimensional cross-validation on the fields in the line dimension container, the power supply dimension container, and the mileage dimension container to determine conflicting models; mark the conflicting models to obtain the fault location model.
[0010] Based on the above technical means, by using five-dimensional containers to carry heterogeneous elements in different dimensions, trace the entire process of attached metadata, and cross-dimensional cross-verification, the fault location information has a unified structured carrier and credibility measurement, providing a scoring benchmark for downstream conflict resolution and avoiding incorrect location caused by field conflicts.
[0011] One possible implementation method for the conflict resolution analysis is as follows: extracting the line field from the fault location model, performing a first-level resolution on the fault location elements to retain the candidate set under the same line; when both the station area field and the substation or feeder field exist simultaneously, performing a second-level resolution on the candidate set based on the power supply affiliation to obtain a resolved candidate set; performing a third-level resolution on the resolved candidate set to merge multiple discrete kilometer marker candidates into a single mileage interval, so as to determine the pillar with the smallest difference from the target analytical mileage as the optimal candidate; and determining the target fault location information containing single-point or interval mileage based on the mileage pattern of the optimal candidate.
[0012] Based on the above technical means, through a three-layer elimination mechanism that narrows layer by layer, irrelevant candidates across lines are eliminated in one go, resolving ambiguity of multiple lines with the same name station area; cross-business conflict candidates are eliminated by utilizing power supply affiliation; and multiple discrete mileage candidates are merged into a single interval to eliminate multi-point conflicts, thereby stably determining the unique target fault location information from multiple candidate results.
[0013] One possible implementation method for determining the fault map corresponding to the fault event based on the target fault location information can be specifically implemented as follows: Based on the target fault location information, determine whether the fault type is a single-point fault or an interval fault; load the trajectory point sequence and cumulative mileage sequence of the line corresponding to the target fault location information; perform interpolation calculation on the single-point or interval mileage based on the cumulative mileage sequence to obtain the track centerline reference coordinates; perform lateral correction based on the unit normal vector and lateral offset distance of the trajectory segment where the track centerline reference coordinates are located to obtain the fault center point; based on the fault type, determine the corresponding fault range circle and / or fault line segment; based on the fault center point, the fault range circle and / or the fault line segment, determine the map spatial object; based on the fault center point, determine the surrounding objects; based on the map spatial object and the surrounding objects, determine the fault map.
[0014] Based on the above technical means, the business mileage is converted into map coordinates by cumulative mileage interpolation, and the lateral deviation between the track centerline and the actual position of the contact network is eliminated by lateral correction, so that the fault location is more accurate; at the same time, differentiated geometric objects such as points, circles and line segments are generated according to the fault type to realize a unified spatial expression of single-point faults and section faults.
[0015] One possible implementation method is as follows: the interpolation calculation can be specifically implemented by dividing the entire line into multiple curve units based on the curvature difference between adjacent trajectory points, and binding the start and end cumulative mileage and curve control point to each curve unit; determining a target curve unit containing the single point or interval mileage in the curve unit, and performing quadratic curve interpolation on the single point or interval mileage based on the target curve unit to obtain the reference coordinates of the track centerline.
[0016] Based on the above technical means, by dividing curve units and using quadratic curve interpolation, the conversion from mileage to coordinates is made to fit the actual track direction in the curve section, which significantly reduces the positioning error in the curve section compared with polyline linear interpolation.
[0017] One possible approach to determining the lateral offset distance is as follows: A lateral offset feature database is constructed, storing the correspondence between line type, row type, section curvature level, standard lateral offset, and historical average positioning error. Using line type, row type, and section curvature level as search criteria, and the fault type as filtering criteria, the standard lateral offset is matched against the lateral offset feature database, and adaptive adjustment is performed in conjunction with the current section type to obtain the lateral offset distance.
[0018] Based on the above technical means, through multi-dimensional retrieval and adaptive adjustment of the lateral offset feature database, the lateral correction amount is dynamically adapted to the line type, row, and section curvature, avoiding systematic deviations caused by fixed offset amounts in curve sections or special sections.
[0019] One possible implementation method is as follows: the types of surrounding objects include point-like surrounding objects, line-like surrounding objects, and work area objects; the determination of surrounding objects based on the fault center point can be specifically implemented as follows: for point-like surrounding objects, calculate the distance between them and the fault center point, and determine point-like surrounding objects whose distance is less than or equal to a preset search radius as first candidate surrounding objects; for line-like surrounding objects, calculate the shortest distance from the fault center point to the line-like surrounding objects, and determine line-like surrounding objects whose shortest distance is not greater than the preset search radius as second candidate surrounding objects; for work area objects, match the jurisdiction mileage of the work area object with the fault mileage interval, and determine work area objects whose jurisdiction range corresponding to the jurisdiction mileage includes the fault mileage interval as third candidate surrounding objects; sort the three types of candidate surrounding objects based on distance and / or business priority to determine a candidate surrounding object sequence; when the number of candidate surrounding objects in the candidate surrounding object sequence exceeds a first number, trim the second number of candidate surrounding objects at the end of the sequence to obtain the surrounding objects.
[0020] Based on the above technical means, by using spherical distance, shortest point-to-line distance and jurisdictional mileage inclusion relationship to filter three types of objects—point, line, and work area—and uniformly sorting and cropping them, we can achieve joint filtering of surrounding objects with fault location as the core, ensuring the complete fault context and controllable display scale.
[0021] One possible implementation method for displaying the fault map is as follows: determining a view fitting carrier from the map spatial objects and the surrounding objects according to a preset priority; determining a bounding box, zoom level, and map center point based on the view fitting carrier; and displaying the fault map based on the bounding box, the zoom level, and the map center point.
[0022] One possible implementation method for determining the bounding box, zoom level, and map center point based on the view fitting carrier is as follows: extracting the extreme coordinates of the view fitting carrier to generate an initial bounding box; obtaining the coordinates of key surrounding objects with a business priority higher than a preset threshold, widening the initial bounding box based on the coordinates of the key surrounding objects to obtain a target bounding box; calculating the base zoom level based on the size of the target bounding box, comparing the base zoom level with preset minimum and maximum zoom thresholds, correcting the base zoom level based on the comparison result to obtain the target zoom level; and determining the geometric center of the target bounding box as the map center point.
[0023] Based on the above technical means, by selecting the view fitting carrier according to priority, expanding the bounding box by combining key surrounding objects, and performing threshold correction on the zoom level, the map view always adaptively focuses around the fault area, avoiding manual zooming and dragging, and shortening the fault assessment time.
[0024] Secondly, this application provides a contact network fault location and map adaptive display device based on fault text location parsing and spatial constraints. The device includes a processor and a memory. The memory stores a computer program, and when the computer program is executed by the processor, it implements the method described in the first aspect above.
[0025] Thirdly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in the first aspect above.
[0026] Fourthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the method described in the first aspect above.
[0027] The solutions provided in the second to fourth aspects above are used to implement the method provided in the first aspect above, and their specific implementations will not be described in detail here. The technical effects corresponding to any implementation method of the solutions provided in the second to fourth aspects above can be found in the technical effects corresponding to any implementation method in the first aspect above, and will not be described in detail here.
[0028] It should be noted that any of the possible implementations of any of the above aspects can be combined, provided that the solutions do not contradict each other.
[0029] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description
[0030] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the structure of a fault location display system provided in an embodiment of this application; Figure 2 A flowchart illustrating a method for catenary fault location and adaptive map display based on fault text location parsing and spatial constraints, provided in an embodiment of this application; Figure 3 A flowchart illustrating a knowledge extraction process provided in an embodiment of this application; Figure 4 A flowchart illustrating a knowledge processing procedure provided in an embodiment of this application; Figure 5 A flowchart illustrating a conflict resolution process provided in an embodiment of this application; Figure 6 A flowchart illustrating a fault map determination process provided in an embodiment of this application; Figure 7 A flowchart illustrating a map adaptive display process provided in an embodiment of this application; Figure 8 A schematic diagram of a contact network fault location and map adaptive display device based on fault text location parsing and spatial constraints provided in this application embodiment; Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation
[0031] In the embodiments of this application, in order to clearly describe the technical solutions of the embodiments of this application, the terms "first" and "second" are used to distinguish the same or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and the terms "first" and "second" are not necessarily different.
[0032] In the embodiments of this application, the words "exemplarily" or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplarily" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design solutions. Specifically, the use of the words "exemplarily" or "for example" is intended to present the relevant concepts in a specific manner to facilitate understanding.
[0033] In the embodiments of this application, at least one can also be described as one or more, and multiple can be two, three, four or more, and this application does not impose any restrictions.
[0034] To facilitate understanding, the terms used in the embodiments of this application will be explained first.
[0035] Overhead contact system: A power transmission facility that is erected above the railway line to continuously supply power to electric locomotives. It is usually composed of components such as supports, cantilever arms, catenary wires, and contact wires.
[0036] Kilometer marker: A mileage coordinate on a railway line that marks the location of the line in kilometers. It is usually written in the form of K23+400, where K is the mileage identifier, 23 represents the number of kilometers, and 400 represents the number of meters. That is, the location is 23 kilometers and 400 meters from the starting point of the line.
[0037] Line type: The direction of operation of a railway line, such as up or down; in a double-track railway, up and down correspond to two independent lines.
[0038] Track: The line number within a station or station area for trains to stop and pass through, such as track 3 or track 5 in a certain station.
[0039] Station area: a collective term for all tracks and related equipment within the station area; Section: a section of track between two stations.
[0040] Substation: A facility that reduces the voltage of high-voltage electrical energy transmitted by the power system and converts it into traction power supply energy; also known as a traction substation.
[0041] Feeder: A power supply branch that transmits electrical energy from a substation to the overhead contact line. It is usually identified by a number, such as feeder 213, which indicates the power supply branch numbered 213.
[0042] Support pole: A pole or tower structure arranged at certain intervals along the railway line to support the contact wire. Each support pole corresponds to a specific mileage position on its respective line.
[0043] GIS (Geographic Information System): A technical system used to collect, store, manage, calculate and visualize geospatial data. Spatial objects such as latitude and longitude coordinates and geometric figures in map displays are usually managed and rendered using GIS.
[0044] Track point sequence: A set of latitude and longitude coordinate points collected along the centerline of the route in the order of travel; Cumulative mileage sequence: A sequence of cumulative mileage values from the starting point of the route to each track point, corresponding to each point in the track point sequence.
[0045] Interpolation: A method for estimating intermediate locations or values among known discrete data points; in this embodiment of the disclosure, interpolation is used to convert mileage values into geographic coordinates.
[0046] Normal vector: A unit direction vector perpendicular to the direction of travel of the line and pointing to the side of the line; Curvature: A measure of the degree of curvature of a curve. The greater the curvature, the steeper the curve. A curvature of approximately zero indicates a straight line.
[0047] Bounding box: The rectangular area that surrounds the target object on the map view; Zoom level: The scale of the map display. The larger the zoom level, the smaller the map display area and the richer the details.
[0048] Spherical distance: The shortest distance between two points on the Earth's surface along a great circle. Taking into account the curvature of the Earth, it is closer to the actual geographical distance than the straight-line distance on a plane.
[0049] Lateral offset distance: The lateral offset of the contact wire support and fault location relative to the track centerline; the track centerline is the geometric centerline of the line, and the contact wire facilities are distributed laterally along the track. Therefore, the track centerline coordinates obtained by mileage interpolation need to be laterally corrected to represent the actual location of the fault.
[0050] The solutions provided in the embodiments of this application will be described in detail below with reference to the accompanying drawings.
[0051] like Figure 1 As shown in the figure, a fault location display system provided in this application embodiment may include: a fault processing server 101, an upstream data source 102, and a map service terminal 103.
[0052] Specifically, the upstream data source 102 includes a scheduling input terminal, a historical fault recording system, and an upstream monitoring system, which are used to push fault event data packets to the fault processing server 101; the fault processing server 101 is used to receive fault events and original fault location text, and perform knowledge extraction, knowledge processing, conflict resolution analysis, and fault map generation; the map service terminal 103 is used to load line trajectory data and display the fault map generated by the fault processing server 101.
[0053] For example, the dispatcher enters the fault description "Upstream Yizhuang Station K23+400" into the dispatch input terminal. The upstream data source 102 pushes the fault event data packet to the fault processing server 101. The fault processing server 101 automatically completes text parsing, location resolution and coordinate conversion. The map service terminal 103 automatically focuses the map view on the catenary section corresponding to K23+400 and displays the surrounding supports, key equipment and the work area to which it belongs.
[0054] like Figure 2 As shown in the figure, the contact network fault location and map adaptive display method based on fault text location parsing and spatial constraints provided in this application embodiment may include: S201: In response to receiving a fault event, obtain the original fault location text and basic context information, perform knowledge extraction on the original fault location text, and determine the fault location elements.
[0055] Among them, a fault event refers to a notification message triggered by an upstream detection device, monitoring system or manual reporting, used to indicate an abnormality in the overhead contact line. The fault event must carry at least the fault location text and basic event metadata.
[0056] Original fault location text refers to information describing the location of a fault in natural language or semi-structured text, such as "Uplink XX Station K23+400" or "XX Substation 213 Feeder 23.400".
[0057] Basic context information refers to the associated information obtained along with the fault event and used to constrain the subsequent analysis process. Basic context information may include at least one of the following: line name, fault type, entry time, fault source, historical fault number, user-selected station area, user-selected line type, user-selected substation, and user-selected feeder number.
[0058] Basic contextual information serves as a strong constraint for subsequent parsing, which can significantly narrow the range of parsing candidates and reduce ambiguity.
[0059] In one implementation, a unified event receiving interface is built to block and listen for fault event data packets pushed by various upstream data sources; the basic event metadata of the data packets is parsed to obtain the unique fault number, fault occurrence time, fault type, and fault reporting source channel; the original fault location text in the data packets is extracted synchronously, and basic context information is obtained from the fault event entry page, upstream system fields, or historical fault archives.
[0060] Among them, blocking listening refers to the listening method in which the interface continuously occupies the listening channel after startup, and immediately receives and processes any data packets that arrive, so as to avoid missing fault events.
[0061] Optionally, the original fault location text can be derived from one of two typical scenarios.
[0062] Scenario 1: Users complete data entry using structured fields, such as selecting fields from dropdown menus like line, row, station area, substation, and feeder number. The system then concatenates these structured fields into standard text according to a predefined template. In this scenario, the text format is standardized and the parsing difficulty is low, making it the basic application scenario for this embodiment. Scenario 2: The original fault location text originates from historical fault records, upstream system pushes, or manual descriptions. In this scenario, the order, abbreviations, and separators of elements such as direction, station area, substation, and feeder in the text vary.
[0063] For Scenario 2, through subsequent steps of text standardization preprocessing and structured parsing, non-standard text can be transformed into structured elements consistent with Scenario 1, allowing both scenarios to share the same processing flow.
[0064] S202: Perform knowledge processing based on fault location elements to determine the fault location model.
[0065] Among them, the fault location model is a structured model obtained by classifying and filling fault location elements into the fields of line, power supply, mileage, spatial representation and credibility.
[0066] The aforementioned line field is used to carry location elements under the semantics of line operation, such as line name, station, section, line type, and track.
[0067] The power supply field is used to carry location elements under the power supply operation and maintenance semantics, such as substation and feeder number; the mileage field is used to carry mileage information such as single-value kilometer markers, start and end interval mileage, and multi-value discrete mileage; the spatial representation field is used to carry spatial morphology identifiers such as single points and interval segments; and the credibility field is used to carry quality metadata such as the source, hit rules, conflict markers, and comprehensive scores of each element.
[0068] It should be understood that by filling in the scattered elements obtained from knowledge extraction by categorizing them according to dimensions, the fault location model transforms unstructured text into structured data with clear fields, traceable sources, and complete dimensions, which serves as the unified data foundation for subsequent conflict resolution and map mapping.
[0069] S203: Based on the fault location model, perform conflict resolution analysis on the fault location elements to determine the target fault location information, which includes single points or interval mileage.
[0070] Among them, conflict resolution analysis refers to the analysis process of filtering and merging candidates layer by layer based on field constraints in the model when there are multiple candidates for fault location elements, business contradictions or numerical ambiguities between elements, and finally obtaining a unique and reliable location result.
[0071] Single-point mileage refers to mileage expressed using a single kilometer marker; interval mileage refers to mileage expressed in pairs, such as K23+400 to K25+600.
[0072] For example, when there are two candidate types, "Yizhuang Station" and "Yizhuang Substation 01 Feeder", in the fault location model, the station area and the substation feeder are checked to see if they belong to the same power supply range based on the power supply affiliation relationship, and candidates without affiliation are eliminated; when there are multiple discrete kilometer marker candidates, the multiple candidates are merged into a single mileage interval, and the support with the smallest difference from the target analytical mileage is selected as the optimal candidate.
[0073] S204: Based on the target fault location information, determine the fault map corresponding to the fault event. The fault map includes at least map spatial objects and surrounding objects.
[0074] Among them, the fault map refers to a collection of map data that organizes spatial display content around the fault location. The fault map includes at least map spatial objects and surrounding objects.
[0075] Map spatial objects refer to a collection of geometric objects generated with the fault location as the core, such as the fault center point, the fault range circle, and the fault line segment.
[0076] Surrounding objects refer to business objects related to fault handling, selected based on spatial constraints and centered on the fault location, such as pillars, key equipment, bridges, and work areas.
[0077] In one implementation, the fault type is determined to be either a single-point fault or an area fault based on the target fault location information. The mileage information in the target fault location information is interpolated and converted into geographic coordinates and lateral correction is performed to obtain the fault center point. Then, map spatial objects such as fault range circles and / or fault line segments are generated. Surrounding objects are filtered with the fault center point as the core, and finally, the map spatial objects and surrounding objects are organized into a fault map.
[0078] S205: Display the fault map.
[0079] In one implementation, a view fitting carrier is determined from map spatial objects and surrounding objects according to a preset priority. The bounding box, zoom level and map center point are automatically calculated based on the view fitting carrier, and the map layer is rendered according to the calculation results, so that the map view automatically focuses on the fault area.
[0080] It should be understood that in the method provided in the embodiments of this disclosure, the outputs of each step are consumed in subsequent steps in sequence: the fault location elements obtained by knowledge extraction are used as input for knowledge processing, the fault location model obtained by knowledge processing is used as the constraint basis for conflict resolution analysis, the target fault location information obtained by conflict resolution is used as the spatial reference for fault map determination, and the fault map is used as the direct data source for the display step, forming a full-link data closed loop from text to structured model, from model to spatial location, and from location to map display.
[0081] This disclosure provides a schematic flowchart of a knowledge extraction process. For example... Figure 3 As shown, in one embodiment, the above-described S201, in response to receiving a fault event, involves obtaining the original fault location text and basic context information, performing knowledge extraction on the original fault location text, and determining fault location elements, including the following steps: S301: Perform text standardization preprocessing on the original fault location text to obtain standardized text.
[0082] Among them, text standardization preprocessing refers to the operation of standardizing the content with inconsistent format and non-standard expression in the original fault location text. The processed text is the standardized text.
[0083] In one implementation, spaces, underscores, forward slashes, Chinese commas, and Chinese dashes in the original fault location text are standardized as delimiters.
[0084] For example, both "Uplink XX Station / K23+400" and "Uplink XX Station, K23+400" are standardized by using a standard delimiter to ensure a consistent text structure for subsequent processing. A complete copy of the original text is retained during preprocessing for subsequent error tracking and result verification.
[0085] In one implementation, different formats of kilometer markers are unified into a standard mileage value.
[0086] For example, K23+400 represents the 23rd kilometer plus 400 meters, which is normalized to the value 23.400 in kilometers; 23Km400m also represents 23.400 kilometers; 23.400 is directly resolved to 23.400 kilometers; 23+400 is resolved to 23.400 kilometers; the prefix JJK in JJK23+400 is the line marking code, and during parsing, the prefix is stripped and 23+400 is normalized to 23.400.
[0087] It should be understood that kilometer markers are expressed in various styles, including national standard formats, industry standard formats, and local custom formats for each route. By using a built-in kilometer marker format recognition rule library for batch normalization, different styles of mileage expressions are converted into standard mileage values under a unified dimension, providing directly usable numerical input for subsequent interpolation calculations.
[0088] Optionally, for the original fault location text derived from free text scenarios, directional terms and business terms can also be identified. Directional terms include up, down, and main line, while business terms include station, section, station, substation, and feeder. Furthermore, business abbreviations can be mapped back to their standard names; for example, "Yizhuang Substation" can be restored to "Yizhuang Substation" and "Yongle Station" can be restored to "Yongle Station Standard Name". For scenarios where data is entered through structured fields, the directional terms and business terms are explicitly selected by the user during the data entry process, allowing for direct reuse of the structured fields without repeated recognition.
[0089] In one implementation, isolation marking is applied to device numbering numbers with special suffixes, such as feeder numbers, to obtain protected numbers.
[0090] For example, the number 213 in "213 feeder" is the feeder number rather than the mileage. The preprocessing stage identifies the combination of this number and a special suffix and adds an isolation mark to it. When performing mileage matching later, the number segment with the isolation mark is automatically skipped to avoid the feeder number being misidentified as a kilometer marker.
[0091] It should be understood that this protected digital isolation mechanism cuts off the path to digital misjudgment at the source and is a key preprocessing operation to ensure the accuracy of positioning.
[0092] In one implementation, single-value, range-value, and multi-value kilometer markers and their connector relationships are identified. A single-value kilometer marker represents a specific kilometer point, such as K23+400; a range-value kilometer marker represents two kilometer points, such as K23+400 to K25+600, generating two standardized values (start and end mileage) during identification; a multi-value kilometer marker represents multiple discrete kilometer points, such as K23+400, K25+600, and K28+800, generating a standardized mileage array during identification. Connectors include "to", "to", "~", "-", "~", and space-separated consecutive numbers, determining the combination relationship between adjacent mileages based on the connector type.
[0093] S302: Load line operation semantic template and power supply operation and maintenance semantic template.
[0094] Semantic templates refer to a rule template library that defines the combination rules and arrangement order of location elements. Line operation semantic templates include combination rules for line name, line type or track, station or section, and kilometer marker, i.e., "line name + line type / track + station / section + kilometer marker"; power supply operation and maintenance semantic templates include combination rules for substation, feeder number, and kilometer marker, i.e., "substation + feeder number + kilometer marker".
[0095] For example, text conforming to the line operation semantic template, such as "Beijing-Tianjin Intercity Railway Upward K23+400", has the element sequence of line name Beijing-Tianjin Intercity Railway, line type Upward, and kilometer marker K23+400; text conforming to the power supply operation and maintenance semantic template, such as "Yizhuang Substation 01 Feeder K23+400", has the element sequence of substation Yizhuang Substation, feeder number 01, and kilometer marker K23+400.
[0096] It should be understood that the two sets of semantic templates correspond to two parallel location expression systems in the overhead contact line operation and maintenance scenario. Simultaneously loading the two sets of templates enables subsequent parsing to recognize location texts of both calibers, avoiding the inability to automatically locate faults due to insufficient parsing capabilities of a single semantic system.
[0097] S303: Based on standardized text and basic context information, the standardized text is structured and parsed using line operation semantic templates and power supply maintenance semantic templates to extract fault location elements.
[0098] Structured parsing refers to the process of breaking down standardized text into field elements with clear business meanings according to the combination rules of semantic templates, and then performing quality assessment and classification of each element. Fault location elements refer to structured fields extracted from standardized text that can describe the fault location, including line-related elements, power supply-related elements, mileage-related elements, and auxiliary location-related elements.
[0099] In one implementation, the structured parsing process includes pre-constraint filtering and parsing initialization.
[0100] Specifically, four major business dictionaries are loaded: lines, stations, substations, and feeders. These business dictionaries are pre-collected lexicons of standard names and alternative names for each line, station, substation, and feeder. The business dictionaries are then pruned using basic context information. For example, if the basic context information specifies a high-speed rail line, other lines and their associated station entries are filtered out, retaining only the business terms that match the line. If the basic context information pre-selects a substation or feeder, a subset of the power supply semantic dictionary is locked to reduce matching interference from irrelevant entries. In the absence of context constraints, a complete dictionary including two sets of vocabulary systems—line operation semantics and power supply maintenance semantics—is loaded.
[0101] In one implementation, standardized text is segmented into words and blocks to obtain independent semantic fragments. Specifically, protected numbers marked in the standardized text are read and kept isolated. The text is then segmented into words and blocks based on standard delimiters to separate independent semantic fragments such as line type, station location, mileage, and feeder line. Each semantic fragment is bound to attributes such as the original text, standardized text, semantic type, and location range to form a sequence of word blocks with semantic tags for subsequent matching and extraction.
[0102] In one implementation, four types of matching extraction are performed in parallel at different levels.
[0103] Specifically, the first category is business dictionary entity matching extraction, which uses the forward maximum matching algorithm to traverse the segmented word fragments; after matching, the extracted entity is immediately subjected to compatibility verification with the basic context information. For example, if the context specifies the line as line A but the extracted station area belongs to line B, the entity is determined to be incompatible with the context, and it is marked as a low-confidence candidate and temporarily stored.
[0104] Among them, positive maximum matching refers to the word segmentation strategy that starts from the beginning of the segment and prioritizes matching the longest dictionary entry to extract standard entities such as line, station area, interval, substation, feeder, and line.
[0105] The second category is kilometer marker pattern matching extraction, which uses regular expression matching to identify standardized single values, start and end intervals, and multiple sets of discrete mileage segments. Regular expression matching refers to the method of pattern recognition and extraction of text according to predefined character pattern rules. During matching, protected numbers are avoided so that numbers such as feeder numbers do not participate in mileage recognition, and mileage type identifiers are recorded simultaneously. Mileage type identifiers include three types: single point, interval, and multi-value mileage set.
[0106] The third category is field order template matching, which performs sequence matching between the extracted entities and mileage fragments and the line operation semantic template and the power supply operation and maintenance semantic template to determine whether the standardized text as a whole belongs to the line scope or the power supply scope; the matching process allows the element order to be misplaced, and the template matching priority is corrected by the context line or substation constraints.
[0107] The fourth category is the extraction of additional directional reference elements, which extracts descriptive directional elements such as "nearby", "upward side", and "middle section of the interval" as well as adjacent reference objects, and stores them in the candidate set as auxiliary location elements for subsequent auxiliary positioning and result verification.
[0108] In one implementation, reliability scoring and grading are performed on all extracted candidate location elements. Specifically, candidate location elements are weighted and scored based on four dimensions: context matching degree, completeness of business dictionary hit, reasonableness of template sequence, and legality of kilometer marker format, to obtain a comprehensive candidate score.
[0109] Among them, context matching degree is used to characterize the consistency between candidate elements and basic context information. A higher score is obtained when the candidate element is completely consistent with the pre-selected line, substation, or feeder in the context. The score is significantly reduced when there are conflicts across lines or substations. No adjustment is made when there is no context. Business dictionary hit completeness is used to characterize the completeness of the candidate element's hit in the standard dictionary. A higher score is obtained when all elements match the standard dictionary. The score is reduced accordingly when there are unrecognized unfamiliar abbreviations. Template sequence rationality is used to characterize the degree of matching between the element combination and the standard word order of the semantic template. A higher score is obtained when the element combination completely matches the standard word order. The score is reduced accordingly when the elements are scattered and the word order is disordered. Kilometer mark format legality is used to characterize the standardization of mileage data. A higher score is obtained when the mileage has been standardized and there are no numerical confusion conflicts. The score is reduced accordingly when multiple sets of mileage have no clear interval boundaries.
[0110] Optionally, the scoring weights, the magnitude of increase or decrease in scoring, and the grading thresholds can all be flexibly configured according to the actual business scenario.
[0111] For example, candidate elements whose context specifies line A, whose extracted line name is also line A, whose entities all match the standard dictionary, whose element combinations completely match the semantic template word order, and whose mileage format is standardized and unambiguous, are all rated as highly reliable if they score full marks in all four dimensions. Candidate elements with partial element matching, minor ambiguity but no serious conflicts across lines or substations are rated as moderately reliable. Candidate elements with a large number of unrecognizable entities and cross-category conflicts between the element and the context are rated as low reliable. Conflict markers are added to elements with multiple mutually exclusive candidate sets, numerical ambiguity, or semantic contradictions.
[0112] In one implementation, all candidate elements are categorized and aggregated according to line dimension, power supply dimension, mileage dimension, and auxiliary orientation dimension. Line elements include line name, type, track, station area, and section; power supply elements include substation and feeder number; mileage elements include single-value mileage, starting mileage, ending mileage, and multi-value standardized mileage; and auxiliary orientation elements include adjacent reference objects and orientation description. Each element is bound to three types of auxiliary metadata: matching rule template, conflict marker, and reliability score. The categorized and aggregated elements are encapsulated into a data package of structured parsing results. The data package includes a full-dimensional location element candidate set, the reliability level and conflict marker corresponding to each element, the original standardized text, and a snapshot of basic context information.
[0113] It should be understood that the data packets output by the structured parsing, as a set of fault location elements, are output to the knowledge processing steps and serve as the data source for filling various field containers. The reliability score and conflict marker carried by each element will continue to participate in the filtering and sorting during the knowledge processing and conflict resolution stages to ensure that data quality information is not lost throughout the entire process.
[0114] Figure 4 This is a flowchart illustrating a knowledge processing procedure provided in an embodiment of this disclosure. Figure 4 As shown, in one embodiment, the above-described S202, performing knowledge processing based on fault location elements to determine a fault location model, includes the following steps: S401. Initialize the empty model carrier and create the line dimension container, power supply dimension container, mileage dimension container, spatial representation container, and credibility container.
[0115] Among them, the model carrier refers to the storage structure used to carry all field data of the fault location model.
[0116] Specifically, the line dimension container is used to store fields under the semantics of line operation, such as line name, station, section, line type, and track; the power supply dimension container is used to store fields under the semantics of power supply operation and maintenance, such as substation and feeder number; the mileage dimension container is used to store single-value kilometer markers, starting mileage, ending mileage, and multi-value discrete mileage sets; the spatial representation container is used to store spatial morphology identifiers such as single points and section segments; and the credibility container is used to store quality metadata such as field source, hit rule number, conflict marker, and candidate comprehensive score.
[0117] In one implementation, after initializing the empty model carrier, five sets of independent field storage containers are created in parallel, and basic context information is loaded as the filling priority rule. That is, high-reliability features that match the context are written into the main field of the container first, while low-reliability features or features with conflict markers are stored in the alternative cache area to ensure the filling quality of the main field.
[0118] S402. Perform semantic standardization processing on the fault location elements, map the original descriptions to standard names, and bind auxiliary metadata to each fault location element.
[0119] Semantic standardization processing refers to the process of converting the original expressions obtained from knowledge extraction into business standard names.
[0120] For example, the abbreviation "Yizhuang Substation" is mapped to the standard name "Yizhuang Substation", and the local customary station name is mapped to the standard station name in the business dictionary, so that elements from different sources and with different expression habits have a unified name standard at the model layer.
[0121] The ancillary metadata includes at least the source, conflict markers, and a comprehensive score. The source distinguishes whether the element was obtained from original text parsing, directly assigned values based on pre-selected contextual information, or generated through multi-candidate derivation. The conflict marker indicates whether the element exhibits cross-dimensional contradictions, numerical ambiguities, or other conflicts. The comprehensive score uses the weighted scoring results from the structured parsing stage. It should be understood that the ancillary metadata makes the quality information of each element traceable and serves as the scoring benchmark for subsequent field filtering and conflict resolution.
[0122] S403. Fill the processed fault location elements into the corresponding containers according to their respective dimensions.
[0123] Among them, the line dimension container should at least be filled with the line name and station area, the power supply dimension container should at least be filled with the substation and feeder number, the mileage dimension container should at least be filled with the single-value kilometer mark or the start and end interval mileage, and the spatial expression container should be filled with the single point or interval segment identifier.
[0124] In one implementation, during the filling of the line dimension field, all candidate elements of the line type in the parsing results are traversed. Based on the sorting rule of prioritizing basic context matching degree over reliability level, the best single element is filled into the main field, and the remaining second-best candidates are stored in the candidate cache. Only high-reliability terms consistent with the context line are retained for the line name; if multiple conflicting line candidates exist, low-matching terms across lines are directly discarded. During the filling of the power supply dimension field, the context-preselected substation or feeder is used as the highest filtering weight. If a preselected substation exists in the context, only the matching substation and corresponding feeder are retained. Candidates with feeder numbers marked with isolation tags in the text and conflicting with the context are directly filtered, and the best substation and feeder are written into the power supply dimension main field. During the filling of the mileage dimension field, mileage is filled according to its form: single-value mileage is filled into the single-value kilometer marker field; start-end paired mileage is filled into the start mileage field and end mileage field respectively; multiple sets of discrete mileage are all stored in a multi-value discrete mileage set.
[0125] In one implementation, auxiliary mileage values are automatically derived during the mileage filling process. For example, if there are start and end mileages, the center mileage is automatically calculated and stored in the mileage dimension container. If there are multi-value discrete mileages, they are automatically merged to generate the minimum coverage start and end interval, providing directly usable data for subsequent spatial expression judgment and interval fault display.
[0126] In one implementation, the spatial representation field is automatically filled based on mileage dimension data: when only a single-value kilometer marker exists, the spatial representation container is marked as a single point; when there are start and end interval mileages or the smallest coverage interval after merging multiple discrete mileages, it is marked as an interval segment. It should be understood that the spatial representation field only stores spatial morphological identifiers, and the actual geometric coordinates are reserved for generation in subsequent map mapping steps, thereby establishing a unified representation of single-point faults and interval faults at the model layer.
[0127] S404. Perform cross-dimensional cross-validation on the fields in the line dimension container, power supply dimension container, and mileage dimension container to identify conflicting models.
[0128] Cross-dimensional cross-validation refers to the mutual validation performed by the three main fields of line, power supply, and mileage, which is used to discover business contradictions that cannot be detected within a single dimension.
[0129] In one embodiment, S404 above, performing cross-dimensional cross-validation on the fields in the line dimension container, power supply dimension container, and mileage dimension container to determine conflicting models, includes the following steps: Ⅰ. Perform a matching attribution verification on the line name, substation, and feeder. If the affiliated lines are inconsistent, it is determined that there is a cross-business conflict.
[0130] In one implementation, the business association ledger is retrieved to verify whether there is a matching affiliation between the line name in the line dimension container and the substation and feeder in the power supply dimension container. That is, whether the substation is a matching traction substation for the line and whether the feeder is supplied by the substation. If the line to which the substation or feeder belongs is inconsistent with the line name in the line field, a cross-business conflict is determined. For example, if the model's line field is line A, but the substation in the power supply field belongs to line B, a cross-business conflict exists.
[0131] II. Perform range verification on the kilometer markers and the operating mileage range of the line. If the kilometer markers exceed the operating mileage range, it is determined that there is a business conflict between the mileage and the line.
[0132] In one implementation, the system retrieves the basic route ledger to obtain the operating mileage range of the route corresponding to the route field. The kilometer markers in the mileage dimension container are then compared with this operating mileage range. If the kilometer marker exceeds the operating mileage range, is negative, or clearly contradicts the route's origin and destination, a business conflict between mileage and route is identified. For example, if the operating mileage range corresponding to the route field is 0 to 120 kilometers, and the kilometer marker in the mileage field is 150.500, it is determined that the kilometer marker exceeds the operating mileage range, indicating a business conflict between mileage and route.
[0133] III. Perform coverage verification on the kilometer marker and the power supply coverage area. If the kilometer marker is not within the power supply coverage area, it is determined that there is a cross conflict between power supply and mileage.
[0134] The power supply coverage area refers to the mileage section of the overhead contact line that is supplied by a substation or feeder. Each feeder in the traction power supply system has a defined power supply range. In one implementation, the kilometer marker in the mileage dimension container is compared with the power supply coverage area of the substation or feeder in the power supply dimension container. If the kilometer marker is not within the power supply coverage area, a conflict between power supply and mileage is determined.
[0135] IV. Perform power supply affiliation verification on the station area and substation. If there is no power supply affiliation, determine that there is a cross conflict between the station area and the power supply.
[0136] In one implementation, if both the station area field and the substation field exist, the business association ledger is retrieved to verify whether there is a power supply affiliation between the station area and the substation; if there is no power supply affiliation, it is determined that there is a cross conflict between the station area and the power supply.
[0137] In one implementation, if any of the above contradictions are detected, the current model is determined to be a model with a conflict, and the contradictory dimensions and contradictory content found are synchronously recorded to the credibility cache; if all combination verifications pass, the current values of each field in the model remain unchanged.
[0138] In one implementation, model integrity verification is performed synchronously during cross-dimensional cross-validation. If there is no parsing result for a certain dimension, the corresponding field is set to empty and marked as no matching candidate in the credibility container. Invalid candidate cache is filtered synchronously, and conflict candidates with a comprehensive score lower than a preset threshold are directly removed, while only high-reliability and medium-reliability candidates are retained for subsequent conflict resolution calls.
[0139] It should be understood that cross-dimensional cross-validation places the fields of the three dimensions of line, power supply, and mileage under the same business rules for mutual verification. It explicitly exposes contradictions that cannot be detected by single-dimensional analysis at the model layer, so that conflicting models carry clear contradiction information into the conflict marking and subsequent resolution process, thus ensuring the integrity and credibility of the fault location model.
[0140] S405. Conflict marking is performed on the conflicting models to obtain the fault location model.
[0141] In one implementation, for models with conflicts, the conflict type and conflict description are written into the conflict flag field of the credibility container, such as cross-business conflict, risk of numerical misjudgment, mileage interval contradiction, etc. After the conflict flag is written, a fault location model with complete fields and traceable quality information is obtained.
[0142] In one implementation, after obtaining the fault location model, a unified spatial constraint condition is generated by integrating the five fields of the fault location model.
[0143] The aforementioned unified spatial constraints include line constraints, power supply constraints, mileage constraints, spatial morphology identifiers, and credibility weights, which serve as the constraint basis for conflict resolution analysis and are output to downstream steps.
[0144] It should be understood that unified spatial constraints integrate the scattered fields in the model into a set of constraints that can be directly consumed for conflict resolution, ensuring that all fields in the model continue to play a benchmark role in subsequent processes.
[0145] Figure 5 This is a flowchart illustrating a conflict resolution process provided in an embodiment of this disclosure. Figure 5 As shown, in one embodiment, the above-mentioned S203, which involves performing conflict resolution analysis on fault location elements based on the fault location model to determine target fault location information containing single points or interval mileage, includes the following steps: S501. Extract the line field from the fault location model and perform the first-level elimination on the fault location elements to retain the candidate set under the same line.
[0146] In one implementation, the line name of the line dimension container in the fault location model is extracted as a mandatory filtering condition. All candidate location elements are traversed, and all station areas, substations, mileages, and support poles that do not match the line are eliminated. After filtering, only the entire set of candidates under the same line is retained.
[0147] Optionally, after filtering, the remaining candidates are checked. If there are no remaining candidates, an anomaly marker is triggered, and all original candidates are retained for manual review. If there are remaining candidates, the process proceeds to the next layer of resolution.
[0148] It should be understood that the first layer of resolution eliminates irrelevant candidates across lines in one go, significantly reducing the scale of the conflict and resolving the core ambiguity issue of multiple lines having the same station area.
[0149] S502. When both the station area field and the substation or feeder field exist, perform a second-level elimination on the candidate set based on the power supply affiliation to obtain the eliminated candidate set.
[0150] Among them, the power supply affiliation refers to the power supply jurisdiction correspondence between the station area and the substation and feeder, that is, which station area the substation or feeder is actually responsible for supplying power to.
[0151] The business association ledger refers to a pre-maintained business data table that records the affiliation of each station area and substation feeder under the line.
[0152] In one implementation, when the fault location model has both a station area field and a substation or feeder field, that is, when the candidate set has both line operation semantics and power supply operation and maintenance semantics, the business association ledger is retrieved to verify whether there is a power supply affiliation relationship between the station area field and the substation or feeder field under the current line. Candidates without affiliation relationship are directly downgraded and reordered.
[0153] Optionally, if only line operation semantics exist (i.e., no substation field) or only power supply operation and maintenance semantics exist (i.e., no station area field), cross-validation can be skipped and all candidates for this dimension can be directly retained.
[0154] In one implementation, during the second-layer resolution process, a special digital conflict resolution is performed simultaneously. The isolation marker in the confidence container of the fault location model is read. If a candidate number matches the feeder protection rule, the weighted score of that number as a mileage candidate is directly reduced, and the standard mileage field without isolation marker is selected first to further eliminate the confusion between the feeder number and the kilometer marker.
[0155] S503. Perform a third-level elimination on the eliminated candidate set, merging multiple discrete kilometer marker candidates into a single mileage interval, so as to determine the pillar with the smallest difference from the target analytical mileage as the optimal candidate.
[0156] In one implementation, the third-layer resolution process includes multi-mileage candidate merging and spatial proximity selection.
[0157] In the multi-mileage candidate merging process, when there are multiple discrete kilometer marker candidates in the mileage dimension of the fault location model, the minimum coverage start and end interval of all mileages is automatically calculated, that is, the minimum interval that can cover all discrete mileage candidates, and a unified center mileage is generated, merging multiple discrete candidates into a single mileage interval constraint to eliminate multi-point conflicts.
[0158] In the spatial proximity selection process, the absolute value of the difference between the mileage of each candidate post and the target analytical mileage is calculated for each candidate post within the line. Posts are then sorted by difference from smallest to largest, and the post with the smallest difference is selected as the optimal candidate under the corresponding constraints. Here, post mileage refers to the standard kilometer marker position of the post on its respective line, and target analytical mileage refers to the target mileage value obtained after standardization and merging. For example, if the target analytical mileage is 23.400, and there are three candidate posts on the line with mileages of 23.320, 23.385, and 23.510, with differences of 0.080, 0.015, and 0.110 from the target analytical mileage respectively, then the post corresponding to mileage 23.385 is determined as the optimal candidate.
[0159] It should be understood that the catenary supports are arranged at fixed intervals along the line and each is bound with an accurate mileage coordinate. Anchoring the fault location to the support with the smallest difference is equivalent to calibrating the approximate mileage obtained from text parsing to the precise benchmark point in the business ledger, which significantly improves the positioning accuracy.
[0160] In one implementation, after the third-layer resolution is completed, a global conflict secondary review is performed to check whether there are still cross-dimensional contradictions in the optimal candidate, such as line matching but feeder not belonging, mileage and station area interval being too far apart; in the case of minor conflict, the optimal item is retained and a conflict note is added to the credibility field; in the case of serious cross-business conflict, a high-priority second-best candidate is output and an anomaly mark is added.
[0161] S504. Based on the optimal candidate mileage pattern, determine the target fault location information, which includes single-point or interval mileage.
[0162] In one implementation, the mileage pattern corresponding to the optimal candidate is encapsulated to obtain the target fault location information.
[0163] The target fault location information includes the unique line, the identified line type, station area, substation and feeder, the standard target single-point mileage or standard start and end interval mileage, the optimal matching reference support number and corresponding kilometer marker, the initial value of the fault range radius, conflict notes and the final credibility rating.
[0164] For example, when the optimal candidate mileage format is a single-value kilometer marker, the target fault location information includes the single-point mileage; when the optimal candidate mileage format is a start-end interval or the minimum coverage interval generated by merging, the target fault location information includes the interval mileage composed of the start mileage and the end mileage.
[0165] In one implementation, after determining the target fault location information, data distribution and bidirectional feedback are performed: the forward output transmits the complete data packet of the target fault location information to the subsequent map space mapping steps, serving as the sole reference for mileage interpolation and latitude-longitude conversion; the reverse feedback writes the constraint rules, conflict markers, and optimal candidate numbers used in this resolution back into the credibility container of the fault location model, and the model is updated synchronously for reuse in subsequent fault type determination and result output.
[0166] It should be understood that through a multi-layered progressive resolution process involving the first layer of line filtering, the second layer of power supply affiliation cross-verification, and the third layer of mileage merging and selection, the candidate range is narrowed layer by layer. The final output target fault location information is unique and highly reliable, providing an accurate and unambiguous mileage benchmark for subsequent map spatial mapping.
[0167] Figure 6 This is a flowchart illustrating a fault map determination process provided in an embodiment of this disclosure. Figure 6 As shown, in one embodiment, the above-described S204, determining the fault map corresponding to the fault event based on the target fault location information, includes the following steps: S601. Based on the target fault location information, determine whether the fault type is a single-point fault or an area fault.
[0168] The fault type is used to characterize the spatial form of the fault in the mileage dimension. For example, if the target fault location information includes the standard target single-point mileage, the fault type is determined to be a single-point fault; if the target fault location information includes the start and end interval mileage, the fault type is determined to be an interval fault.
[0169] In one implementation, the fault type is determined based on the spatial morphological identifier in the spatial representation container of the fault location model. If only a single-value kilometer marker exists, the fault type is determined to be a single-point fault. If there is a start-end interval mileage or the smallest coverage interval after merging, the fault type is determined to be an interval fault.
[0170] S602. Load the trajectory point sequence and cumulative mileage sequence of the line corresponding to the target fault location information.
[0171] The trajectory point sequence is a set of latitude and longitude coordinates collected along the centerline of the route in the order of travel.
[0172] The cumulative mileage sequence is a sequence of cumulative mileage values from the starting point of the route to each trajectory point, corresponding point by point to the trajectory point sequence.
[0173] In one implementation, based on the unique route bound in the target fault location information, the trajectory point sequence exclusive to that route and the pre-calculated cumulative mileage sequence are loaded; at the same time, the fault line parameters are read, the unit normal vector and lateral offset distance of the corresponding trajectory segment are cached, and the geometric object storage container is initialized, reserving storage space for center point, interval endpoint, line segment, and search circle, to prepare data for subsequent interpolation and correction calculations.
[0174] It should be understood that the trajectory point sequence and the cumulative mileage sequence constitute the basis for the mapping between mileage values and geographic coordinates: the same physical location on the line can be expressed by either mileage or latitude and longitude, and the conversion between the two is completed based on the trajectory point sequence and the cumulative mileage sequence.
[0175] S603. Based on the cumulative mileage sequence, interpolate the mileage of a single point or interval to obtain the reference coordinates of the track centerline.
[0176] In one implementation, a binary search is performed in the cumulative mileage sequence for the target mileage to locate the line segment where the target mileage is located.
[0177] The binary search mentioned above refers to a search method in an ordered sequence that halves the search range each time until the target interval is located.
[0178] Specifically, in the cumulative mileage sequence, locate the line segment that satisfies the condition that the initial cumulative mileage is less than or equal to the target mileage, and the target mileage is less than or equal to the next adjacent cumulative mileage.
[0179] The above-mentioned broken line segment is formed by connecting two adjacent trajectory points.
[0180] In one implementation, an interpolation ratio coefficient is calculated. The interpolation ratio coefficient is used to characterize the relative position of the target mileage within the broken line segment, and its value ranges from 0 to 1. Based on the interpolation ratio coefficient, linear interpolation is performed on the coordinates of the trajectory points at both ends of the broken line segment, that is, the latitude and longitude coordinates of the trajectory points at both ends are weighted and summed according to the interpolation ratio coefficient to obtain the reference coordinates of the track centerline.
[0181] Among them, the reference coordinates of the track centerline refer to the latitude and longitude coordinates of the target mileage on the geometric centerline of the line.
[0182] In one implementation, coordinates are calculated in batches according to their morphology: when the fault type is a single-point fault, only one set of track centerline reference coordinates is calculated; when the fault type is an interval fault, two sets of reference coordinates corresponding to the starting mileage and the ending mileage are calculated respectively, and the interval temporary center point is generated by the average of the two sets of coordinates.
[0183] In one embodiment, the above-mentioned S603, which involves interpolating the mileage of a single point or interval based on the cumulative mileage sequence to obtain the reference coordinates of the track centerline, includes the following steps: I. The entire line is divided into multiple curve units based on the curvature difference between adjacent trajectory points. Each curve unit is bound to the cumulative mileage from start to end and the curve control point.
[0184] Among them, the curve unit refers to the geometric unit obtained by segmenting the track according to the curvature change; the curve control point refers to the auxiliary coordinate point used to characterize the curvature of the curve, which is obtained by weighted fitting of all track points in the segment.
[0185] In one implementation, for three consecutive trajectory points in the sequence of trajectory points, the local curvature at each trajectory point is calculated using the radius of the circumcircle of the three points. The larger the local curvature, the sharper the curve; when the local curvature is approximately zero, it represents a straight line segment. A curvature abrupt change threshold is preset. If the difference in local curvature between adjacent trajectory points is greater than or equal to the curvature abrupt change threshold, the trajectory point is marked as a segment node. If the difference is less than the curvature abrupt change threshold, the trajectory point is added to the current segment for further traversal. After traversal, the entire line is divided into multiple independent curve units. Each curve unit stores the unit start point, unit end point, cumulative mileage from start to end, curve control points, unit average curvature, and a set of lateral normal vectors of the line.
[0186] For example, the curvature abrupt change threshold can be set to 0.0005 per meter, straight segments can be merged into longer segments, and curves and transition curves can be segmented into independent units at curvature changes. It should be understood that segmenting by curvature difference can adaptively fit the actual geometry of the track, keeping straight segments simple and providing a more refined description of curved segments.
[0187] II. In the curve unit, determine the target curve unit containing single point or interval mileage, and perform quadratic curve interpolation on the single point or interval mileage based on the target curve unit to obtain the reference coordinates of the track centerline.
[0188] In one implementation, a binary search is performed on the target mileage among the cumulative mileages of all curve units to locate the unique target curve unit containing the target mileage; the total mileage length of the target curve unit is calculated, and the normalized curve interpolation parameters of the target mileage within the unit are obtained. The normalized curve interpolation parameters are values between 0 and 1, used to characterize the relative position of the target mileage from the unit's starting point to its ending point; based on the unit's starting point coordinates, curve control point coordinates, and unit ending point coordinates, quadratic curve interpolation is performed according to the normalized curve interpolation parameters to obtain the track centerline reference coordinates.
[0189] For example, quadratic curve interpolation is performed as follows: the reference coordinates of the track centerline are obtained by a weighted combination of the coordinates of the unit's starting point, the coordinates of the curve control points, and the coordinates of the unit's ending point. The weight of the starting point coordinates is the square of the difference between the normalized curve interpolation parameter and 1; the weight of the control point coordinates is twice the product of the normalized curve interpolation parameter and the difference between that parameter and 1; and the weight of the ending point coordinates is the square of the normalized curve interpolation parameter. The sum of the three weights is 1, ensuring that the interpolation result falls within the geometric range of the curve unit.
[0190] It should be understood that, compared to linear interpolation which approximates the line as a broken line segment, quadratic curve interpolation uses curve control points to depict the arc of the curve, making the track centerline reference coordinates obtained from mileage conversion more closely match the actual curve shape of the line. In particular, it can significantly reduce coordinate conversion deviation in the curve section of the line, and further improve the accuracy of fault point geographic location.
[0191] In the case of a single-point fault, the unique track centerline reference coordinates are directly calculated by substituting the normalized curve interpolation parameters; in the case of an interval fault, the curve units to which the starting mileage and the ending mileage belong are located respectively, the reference coordinates of the starting point and the reference coordinates of the ending point are calculated, and the temporary center point of the interval is generated by averaging the two points.
[0192] S604. Based on the unit normal vector and lateral offset distance of the trajectory segment where the track centerline reference coordinates are located, perform lateral correction to obtain the fault center point.
[0193] The unit normal vector refers to the unit direction vector perpendicular to the direction of track travel and pointing towards the up and down contact wire sides; the lateral offset distance refers to the lateral offset distance of the fault point or contact wire support relative to the track centerline. It should be understood that the track centerline reference coordinates only represent the geometric centerline position of the track, while the contact wire supports and fault points are usually arranged laterally along the track, thus requiring lateral correction.
[0194] In one implementation, the local unit normal vector of the trajectory segment where the reference coordinates of the track centerline are located is taken, and the reference coordinates of the track centerline are offset laterally along the direction of the unit normal vector to obtain the corrected coordinates. The corrected coordinates are then determined as the fault center point.
[0195] For example, lateral correction is performed as follows: the corrected coordinates are equal to the sum of the reference coordinates of the track centerline, the product of the lateral offset scalar and the unit normal vector, where the lateral offset scalar is the lateral offset distance corresponding to the current segment, and the direction of the unit normal vector is determined by the fault line parameter.
[0196] Optionally, in the case of a section fault, lateral corrections are performed on the reference coordinates of the section's starting point and ending point, respectively. After the corrections are completed, the fault center point is recalculated based on the corrected coordinates of the two points. It should be understood that the curvature of the line may be uneven and the offset may be inconsistent within a long section. Correcting the starting and ending points separately before calculating the center point can avoid the cumulative deviation caused by overall correction.
[0197] In one embodiment, the process of determining the lateral offset distance includes: constructing a lateral offset feature database that stores the correspondence between line type, row type, section curvature level, standard lateral offset, and historical average positioning error; using line type, row type, and section curvature level as search conditions and fault type as filtering conditions, matching the standard lateral offset in the lateral offset feature database, and performing adaptive adjustment in combination with the current section type to obtain the lateral offset distance.
[0198] The lateral offset feature database refers to a pre-constructed dataset that stores the offset feature parameters of the overhead contact line supports by section. Each data sample in the lateral offset feature database is bound to a unique line identifier and stores multi-dimensional offset feature parameters.
[0199] For example, the basic line attribute fields include line type, row type, section curvature level, design gauge, and track superelevation value. The line type includes intercity high-speed railway, conventional railway, station turnout section, and curved ramp. The section curvature level includes straight section, gentle curve section, and sharp curve section. The offset feature core fields include the standard lateral offset of the catenary support under the corresponding working condition, the offset fluctuation threshold, the optimal correction coefficient, and the historical average positioning error. The fault condition association field includes the offset correction priority weight corresponding to different fault types. The data iteration field includes the actual deviation verification value after each fault positioning correction, which is used for subsequent parameter iteration optimization.
[0200] In one implementation, when constructing the lateral offset feature database, the entire line section is divided into grids according to curvature, line type, and row. Each grid cell is bound to a unique offset feature parameter. Historical abnormal fault location data is removed, and the standard lateral offset is calibrated through a weighted average algorithm to ensure the accuracy of the basic data.
[0201] In one implementation, during the matching process, the line type, row type, and section curvature level are used as primary search conditions, and the fault type is used as a secondary filtering condition. The corresponding standard lateral offset is matched in the lateral offset feature database. Adaptive adjustment is then performed in conjunction with the current section type to obtain the lateral offset distance.
[0202] For example, when the current section type is a straight conventional section, a standard fixed correction method is adopted, directly calling the matched standard lateral offset as the lateral offset distance, which has stable correction accuracy and high calculation efficiency; when the current section type is a curved section, a curvature linkage dynamic correction method is adopted, which combines the curvature of the current section to fine-tune the offset, and the lateral offset correction amplitude increases adaptively as the curvature increases, which conforms to the offset law of the curve support; when the current section type is a station turnout complex section, a high-precision weighted correction method is adopted, which superimposes the fault type weight and historical error compensation parameters to perform fine correction.
[0203] Optionally, after each fault location correction is completed, the line parameters, curvature data, fault type, selected correction strategy, and final location error value are synchronously fed back to the lateral offset feature database, automatically updating the offset mean and correction weight of the corresponding grid cell to achieve adaptive iterative update of the database.
[0204] It should be understood that through multi-condition linkage matching and adaptive adjustment of the lateral offset feature database, the lateral offset distance can dynamically change with line type, row, curvature level and fault type, which solves the problem that fixed offset cannot adapt to the difference in catenary support offset under different lines and working conditions, and further improves the accuracy of fault point coordinate correction in complex sections.
[0205] S605. Based on the fault type, determine the corresponding fault range circle and / or fault line segment.
[0206] The fault range circle refers to a circular area with the fault center point as the center and a radius determined by business rules, which is used to characterize the impact range of the fault.
[0207] A fault segment is a line segment that connects the coordinates of the start and end points of a fault in a given section, and is used to characterize the extent of the fault zone.
[0208] In one implementation, when the fault type is a single-point fault, the fault range circle is determined with the fault center point as the center and the distance from the fault center point to the optimal support as the radius; when the fault type is an interval fault, the fault line segment is determined with the corrected interval start coordinates and interval end coordinates, and the fault range circle is determined with the fault center point as the center and the maximum distance from the fault center point to both ends of the interval as the radius.
[0209] It should be understood that the fault range circle and fault line segment transform the fault pattern in the mileage dimension into a geometrically visible area in the geographic space. For single-point faults, the range circle highlights the scope of influence, while for interval faults, the line segment and range circle together express the extension of the interval and the scope of influence.
[0210] S606. Determine map spatial objects based on the fault center point, the fault range circle, and / or the fault line segment.
[0211] Among them, map spatial objects refer to a collection of geometric objects generated with the fault location as the core and used for map rendering. This step organizes the fault center point, fault range circle and / or fault line segment obtained in the previous step into a map spatial object data package in a unified data format, so that the spatial representation of different fault forms has a consistent structure in the map layer.
[0212] In one embodiment, S606 above, determining the map spatial object based on the fault center point, the fault range circle, and / or the fault line segment, includes the following steps: I. Determine the base coordinates and geometric carrier of the data packet based on the fault type.
[0213] In one implementation, when the fault type is a single-point fault, the base coordinates are taken as the coordinates of the fault center point, and the geometric carrier is taken as the fault range circle centered on the fault center point, excluding the fault line segment. When the fault type is an interval fault, the base coordinates are taken as the coordinates of the fault center point, the interval start point coordinates, and the interval end point coordinates, and the geometric carrier is taken as the fault line segment connecting the interval start point coordinates and the interval end point coordinates, as well as the fault range circle. It should be understood that the composition of the data packet automatically adapts to the fault type. Single-point faults form a point-like spatial object with a point marker as the core, while interval faults form a composite spatial object jointly expressed by line segments and circular areas. The two types of forms maintain consistency in data structure.
[0214] II. Bind traceability parameters and fault type flags to data packets.
[0215] Among them, the traceability parameters refer to the set of parameters that record the basis for generating the coordinates of spatial objects. In one implementation, the traceability parameters include the corresponding line, the original standard mileage, the lateral offset scalar, and the interpolation coefficients. The corresponding line records the line name to which the spatial object belongs, the original standard mileage records the standard target mileage before interpolation calculation, the lateral offset scalar records the offset perpendicular to the track centerline during lateral correction, and the interpolation coefficients record the scaling factor used when interpolating coordinates from the cumulative mileage sequence. The fault type marker is used to indicate single-point faults or section faults, serving as the type basis for filtering surrounding objects and differentiating rendering.
[0216] In one implementation, after the data packet is encapsulated, the map space object is output to the fault map organization step as the constituent data of the core layer of the fault map; the fault type marker is synchronously provided for surrounding object filtering and subsequent display steps for reuse.
[0217] It should be understood that the traceability parameters record the complete conversion basis from mileage to coordinates. The positioning result of any spatial object can be traced back to the original mileage and correction parameters, ensuring the interpretability and verifiability of map spatial objects. The unified data packet format ensures consistent processing of single-point faults and interval faults in the fault map organization, surrounding object linkage and view fitting stages.
[0218] S607. Based on the fault center point, determine the surrounding objects.
[0219] Among them, the surrounding objects refer to the business objects related to fault handling that are selected based on the fault center point and spatial constraints. The types of surrounding objects include point-like surrounding objects, line-like surrounding objects, and work area objects.
[0220] Point-like peripheral objects refer to objects that can be located by a single coordinate point, such as pillars, key equipment, and inspection points; linear peripheral objects refer to objects that extend in a linear manner, such as bridges and cables; work area objects refer to operation and maintenance organizational units that undertake overhead contact line maintenance tasks and divide their responsibilities according to the mileage under their jurisdiction.
[0221] S608. Determine the fault map based on map spatial objects and surrounding objects.
[0222] In one implementation, map spatial objects and surrounding objects are organized into a fault map according to layer relationships. The map spatial objects serve as the core layer, carrying the fault location and scope of influence, while the surrounding objects serve as associated layers, carrying fault context information.
[0223] Once the fault map is determined, it is output to the display step as a direct data source for adaptive view fitting and rendering.
[0224] In one embodiment, the above-mentioned S607, determining surrounding objects based on the fault center point, includes the following steps: S6071. For point-like peripheral objects, calculate the distance between the point-like peripheral object and the fault center point, and determine the point-like peripheral objects whose distance is less than or equal to the preset search radius as the first candidate peripheral objects.
[0225] The preset search radius refers to a configurable distance threshold used to define the scope of surrounding object filtering. For example, the preset search radius can be set to 3 kilometers and can be dynamically adjusted according to the business scenario.
[0226] In one implementation, for point-like surrounding objects such as pillars, key equipment, and detection points, the spherical distance between each object and the fault center point is calculated. Point-like surrounding objects with a spherical distance less than or equal to a preset search radius are identified as first candidate surrounding objects. Here, spherical distance is the shortest distance between two points on the Earth's surface along a great circle, which is closer to the real geographical relationship than the straight-line distance on a plane, and can avoid distance deviations caused by projection distortion, especially in large-scale map scenarios.
[0227] S6072. For linear peripheral objects, calculate the shortest distance from the fault center point to the linear peripheral object, and determine the linear peripheral objects whose shortest distance is not greater than the preset search radius as the second candidate peripheral objects.
[0228] In one implementation, for linear peripheral objects such as bridges and cables, the shortest distance from the fault center point to the corresponding line segment of the linear peripheral object is calculated. Linear peripheral objects whose shortest distance is not greater than a preset search radius are identified as second candidate peripheral objects. It should be understood that linear objects have a large extension range, and their endpoints or midpoints may be far from the fault center, while a segment of a linear object may be adjacent to the fault location. Therefore, using the shortest distance as the criterion can accurately capture linear objects related to the fault space.
[0229] S6073. For work area objects, match the jurisdiction mileage and fault mileage interval of the work area object, and determine the work area objects whose jurisdiction range corresponding to the jurisdiction mileage includes the fault mileage interval as the third candidate surrounding objects.
[0230] Among them, the jurisdictional mileage refers to the mileage range within which the work area is responsible for the maintenance of the overhead contact line. In one implementation, for the selected work area objects, the jurisdictional mileage and fault mileage range of the work area objects are additionally matched, and the work area objects whose jurisdictional range includes the fault mileage range are identified as the third candidate surrounding objects and marked as the work area to which they belong with high priority.
[0231] Optionally, the work area jurisdiction data can be pre-loaded and indexed for quick matching. It should be understood that the work area is the responsible entity for fault handling, and the work area whose jurisdiction includes the fault mileage range is the work area to which this fault belongs. Including it in the surrounding objects with high priority ensures the direct accessibility of fault handling responsibility information.
[0232] Optionally, the screening process simultaneously includes facilities on both the uplink and downlink sides of the line to ensure that facilities on the opposite side that are in the same spatial range as the fault location are not missed.
[0233] S6074. Sort the first candidate surrounding objects, the second candidate surrounding objects, and the third candidate surrounding objects based on distance and / or business priority to determine the candidate surrounding object sequence.
[0234] Here, business priority refers to the pre-defined ranking weights assigned to different types of objects based on their business importance. For example, the business priorities of objects, from highest to lowest, are: critical equipment, work area, inspection point, and ordinary support pillar.
[0235] In one implementation, all candidate surrounding objects are sorted in ascending order according to their distance and business priority to obtain a sequence of candidate surrounding objects. Each candidate surrounding object carries the distance to the fault center point, the attribution mark, and the resource association information. The attribution mark is used to indicate the object's category and relationship, and the resource association information is used for extended data such as ledger information and detection records of the associated objects.
[0236] S6075. If the number of candidate surrounding objects in the candidate surrounding object sequence exceeds the first number, trim the second number of candidate surrounding objects at the end of the sequence to obtain the surrounding objects.
[0237] The first quantity refers to the upper limit of the number of candidate surrounding objects that can be retained in the sequence, and the second quantity refers to the number of candidate surrounding objects that need to be pruned. The second quantity is the difference between the number of candidate surrounding objects and the first quantity.
[0238] For example, when the number of candidate surrounding objects is 60 and the first number is 30, the 30 candidate surrounding objects at the end of the sequence are pruned, that is, candidate surrounding objects that are far away and have low business priority are pruned, and the first 30 candidate surrounding objects are retained as surrounding objects.
[0239] It should be understood that, since candidate objects that are close to each other are placed at the beginning of the sequence during sorting, and candidate objects with high business priority are given a higher position, the end of the sequence is trimmed to prioritize retaining objects that are close to each other and important to the business. This makes the number of surrounding objects displayed in the end controllable and highlights the key points, taking into account both display effect and information integrity.
[0240] Figure 7 This is a flowchart illustrating a map adaptive display process provided in an embodiment of this disclosure. Figure 7 As shown, in one embodiment, the above-described S205, displaying the fault map, includes the following steps: S701. Determine the view fitting carrier from map spatial objects and surrounding objects according to preset priority.
[0241] Here, the view fitting carrier refers to the spatial graphic object that serves as the reference for map view zooming and positioning. Preset priority refers to the pre-defined priority rules used to select the view fitting carrier from multiple spatial objects.
[0242] For example, the preset priorities from high to low are as follows: the fault range circle and fault line segment corresponding to the interval fault are the highest priority carriers; the center point of the single-point fault and high-priority surrounding key facilities are secondary carriers; the entire line serves as the fallback background carrier when there are no fault objects or surrounding objects. The system automatically discards low-priority carriers and uses the highest priority spatial graphics as the view reference boundary.
[0243] It should be understood that the view fitting carrier is determined according to the preset priority, so that the focus range of the map view is automatically matched with the actual spatial form of the fault. The view of the section fault covers the entire fault section, and the view of the single-point fault focuses on the fault location and its key related facilities, so as to avoid the key information being overwhelmed by the fixed display of the entire line.
[0244] S702. Determine the bounding box, zoom level, and map center point based on the view fitting carrier.
[0245] In one embodiment, the above-described S702, determining the bounding box, zoom level, and map center point based on the view fitting carrier, includes the following steps: Ⅰ. Extract the extreme coordinates of the view fitting carrier to generate the initial bounding box.
[0246] Among them, extreme coordinates refer to the maximum and minimum coordinates of the geometry of the view fitting carrier in various directions, such as the latitude and longitude coordinates of the northernmost, southernmost, easternmost, and westernmost points. In one implementation, when the view fitting carrier is the fault range circle or fault line segment of an interval fault, all extreme latitude and longitude coordinates of the fault graphic are extracted to generate a bounding rectangle as the initial bounding box; when the view fitting carrier is the center point of a single-point fault, the initial bounding box is generated by expanding the fault range circle by a small buffer distance.
[0247] II. Obtain the coordinates of key surrounding objects whose business priority is higher than a preset threshold. Widen the initial bounding box based on the coordinates of the key surrounding objects to obtain the target bounding box.
[0248] In one implementation, high-priority key surrounding objects are selected from the surrounding objects based on business priority, such as the work area and core critical equipment to which the high priority belongs. The coordinates of the key surrounding objects are included in the boundary range, and the coordinates of all core objects are merged to obtain the overall minimum enclosing rectangle, thus obtaining the target bounding box. It should be understood that the widening process ensures that the display range focuses on the core fault area while not omitting key objects directly related to fault assessment, avoiding critical facilities from being outside the field of view.
[0249] Ⅲ. Calculate the base scaling level based on the size of the target bounding box, compare the base scaling level with the preset minimum scaling threshold and maximum scaling threshold, and correct the base scaling level based on the comparison results to obtain the target scaling level.
[0250] The minimum and maximum zoom thresholds refer to the upper and lower limits of the system's preset zoom levels. In one implementation, the base zoom level is calculated based on the length and width of the target bounding box. If the base zoom level is greater than the maximum zoom threshold (i.e., the view is too small and details are lost), the target zoom level is locked to the maximum zoom threshold. If the base zoom level is less than the minimum zoom threshold (i.e., the view is too empty), the target zoom level is locked to the minimum zoom threshold. If the base zoom level is between the two thresholds, the base zoom level is determined as the target zoom level.
[0251] Optionally, if the number of surrounding objects exceeds the display threshold, the zoom level remains unchanged, and only the distant low-priority facilities are cropped; if the faulty objects are sparse and the surrounding facilities are few, the view is automatically zoomed in slightly to fill in the nearest pillars or key points, thus stabilizing the display scale.
[0252] IV. Determine the geometric center of the target bounding box as the map center point.
[0253] In one implementation, the geometric center of the target bounding box rectangle is used as the center point for map display; alternatively, if the fault type is an interval fault, the midpoint of the fault interval is taken as the center point for map display to ensure that the fault segment covers the core area of the screen.
[0254] S703 displays fault maps based on bounding boxes, zoom levels, and map center points.
[0255] In one implementation, the fault map is rendered using the map's center point as the view positioning center and the target zoom level as the display ratio, and differentiated layer rendering is performed according to the fault type. For example, when the fault type is a single-point fault, a prominent fault point marker is drawn on the map; when the fault type is an area fault, a semi-transparent circular influence range overlay layer is superimposed; surrounding objects are rendered differently, with high-priority work areas and core critical equipment highlighted at the top, while ordinary support pillars are displayed in a lighter color.
[0256] Optionally, the fault location model and map positioning results can be synchronously output to the page text area. The output includes fields such as a unique fault number, fault occurrence time, fault type, fault location description text, map center point coordinates, zoom level, and surrounding object statistics. This output data is also used by multiple front-end components, including fault card display, map description area updates, detail title population, and map marker rendering. It should be understood that the text display and map display share the same data source, ensuring consistency between the text and map display perspectives. This avoids situations where the text displays a certain station area while the map focuses on another route, increasing user trust in the positioning results.
[0257] In one embodiment, S701, determining the view fitting carrier from map spatial objects and surrounding objects according to a preset priority includes: Ⅰ. When there is a fault line segment in the fault space object, the fault range circle and the fault line segment are determined as the view fitting carrier.
[0258] In one implementation, the fault line segment is a unique geometric carrier of interval faults. The existence of a fault line segment indicates that the current fault is an interval fault. The fault range circle and the fault line segment as a whole are determined as the view fitting carrier. Subsequent boundary box calculations are based on the overall range of the fault segment, so that the view completely covers the fault segment.
[0259] II. In the absence of fault line segments in the fault space object, the fault center point and key surrounding objects with business priority higher than the preset threshold are determined as the view fitting carriers.
[0260] In one implementation, the absence of a fault line segment indicates that the current fault is a single-point fault. The fault center point and key surrounding objects selected according to business priority are jointly determined as the view fitting carrier, such as the work area and core key equipment, so that the view can focus on the fault location while accommodating key facilities directly related to the fault assessment.
[0261] III. In the case where both the fault space object and the surrounding objects are missing, the trajectory of the entire line is determined as the catch-all background carrier.
[0262] In one implementation, when both the fault space object and its surrounding objects are missing, an effective fault view range cannot be formed. In this case, the trajectory range of the entire line is used as the view fitting carrier to ensure that the map view always has a definite display benchmark.
[0263] In one implementation, when multiple candidate carriers coexist, low-priority carriers are discarded according to a preset priority, and the spatial graphic with the highest priority is used as the view reference boundary.
[0264] It should be understood that the view fitting carrier is determined according to the preset priority, so that the focus range of the map view is automatically matched with the actual spatial form of the fault. The view of the section fault covers the entire fault section, and the view of the single-point fault focuses on the fault location and its key related facilities, so as to avoid the key information being overwhelmed by the fixed display of the entire line.
[0265] like Figure 8 The diagram shown is a structural schematic of a contact network fault location and map adaptive display device based on fault text location parsing and spatial constraints provided in an embodiment of this application. The device may include: an acquisition module 801, a processing module 802, a resolution module 803, a generation module 804, and a display module 805.
[0266] The acquisition module 801 is used to execute Figure 2 In the illustrated method, the operation of S201 is performed by the processing module 802; the digestion module 803 is used to perform the operation of S203; the generation module 804 is used to perform the operation of S204; and the display module 805 is used to perform the operation of S205.
[0267] In some embodiments, the device includes hardware structures and / or software modules corresponding to the execution of each function in order to achieve the above-described functions. Those skilled in the art will readily recognize that, based on the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution.
[0268] The module division in this embodiment is illustrative and represents only one logical functional division. In actual implementation, there may be other division methods.
[0269] Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 9 As shown, the electronic device 900 includes, but is not limited to, a processor 901 and a memory 902.
[0270] The memory 902 stores executable instructions of the processor 901; the processor 901 is configured to execute instructions to implement the methods in the above embodiments. The processor 901 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices.
[0271] The memory 902 may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage, flash memory, read-only memory (ROM), etc.
[0272] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the methods described in the above-described method embodiments. This application also provides a computer program product, including a computer program, which, when executed by a processor, implements the methods described in the above-described method embodiments.
[0273] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered 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 method for catenary fault location and adaptive map display based on fault text location parsing and spatial constraints, characterized in that, The method includes: In response to receiving a fault event, the system obtains the original fault location text and basic context information, performs knowledge extraction on the original fault location text, and determines the fault location elements. Based on the fault location elements, knowledge processing is performed to determine the fault location model. The fault location model is a structured model obtained by classifying and filling the fault location elements into the fields of line, power supply, mileage, spatial representation and credibility. Based on the fault location model, conflict resolution analysis is performed on the fault location elements to determine the target fault location information, which includes single points or interval mileage. Based on the target fault location information, a fault map corresponding to the fault event is determined, and the fault map includes at least map spatial objects and surrounding objects; The fault map is then displayed.
2. The contact network fault location and adaptive map display method based on fault text location parsing and spatial constraints according to claim 1, characterized in that, The step of performing knowledge extraction on the original fault location text to determine fault location elements includes: Perform text standardization preprocessing on the original fault location text to obtain standardized text; Load line operation semantic templates and power supply operation and maintenance semantic templates; the line operation semantic templates include combination rules for line name, line type or track, station or section, and kilometer marker; the power supply operation and maintenance semantic templates include combination rules for substation, feeder number, and kilometer marker. Based on the standardized text and the basic context information, the standardized text is structured and parsed using the line operation semantic template and the power supply operation and maintenance semantic template to extract fault location elements.
3. The contact network fault location and adaptive map display method based on fault text location parsing and spatial constraints according to claim 1, characterized in that, The step of performing knowledge processing based on the fault location elements to determine the fault location model includes: Initialize the empty model carrier, and create line dimension container, power supply dimension container, mileage dimension container, spatial representation container and credibility container; Semantic standardization processing is performed on the fault location elements to map the original descriptions to standard names, and auxiliary metadata is bound to each fault location element; the auxiliary metadata includes at least the source, conflict marker, and comprehensive score. The processed fault location elements are filled into the corresponding containers according to their respective dimensions. Among them, the line dimension container should at least fill in the line name and station area, the power supply dimension container should at least fill in the substation and feeder number, the mileage dimension container should at least fill in the single-value kilometer mark or the start and end interval mileage, and the spatial expression container should fill in the single point or interval segment identifier. Perform cross-dimensional cross-validation on the fields in the line dimension container, the power supply dimension container, and the mileage dimension container to identify conflicting models; The conflicting models are marked to obtain the fault location model.
4. The contact network fault location and adaptive map display method based on fault text location parsing and spatial constraints according to claim 1, characterized in that, The step of performing conflict resolution analysis on the fault location elements based on the fault location model to determine target fault location information including single points or interval mileage includes: Extract the line field from the fault location model, and perform a first-level resolution on the fault location elements to retain the candidate set under the same line; When both the station area field and the substation or feeder field exist simultaneously, a second-level elimination is performed on the candidate set based on the power supply affiliation to obtain the eliminated candidate set. A third-level resolution is performed on the resolved candidate set to merge multiple discrete kilometer marker candidates into a single mileage interval, so that the pillar with the smallest difference from the target analytical mileage is determined as the optimal candidate; Based on the mileage pattern of the optimal candidate, the target fault location information, including single-point or interval mileage, is determined.
5. The contact network fault location and adaptive map display method based on fault text location parsing and spatial constraints according to claim 1, characterized in that, The step of determining the fault map corresponding to the fault event based on the target fault location information includes: Based on the target fault location information, the fault type is determined to be either a single-point fault or an area fault. Load the trajectory point sequence and cumulative mileage sequence of the line corresponding to the target fault location information; Based on the cumulative mileage sequence, interpolation calculations are performed on the single-point or interval mileage to obtain the reference coordinates of the track centerline. Lateral correction is performed based on the unit normal vector and lateral offset distance of the trajectory segment where the reference coordinates of the track centerline are located, to obtain the fault center point; Based on the fault type, determine the corresponding fault range circle and / or fault line segment; Based on the fault center point, the fault range circle, and / or the fault line segment, determine the map spatial object; Based on the fault center point, determine the surrounding objects; The fault map is determined based on the map spatial object and the surrounding objects.
6. The contact network fault location and adaptive map display method based on fault text location parsing and spatial constraints according to claim 5, characterized in that, The step of interpolating the single-point or interval mileage based on the cumulative mileage sequence to obtain the reference coordinates of the track centerline includes: The entire route is divided into multiple curve units based on the curvature difference between adjacent trajectory points, and each curve unit is bound to the start and end cumulative mileage and curve control points; A target curve unit containing the single point or interval mileage is determined in the curve unit, and quadratic curve interpolation is performed on the single point or interval mileage based on the target curve unit to obtain the reference coordinates of the track centerline.
7. The contact network fault location and adaptive map display method based on fault text location parsing and spatial constraints according to claim 5, characterized in that, The process of determining the lateral offset distance includes: Construct a lateral offset feature database that stores the correspondence between line type, row, section curvature level, standard lateral offset and historical positioning error mean; Using line type, row type, and section curvature level as search criteria, and the fault type as filtering criteria, the standard lateral offset is matched in the lateral offset feature database, and adaptive adjustment is performed in combination with the current section type to obtain the lateral offset distance.
8. The contact network fault location and adaptive map display method based on fault text location parsing and spatial constraints according to claim 5, characterized in that, The types of surrounding objects include: point-like surrounding objects, line-like surrounding objects, and work area objects; determining surrounding objects based on the fault center point includes: For the point-like surrounding objects, calculate the distance between the point-like surrounding objects and the fault center point, and determine the point-like surrounding objects whose distance is less than or equal to the preset search radius as the first candidate surrounding objects; For the linear peripheral objects, calculate the shortest distance from the fault center point to the linear peripheral objects, and determine the linear peripheral objects whose shortest distance is not greater than the preset search radius as the second candidate peripheral objects; For the work area object, the jurisdiction mileage and fault mileage interval of the work area object are matched, and the work area object whose jurisdiction range corresponding to the jurisdiction mileage includes the fault mileage interval is determined as the third candidate surrounding object. The first candidate surrounding objects, the second candidate surrounding objects, and the third candidate surrounding objects are sorted based on distance and / or business priority to determine the candidate surrounding object sequence; When the number of candidate surrounding objects in the candidate surrounding object sequence exceeds a first number, a second number of candidate surrounding objects at the end of the sequence are trimmed to obtain the surrounding objects.
9. The contact network fault location and adaptive map display method based on fault text location parsing and spatial constraints according to claim 1, characterized in that, The process of displaying the fault map includes: A view fitting carrier is determined from the map spatial objects and the surrounding objects according to a preset priority; the bounding box, zoom level and map center point are determined based on the view fitting carrier; The fault map is displayed based on the bounding box, the zoom level, and the map center point.
10. The contact network fault location and adaptive map display method based on fault text location parsing and spatial constraints according to claim 9, characterized in that, The process of determining the bounding box, zoom level, and map center point based on the view fitting carrier includes: Extract the extreme coordinates of the view fitting carrier to generate an initial bounding box; Obtain the coordinates of key surrounding objects whose business priority is higher than a preset threshold among the surrounding objects, and widen the initial bounding box according to the coordinates of the key surrounding objects to obtain the target bounding box; The base scaling level is calculated based on the size of the target bounding box. The base scaling level is compared with the preset minimum scaling threshold and maximum scaling threshold. The base scaling level is corrected based on the comparison result to obtain the target scaling level. The geometric center of the target bounding box is determined as the center point of the map.
