A method and system for laser additive management of rail defects

Abstract

The present application relates to the technical field of rail maintenance, and provides a rail defect laser additive management method and system, the method comprising: after the first laser additive repair of the rail is completed, establishing a regional health record for the repair area of the rail and determining a three-dimensional space anchor point of the regional health record; storing the repair information of the first laser additive repair as the initial content of the regional health record; identifying a newly discovered defect of the rail and obtaining three-dimensional position information of the newly discovered defect; judging whether the three-dimensional position information and the three-dimensional space anchor point in the regional health record exist spatial overlap, obtaining a spatial overlap judgment result; if the spatial overlap judgment result indicates that there is spatial overlap, the repair information of the newly discovered defect is added as additional content of the regional health record for additional supplement. The present application has the effect of improving the scientificity and efficiency of rail maintenance.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of rail maintenance, and in particular to a rail defect laser additive management method and system. BACKGROUND

[0002] In railway operation and maintenance work, as the key infrastructure for carrying train operation, rails have been subjected to huge dynamic loads and environmental erosion for a long time, and thus various forms of damage, such as surface peeling, fatigue cracks and wear, etc., are inevitable. In order to ensure the safety and efficiency of train operation, it is crucial to repair these damages in a timely and effective manner.

[0003] The rail defect detection system cannot directly determine whether the second appearing crack is related to the boundary effect of the first repair, nor can it automatically track the cumulative impact of the second repair on the material performance of the first repair area. It only faithfully records two independent repair operations without considering them as part of a continuous and evolving repair history. This "event isolation" data storage mode makes it impossible for the system to construct a "health evolution path" of the rail after multiple repairs. This data management approach brings significant limitations when conducting in-depth analysis. When data analysts attempt to assess the overall health of a section of rail or predict its remaining service life, they find it difficult to effectively integrate and interpret the data from these "iterative" or "adjacent" repairs. This lack of complex relationships between repair events makes long-term performance predictions and repair strategy optimization based on historical data inaccurate and even potentially leading to incorrect decisions, thus failing to truly achieve the goal of using accumulated data to guide rail life cycle health management. The existing system cannot extract continuous and logically inherent rail health evolution rules from isolated events, which severely restricts the value of data in improving repair processes and extending the service life of rails.

[0004] In view of the above problems, the prior art needs to be improved. SUMMARY

[0005] The present application discloses a rail defect laser additive management method and system, aiming to solve the problem of lack of effective management of the historical health status of the repair area in the existing rail defect laser additive repair process, leading to fragmentation of repair information and difficulty in achieving long-term, dynamic and accurate tracking and management of the rail repair area.

[0006] The technical solution of the present application is as follows:

[0007] In a first aspect, the present application discloses a rail defect laser additive management method, specifically comprising:

[0008] After the first laser additive repair of the steel rail is completed, a regional health record of the repair area of the steel rail is established, and a three-dimensional space anchor point of the regional health record is determined;

[0009] The repair information of the first laser additive repair is stored as initial content of the regional health record;

[0010] A newly discovered defect of the steel rail is identified, and three-dimensional position information of the newly discovered defect is obtained;

[0011] It is judged whether the three-dimensional position information and the three-dimensional space anchor point in the regional health record exist spatial overlap, to obtain a spatial overlap judgment result;

[0012] If the spatial overlap judgment result indicates that there is spatial overlap, the repair information of the newly discovered defect is added as additional content of the regional health record.

[0013] Through the technical scheme, the present application can establish the health record of the repair area of the steel rail, and realize accurate positioning of the repair area and dynamic management of historical information based on the three-dimensional space anchor point, effectively solving the problem of fragmentation of repair information and difficulty in tracing in the prior art, and providing data support for long-term maintenance of the steel rail.

[0014] Further, on the basis described above, the present application further proposes that the step of judging whether the three-dimensional position information and the three-dimensional space anchor point in the regional health record exist spatial overlap, to obtain a spatial overlap judgment result, comprises:

[0015] The laser output power, the laser scanning speed, the powder feeding rate and the preheating temperature of the first laser additive repair are obtained;

[0016] According to the laser output power, the laser scanning speed, the powder feeding rate and the preheating temperature of the first laser additive repair, a three-dimensional ellipsoid space anchor point is calculated and determined as the three-dimensional space anchor point of the regional health record;

[0017] It is judged whether the three-dimensional position information and the three-dimensional ellipsoid space anchor point exist spatial overlap, to obtain a spatial overlap judgment result.

[0018] Through the technical scheme, the present application uses the process parameters of laser additive repair to accurately calculate the three-dimensional ellipsoid space anchor point, so that the determination of the space anchor point is more scientific and accurate, and the accuracy of subsequent spatial overlap judgment is improved, thereby more effectively managing the repair area.

[0019] Further, in some preferred embodiments, the step of judging whether the three-dimensional position information and the three-dimensional ellipsoid space anchor point exist spatial overlap, to obtain a spatial overlap judgment result, comprises:

[0020] Based on the three-dimensional position information, three-dimensional point cloud data of the newly discovered defect is obtained;

[0021] performing local geometric feature analysis on the three-dimensional point cloud data to obtain a geometric boundary feature point set of the newly discovered defect;

[0022] constructing a minimum envelope geometric body of the newly discovered defect based on the geometric boundary feature point set;

[0023] judging whether the minimum envelope geometric body and the three-dimensional ellipsoid space anchor point exist spatial overlap to obtain a spatial overlap judgment result.

[0024] Through the technical scheme, the application performs fine analysis on the three-dimensional point cloud data of the newly discovered defect, constructs a minimum envelope geometric body, makes the spatial overlap judgment more accurate, avoids the error that may be caused by simple point position judgment, and improves the defect recognition and management accuracy.

[0025] As an optional scheme, the step of performing local geometric feature analysis on the three-dimensional point cloud data to obtain a geometric boundary feature point set of the newly discovered defect includes:

[0026] performing local region division on the three-dimensional point cloud data to obtain a plurality of local point cloud regions;

[0027] calculating local geometric feature parameters of each point in each local point cloud region;

[0028] based on the local geometric feature parameters, identifying abnormal points in the local point cloud region that are different from the preset rail healthy surface point cloud feature;

[0029] performing spatial connectivity analysis on the abnormal points, and clustering the spatially connected abnormal points as a defect candidate region;

[0030] extracting points on the spatial boundary of each defect candidate region to obtain a geometric boundary feature point set of the newly discovered defect.

[0031] Through the technical scheme, the application can more accurately extract the geometric boundary feature points of the defect from the point cloud data through local geometric feature analysis and abnormal point identification, provides reliable basic data for subsequent construction of a minimum envelope geometric body, and improves the accuracy of defect boundary identification.

[0032] On the basis described above, the application further proposes a step of adding repair information of the newly discovered defect as additional content of the regional health record, including:

[0033] collecting repair process data of the newly discovered defect; the repair process data includes three-dimensional topographic data, laser power, scanning speed, powder feeding rate, preheating temperature, and quality detection results after repair completion;

[0034] Based on the repair process data, a repair version data packet corresponding to the newly discovered defect is generated, and a unique version number and a timestamp are assigned to the repair version data packet;

[0035] The repair version data packet is added as additional content to the repair history list of the corresponding area health record;

[0036] Based on the unique version number and the timestamp of the repair version data packet, the latest version identifier and the last update time of the area health record are updated.

[0037] Through the technical scheme, the repair version data packet with a unique version number and a timestamp is generated, and it is added to the area health record, which realizes fine management and version control of the repair history, and ensures the integrity and traceability of the repair information.

[0038] In some preferred embodiments, the step of extracting points on the spatial boundary of each defect candidate region comprises:

[0039] Local surface reconstruction is performed on the point cloud data in the defect candidate region to obtain a local surface mesh;

[0040] On the local surface mesh, edge points with curvature changes exceeding a preset threshold or with abrupt changes in normal direction are identified as abnormal edge points;

[0041] The abnormal edge points are subjected to connectivity analysis, and isolated abnormal edge points are removed, and a set of connected abnormal edge points is retained;

[0042] Based on the retained set of abnormal edge points, points on the spatial boundary of the defect candidate region are extracted.

[0043] Through the technical scheme, the edge points of the defect region can be more accurately identified through local surface reconstruction and curvature / normal analysis, and connectivity analysis is performed, effectively excluding noise points, and improving the accuracy and robustness of defect boundary extraction.

[0044] To enhance the function, the step of performing local surface reconstruction on the point cloud data in the defect candidate region to obtain a local surface mesh comprises:

[0045] Spatial neighborhood search is performed on the point cloud data in the defect candidate region to determine the neighborhood point set of each point;

[0046] According to the neighborhood point set of each point, the local weighted average normal direction and the local density of each point are calculated;

[0047] Based on the local weighted average normal direction and the local density, the points in the unevenly distributed region in the defect candidate region are interpolated to generate a supplementary point cloud;

[0048] The original point cloud data in the fusion defect candidate area and the supplementary point cloud are fused to form enhanced point cloud data;

[0049] Based on the enhanced point cloud data, local surface reconstruction is performed to obtain a local surface mesh.

[0050] Through the technical solution, the application generates supplementary point cloud by interpolating points in the unevenly distributed area, effectively solves the problem of sparse or unevenly distributed point cloud data, makes the local surface reconstruction more complete and accurate, and provides high-quality surface mesh for subsequent edge point recognition.

[0051] On the basis of the above, the application further proposes that the step of performing spatial neighborhood search on the point cloud data in the defect candidate area to determine the neighborhood point set of each point comprises:

[0052] Performing local density estimation on the point cloud data in the defect candidate area to obtain the local point density of each point;

[0053] According to the local point density, dynamically adjusting the parameters of the spatial neighborhood search;

[0054] Based on the adjusted parameters of the spatial neighborhood search, performing spatial neighborhood search on each point to determine the neighborhood point set of each point.

[0055] Through the technical solution, the application can adaptively perform neighborhood search according to the local density of the point cloud by dynamically adjusting the parameters of the spatial neighborhood search, thereby improving the accuracy and efficiency of determining the neighborhood point set, and is especially suitable for point cloud data with uneven density.

[0056] In some preferred embodiments, the step of performing local density estimation on the point cloud data in the defect candidate area to obtain the local point density of each point comprises:

[0057] Performing scan blind area identification on the point cloud data in the defect candidate area to determine the area where the scan blind area or the occlusion exists;

[0058] Performing local outlier detection on the peripheral non-blind area point cloud data of the area where the scan blind area or the occlusion exists to obtain an outlier;

[0059] Removing the outlier to obtain peripheral non-blind area point cloud data after removing the outlier;

[0060] Performing local smoothing processing on the peripheral non-blind area point cloud data after removing the outlier to obtain smoothed peripheral non-blind area point cloud data;

[0061] Based on the smoothed peripheral non-blind area point cloud data, in the corresponding scan blind area or occluded area, according to the interpolation rule of the local geometric feature, a supplementary point cloud is generated;

[0062] The original point cloud data and supplementary point cloud data within the defect candidate region are merged to form complete point cloud data;

[0063] Based on complete point cloud data, local density estimation is performed on each point to obtain the local point density of each point.

[0064] Through this technical solution, this application identifies and processes scanning blind spots and occluded areas, removes outliers and smooths data to generate supplementary point clouds, and finally forms complete point cloud data, thereby making local density estimation more accurate and reliable and overcoming the possible defects of the original point cloud data.

[0065] Secondly, this application also discloses a rail defect laser additive manufacturing management system for performing rail defect laser additive manufacturing management, the system comprising:

[0066] The spatial anchor point determination module is used to establish a regional health history for the repaired area of ​​the rail after the first laser additive repair is completed, and to determine the three-dimensional spatial anchor points of the regional health history.

[0067] The regional history storage module is used to store the repair information of the first laser additive repair as the initial content of the regional health history.

[0068] The location information acquisition module is used to identify newly discovered defects in the rail and acquire the three-dimensional location information of the newly discovered defects;

[0069] The spatial overlap judgment module is used to determine whether there is spatial overlap between the three-dimensional location information and the three-dimensional spatial anchor points in the regional health record, and to obtain the spatial overlap judgment result.

[0070] The supplementary content module is used to add the repair information of newly discovered defects as supplementary content to the regional health record if the spatial overlap judgment result indicates that spatial overlap exists.

[0071] Through this system, this application enables automated and systematic management of rail repair areas. The modular design improves the system's scalability and maintainability, providing an efficient tool for the long-term health management of rails.

[0072] Beneficial Effects: The laser additive manufacturing management method for rail defects disclosed in this application establishes a regional health history and determines three-dimensional spatial anchor points for the repaired area after the initial laser additive repair of the rail, storing the initial repair information as initial content. When a new defect is discovered, its three-dimensional location information is obtained, and it is determined whether there is spatial overlap with the three-dimensional spatial anchor points in the regional health history. If there is overlap, the repair information of the new defect is added to the health history as supplementary content. This method effectively solves the problems of fragmented and difficult-to-trace rail repair information in the prior art. By establishing a regional health history with three-dimensional spatial anchor points as the core, this application can achieve precise positioning of the rail repair area and dynamic management of historical information, ensuring that the data of each repair can be effectively linked and traced, thereby providing comprehensive and continuous data support for the long-term maintenance and condition assessment of rails, significantly improving the scientific nature and efficiency of rail maintenance. Attached Figure Description

[0073] Figure 1 This is a flowchart of a method for managing rail defects using laser additive manufacturing, as described in one embodiment of the present invention.

[0074] Figure 2 This is a flowchart of the sub-steps of a laser additive manufacturing method for managing rail defects in another embodiment of the present invention;

[0075] Figure 3 This is a system block diagram of a laser additive management system for rail defects according to another embodiment of the present invention;

[0076] Explanation of reference numerals in the attached figures:

[0077] 1. Rail Defect Laser Additive Management System; 11. Spatial Anchor Point Determination Module; 12. Regional History Storage Module; 13. Location Information Acquisition Module; 14. Spatial Overlap Judgment Module; 15. Supplementary Content Module. Detailed Implementation

[0078] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0079] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0080] Traditional rail defect repair management methods lack a systematic mechanism for tracing repair history and effectively managing subsequent defects after the initial laser additive repair. This leads to fragmented repair information, making it difficult to perform correlation analysis on multiple repairs of the same area, thus affecting the optimization of repair strategies and the efficiency of rail lifecycle health management. Failure to address these issues could result in repeated repairs, difficulty in assessing repair quality, and increased maintenance costs, severely hindering the further promotion and application of laser additive technology in rail maintenance.

[0081] In response, this application proposes a laser additive manufacturing management method for rail defects, combining... Figure 1 As shown, it includes:

[0082] S1. After the first laser additive repair of the rail is completed, a regional health history is established for the repaired area of ​​the rail, and the three-dimensional spatial anchor points of the regional health history are determined.

[0083] S2, store the repair information of the first laser additive repair as the initial content of the regional health record;

[0084] S3 identifies newly discovered defects in the rails and obtains the three-dimensional location information of the newly discovered defects;

[0085] S4, determine whether there is spatial overlap between the three-dimensional location information and the three-dimensional spatial anchor points in the regional health record, and obtain the spatial overlap judgment result;

[0086] S5. If the spatial overlap judgment result indicates that there is spatial overlap, the repair information of the newly discovered defects will be added as supplementary content to the regional health record.

[0087] Specifically, a "regional health record" can be a digital archive that records all relevant information about a specific repair area of ​​the rail, including but not limited to detailed data from the initial repair, information on subsequent defects discovered, and the detailed process and results of each repair. This record is dynamically updated, and its content will be continuously enriched and improved as the rail is operated and maintained.

[0088] "Three-dimensional spatial anchor points" are precise mappings of regional health records in physical space. They are entities with clearly defined three-dimensional coordinates and geometric extents, used to identify and define the rail repair area corresponding to the regional health record. Through three-dimensional spatial anchor points, it is possible to accurately determine whether newly discovered defects are located within already repaired areas, thereby deciding whether to add the repair information of the new defects to the existing record.

[0089] The working environment of this method typically includes: laser additive repair equipment, high-precision 3D scanning equipment, a data acquisition and processing system, and a back-end data management platform. The laser additive repair equipment performs the rail repair work; the high-precision 3D scanning equipment acquires precise 3D topographic data of the rail defects and the repair area; the data acquisition and processing system collects, processes, and analyzes various parameters and test results during the repair process; and the back-end data management platform stores, manages, and queries all regional health records and related data.

[0090] The main feature of the laser additive management method for rail defects in this application is the refined management and historical traceability of the rail repair area.

[0091] First, after the rail completes its initial laser additive repair, a regional health record is established for the repaired area of ​​the rail, and the three-dimensional spatial anchor points of the regional health record are determined.

[0092] For example, once a section of a rail is repaired for the first time using laser additive manufacturing technology, a unique "regional health record" can be immediately created for that repaired area. This record can be a separate data file, a database record, or a digital archive associated with the physical location of the rail. Simultaneously, the three-dimensional spatial anchor points of this regional health record need to be determined. These anchor points can be the geometric center point of the repaired area and its radius, the smallest bounding rectangle or ellipsoid of the repaired area, or complete point cloud data of the repaired area obtained through 3D scanning and used as the 3D spatial anchor points. This anchor point information will serve as the unique spatial identifier for the regional health record, used for subsequent spatial positioning and overlap determination.

[0093] Secondly, the repair information from the first laser additive repair is stored as the initial content of the regional health record.

[0094] After the regional health record is established, all relevant information generated from the first laser additive repair, such as the defect type and size before repair, process parameters such as laser power, scanning speed, powder feeding rate, and preheating temperature used during repair, quality inspection results after repair (such as hardness, metallographic structure, residual stress, etc.), and metadata such as repair personnel and repair time, will be collected and stored as the initial content of the regional health record. This information can be stored in a database in a structured data format (such as JSON or XML) or in a file system and linked to the regional health record.

[0095] Next, newly discovered defects in the rails are identified, and their three-dimensional location information is obtained.

[0096] After rails have been in use for a period of time, new defects may appear in or near the repaired areas. These newly discovered defects can be identified in various ways, such as through manual inspections and the use of handheld 3D scanners to obtain their approximate location information; or through periodic scanning by automated inspection equipment (such as laser scanning systems or machine vision systems mounted on rail inspection vehicles) to automatically identify defects and obtain their precise 3D location information. This 3D location information can include the defect's center coordinates, geometric dimensions, shape features, etc., and is usually represented in the form of point cloud data, 3D models, or geometric parameters.

[0097] Next, it is determined whether there is spatial overlap between the three-dimensional location information and the three-dimensional spatial anchor points in the regional health record, and the spatial overlap judgment result is obtained.

[0098] After obtaining the 3D location information of the newly discovered defect, it needs to be compared with the 3D spatial anchor points in the established regional health record to determine whether there is any spatial overlap. For example, if the 3D spatial anchor point is an ellipsoid, the distance from the center point of the newly discovered defect to the center of the ellipsoid can be calculated, and it can be determined whether this distance is less than the semi-axis length of the ellipsoid, or whether the minimum envelope of the newly discovered defect intersects with the ellipsoid. If the 3D spatial anchor point is point cloud data, a point cloud registration algorithm can be used to determine the degree of overlap between the new defect point cloud and the anchor point point cloud. In this way, a spatial overlap judgment result can be obtained, indicating whether overlap exists.

[0099] Finally, if the spatial overlap determination result indicates that spatial overlap exists, the repair information of the newly discovered defects will be added as supplementary content to the regional health record.

[0100] If the assessment results indicate that the 3D location information of a newly discovered defect spatially overlaps with the 3D spatial anchor point of a region's health record, this means that the new defect may be located in or adjacent to a previously repaired area. In this case, after repairing the new defect, its repair information will be added as supplementary content to the corresponding region's health record. This ensures that multiple repair histories of the same region can be completely recorded and traced, forming a continuous and comprehensive health profile. For example, the repair information for the new defect can be used to generate a new version data package and added to the repair history list of the region's health record, while simultaneously updating the latest version identifier and last update time of the record.

[0101] The laser additive management method for rail defects proposed in this application achieves refined and traceable management of rail repair areas by introducing the concepts of regional health history and three-dimensional spatial anchor points.

[0102] Specifically, after the rail undergoes its first laser additive repair, the system immediately establishes a unique regional health record for the repaired area and precisely identifies its three-dimensional spatial anchor point. This three-dimensional spatial anchor point is the unique identifier of the repaired area in physical space, accurately defining its scope and location. Subsequently, all detailed information about the first repair, including the defect condition before repair, process parameters during repair (such as laser power, scanning speed, powder feeding rate, preheating temperature, etc.), and post-repair quality inspection results, is stored completely in this regional health record as initial content. This lays the foundation for subsequent traceability and analysis.

[0103] As the rails continue to operate, new defects may appear. When the system detects a newly discovered defect on the rail, it immediately acquires its precise three-dimensional location information. Subsequently, a crucial judgment step is performed: the system determines whether the three-dimensional location information of the newly discovered defect spatially overlaps with the three-dimensional spatial anchor points in the established area health history. This judgment process is intelligent; it can accurately identify whether the new defect is located in or adjacent to a previously repaired area.

[0104] If the assessment results indicate spatial overlap, it means that the newly discovered defect is associated with a previously repaired area. In this case, after the new defect is repaired, its repair information is not simply stored as a new, independent record, but rather added as supplementary content to the corresponding area's health record. In this way, multiple repair histories of the same repaired area are organically integrated, forming a continuous and complete health profile.

[0105] Therefore, this method effectively solves the problems of fragmented and difficult-to-trace repair information in traditional rail maintenance. By linking each repair to a specific spatial area and historical record, maintenance personnel can clearly understand the evolution of the "health status" of each area of ​​the rail, including defect recurrence and the effectiveness of different repair strategies. This not only improves the efficiency and accuracy of rail maintenance but also provides valuable data support for optimizing laser additive repair processes and extending the service life of rails.

[0106] Optional, combined Figure 2 As shown, the steps in S4 to determine whether there is spatial overlap between the three-dimensional location information and the three-dimensional spatial anchor points in the regional health record, and to obtain the spatial overlap determination result, include:

[0107] S41, obtain the laser output power, laser scanning speed, powder feeding rate and preheating temperature of the first laser additive repair;

[0108] S42, based on the laser output power, laser scanning speed, powder feeding rate and preheating temperature of the first laser additive repair, calculate and determine the three-dimensional ellipsoidal spatial anchor point as the three-dimensional spatial anchor point of the regional health record.

[0109] S43 determines whether the three-dimensional position information overlaps with the spatial anchor point of the three-dimensional ellipsoid, and obtains the spatial overlap judgment result.

[0110] Specifically, after the initial laser additive repair of the rail, in order to more accurately define the spatial extent of the repair area, it is necessary to obtain the process parameters directly related to this repair process. These parameters include laser output power, laser scanning speed, powder feeding rate, and preheating temperature. These parameters directly affect the size and shape of the laser molten pool and the material properties after solidification, thus determining the actual geometry and physical boundaries of the additive repair area.

[0111] The three-dimensional ellipsoidal spatial anchor point can be understood as a geometric model that can accurately characterize the spatial range of the laser additive repair area. Parameters such as the major axis, minor axis, and center position of this ellipsoid can be calculated and determined using process parameters obtained above, such as laser output power, laser scanning speed, powder feeding rate, and preheating temperature. For example, laser output power and scanning speed may affect the depth and width of the molten pool, powder feeding rate affects the deposition height, and preheating temperature affects the melting and solidification behavior of the material. The combined effect of these factors results in the repair area exhibiting an approximately ellipsoidal spatial distribution. By establishing a mapping relationship between these parameters and the geometric characteristics of the ellipsoid, a three-dimensional spatial anchor point that highly matches the actual repair area can be obtained.

[0112] In practical applications, determining whether 3D location information overlaps with the spatial anchor points of a 3D ellipsoid involves performing spatial geometric operations on the newly discovered defect's 3D location information (e.g., the defect's center coordinates or the geometric information of its envelope) and the already determined spatial anchor points of the 3D ellipsoid. This can be achieved by calculating whether the defect's location point falls inside the ellipsoid, or whether the defect's smallest envelope intersects with the ellipsoid. From this, a clear result can be obtained: either spatial overlap exists or it does not.

[0113] In some preferred embodiments, a specific example is given below. Assume that during a first laser additive repair of a rail, the recorded laser output power is 3000W, the laser scanning speed is 10mm / s, the powder feeding rate is 15g / min, and the preheating temperature is 200℃. Based on these process parameters, the system can use a pre-established process parameter-geometric dimension mapping model to calculate the approximate dimensions of the repair area. For example, the model might indicate that under these parameters, the repair area extends approximately 50mm in the length direction, approximately 10mm in the width direction, and approximately 5mm in the depth direction. Thus, a three-dimensional ellipsoidal spatial anchor point centered at the geometric center of the repair area can be determined, with semi-axial lengths of 25mm (length direction), 5mm (width direction), and 2.5mm (depth direction). When a new defect is subsequently detected on the rail and its three-dimensional position information is obtained (e.g., the defect center coordinates are (X,Y,Z)), the system will determine whether the defect coordinates fall within the calculated three-dimensional ellipsoid. If the defect falls within the area, spatial overlap is identified, and the repair information for the new defect is added to the corresponding area's health history. This precise anchor point definition based on actual process parameters makes the tracing and management of rail defects more scientific and accurate.

[0114] Optionally, the steps to determine whether the three-dimensional position information overlaps with the spatial anchor points of the three-dimensional ellipsoid and to obtain the spatial overlap determination result include:

[0115] Based on three-dimensional location information, acquire three-dimensional point cloud data of newly discovered defects;

[0116] Local geometric feature analysis is performed on 3D point cloud data to obtain a set of geometric boundary feature points of newly discovered defects;

[0117] Based on the set of geometric boundary feature points, construct the minimum envelope geometry of the newly discovered defects;

[0118] Determine whether there is spatial overlap between the minimum envelope geometry and the spatial anchor points of the three-dimensional ellipsoid, and obtain the spatial overlap determination result.

[0119] Specifically, acquiring 3D point cloud data of newly discovered defects based on 3D location information refers to using 3D scanning equipment (such as laser scanners, structured light scanners, or industrial CT scanners) to scan the surface area of ​​the rail where the newly discovered defect is located, thereby collecting a series of discrete 3D points. These points together constitute the 3D point cloud data of the defect. This 3D point cloud data can more comprehensively and accurately characterize the actual shape and spatial distribution of the defect.

[0120] This process involves performing local geometric feature analysis on 3D point cloud data to obtain a set of geometric boundary feature points for newly discovered defects. This can be understood as identifying points located on the edges or boundaries of defects by calculating the local curvature, normal direction changes, density changes, and other geometric attributes of each point in the point cloud. These points typically exhibit geometric features different from those inside or outside the defect, such as abrupt changes in curvature or sharp changes in normal direction. The aim is to accurately extract the contour information of defects from large amounts of point cloud data.

[0121] In practical applications, based on the set of geometric boundary feature points, a minimum envelope geometry for newly discovered defects is constructed. Specifically, this involves using the identified set of geometric boundary feature points and employing geometric algorithms (such as the convex hull algorithm, the minimum axis-aligned bounding box (AABB) algorithm, the minimum directed bounding box (OBB) algorithm, or the minimum bounding sphere algorithm) to generate a minimum geometric shape that completely contains all boundary feature points. This minimum envelope geometry can represent the overall spatial extent and approximate shape of the newly discovered defect in a concise and accurate manner, facilitating subsequent spatial overlap assessment.

[0122] Furthermore, determining whether there is spatial overlap between the minimum envelope geometry and the three-dimensional ellipsoidal spatial anchor point, and obtaining the spatial overlap judgment result, refers to using a standard geometric collision detection algorithm or spatial relationship judgment algorithm to determine whether there is any intersection between the minimum envelope geometry of the newly discovered defect and the three-dimensional ellipsoidal spatial anchor point corresponding to the first laser additive repair area. If an intersection exists, it indicates that the newly discovered defect and the historical repair area have spatial overlap.

[0123] In some preferred embodiments, a specific example is given below. Assume that after the initial laser additive repair of a rail, the repaired area is identified as a specific three-dimensional ellipsoidal spatial anchor point. Subsequently, a new microcrack defect is discovered on the rail surface during routine inspections. To accurately determine whether this microcrack overlaps with the previously repaired area, a high-precision three-dimensional laser scanner is first used to scan the microcrack and its surrounding area, thereby acquiring three-dimensional point cloud data of the microcrack. Next, local geometric feature analysis is performed on these point cloud data, such as calculating the normal direction and curvature of each point, identifying edge points with drastic curvature changes or abrupt changes in normal direction; these points constitute the set of geometric boundary feature points of the microcrack. Then, based on these boundary feature points, a minimum bounding box is constructed as the minimum envelope geometry of the microcrack, which tightly encloses the actual spatial range of the microcrack. Finally, the system executes a geometric collision detection algorithm to determine whether this minimum bounding box has spatial intersection with the previously determined three-dimensional ellipsoidal spatial anchor point. If spatial overlap is detected, the repair information of the microcrack will be added to the health history of the corresponding area, thereby achieving refined and accurate management of rail defects.

[0124] Optionally, the step of performing local geometric feature analysis on the 3D point cloud data to obtain the set of geometric boundary feature points of the newly discovered defects may include the following specific operations:

[0125] The 3D point cloud data is divided into local regions to obtain several local point cloud regions;

[0126] Calculate the local geometric feature parameters of each point within each local point cloud region;

[0127] Based on local geometric feature parameters, abnormal points in the local point cloud region that differ from the preset point cloud features of the healthy rail surface are identified.

[0128] Spatial connectivity analysis is performed on outliers, and spatially connected outliers are clustered into defect candidate regions.

[0129] Extract points on the spatial boundary of each defect candidate region to obtain a set of geometric boundary feature points of the newly discovered defect.

[0130] Local region partitioning can be achieved using various spatial indexing structures or partitioning methods. For example, 3D point cloud data can be divided into regular voxel grids, octree structures, or KD tree structures to facilitate efficient subsequent processing of local regions. Each local point cloud region represents a small set of discrete points on the rail surface. Local geometric feature parameters may include, but are not limited to, the normal direction, curvature, local density, and roughness of the points. These parameters can be calculated by analyzing the geometric distribution of each point within its neighborhood. For example, the normal direction can be determined using principal component analysis (PCA) or least squares fitting of a local plane; curvature can be calculated by fitting a local quadratic surface or analyzing the rate of change of the normal direction. The preset rail healthy surface point cloud features typically refer to the geometric characteristics exhibited by the point cloud of defect-free areas, such as a smooth surface, uniform normal direction, and stable local density. Outlier identification can be achieved by setting thresholds, statistical analysis methods (such as outlier detection based on standard deviation), or machine learning models. For example, when the curvature or normal direction of a point is significantly different from that of points in the surrounding healthy area, that point may be identified as an outlier. Spatial connectivity analysis aims to group close, geometrically similar outliers together. This can be achieved using clustering algorithms, such as density-based clustering (e.g., DBSCAN) or connected component-based algorithms. In this way, discrete outliers are organized into potential defect regions with a defined spatial extent, thus distinguishing isolated noise points from actual defects. Points on the spatial boundaries refer to those that form the outer contour of the defect candidate region. These points are crucial for accurately defining the shape and size of the defect. Identifying these boundary points provides accurate input data for subsequently constructing the minimum envelope geometry of the defect.

[0131] Optionally, the steps to supplement newly discovered defect remediation information as additional content to the regional health record include:

[0132] Collect repair process data for newly discovered defects; repair process data includes three-dimensional morphology data, laser power, scanning speed, powder feeding rate, preheating temperature, and quality inspection results after repair.

[0133] Based on the repair process data, generate a repair version data package corresponding to the newly discovered defect, and assign a unique version number and timestamp to the repair version data package;

[0134] The patched version data package will be added as additional content to the patch history list of the corresponding region's health record;

[0135] Based on the unique version number and timestamp of the patched version data package, update the latest version identifier and last update time of the regional health record.

[0136] Specifically, collecting data on the repair process of newly discovered defects refers to collecting various parameters and results directly related to the repair operation in real time or after the laser additive repair process. Three-dimensional morphology data can be understood as the geometric shape information of the defect area before, during, or after repair, such as point cloud data or mesh models acquired through a 3D scanner. Its purpose is to record the original morphology of the defect, the material accumulation during the repair process, and the surface smoothness after repair. Laser power, scanning speed, powder feed rate, and preheating temperature are key process parameters in the laser additive repair process. Accurate recording of these parameters helps analyze the relationship between repair quality and process parameters, providing a basis for subsequent repair optimization. Quality inspection results after repair, such as ultrasonic testing, eddy current testing, or visual inspection results, are used to assess the internal quality and surface integrity of the repaired area, ensuring that the repair meets the expected standards.

[0137] This process involves generating repair version data packages corresponding to newly discovered defects based on the repair process data, and assigning each repair version data package a unique version number and timestamp. This can be understood as structurally encapsulating all the collected repair process data into an independent, traceable data unit. The unique version number provides a unique identifier for each repair operation, ensuring no confusion between different repair records. The timestamp records the specific time the repair data package was generated or the repair was completed, facilitating chronological tracing of the repair history. The aim is to achieve fine-grained management and version control of each repair operation, ensuring data integrity and traceability.

[0138] In practical applications, appending repair version data packages as supplementary content to the repair history list of the corresponding regional health record means adding the newly generated repair version data package, according to a preset data structure, to a historical record set maintained within the regional health record corresponding to the rail repair area. This repair history list can be a time-sorted list, a database table, or a linked structure, used to store all version data packages from all repairs in that area. Its purpose is to construct a complete, time-ordered repair history archive, facilitating user or system querying and analysis.

[0139] Furthermore, updating the latest version identifier and last update time of the regional health record based on the unique version number and timestamp of the patch package refers to automatically updating the metadata of the regional health record after a new patch package is successfully appended. The latest version identifier typically points to the unique version number of the latest patch package in the current regional health record, while the last update time records the time when the regional health record was last modified. The purpose is to provide an overview of the latest status of the regional health record, allowing users to quickly understand its current version and update status.

[0140] In some preferred embodiments, a specific example is given below. Suppose that after a section of rail undergoes its first laser additive repair, a regional health history is established for the repaired area, storing the initial information of the first repair. Some time later, a new minor defect is discovered near the repaired area. When laser additive repair is performed on this newly discovered defect, the system collects various data in real time during the repair process, such as recording the laser output power as 3000W, the scanning speed as 10mm / s, the powder feeding rate as 15g / min, and the preheating temperature as 200℃. It also acquires the three-dimensional morphological data before and after repair, as well as the ultrasonic testing results after repair (e.g., no internal defects). This collected data is then packaged into a repair version data package, and the system automatically assigns a unique version number, such as "V20231026-001," and a precise timestamp, such as "2023-10-26 14:30:00." This repair version data package is then appended to the repair history list in the regional health history of the rail repaired area. Meanwhile, the latest version identifier of the regional health record has been updated to "V20231026-001", and the last update time has been updated to "2023-10-26 14:30:00". In this way, even if the region subsequently experiences defects and is repaired, detailed information about each repair will be appended in the form of a new version data package, forming a clear and complete repair history chain, which is convenient for querying and analysis at any time.

[0141] Optionally, the step of extracting points on the spatial boundary of each defect candidate region includes:

[0142] Local surface reconstruction is performed on the point cloud data within the defect candidate region to obtain a local surface mesh;

[0143] Local surface reconstruction refers to the process of converting discrete point cloud data into a continuous geometric surface representation, such as a triangular mesh or parametric surface, using algorithms. Its purpose is to provide a structured, continuous geometric model for subsequent edge point identification, enabling more accurate analysis of surface features.

[0144] On a local surface mesh, edge points where the curvature change exceeds a preset threshold or where there is an abrupt change in the normal direction are identified and recorded as abnormal edge points.

[0145] Specifically, identifying surface boundaries or feature edges by varying curvature beyond a preset threshold or by abrupt changes in the normal direction is a common method. On a local surface mesh, this can be determined by calculating the discrete curvature (e.g., Gaussian curvature, mean curvature) of each vertex or the change in its neighborhood normal. When these values ​​exceed a preset threshold, it indicates that the point is located in a region where the geometry has significantly changed, typically the boundary of a defect. These identified points are recorded as anomalous edge points.

[0146] Perform connectivity analysis on anomalous edge points, remove isolated anomalous edge points, and retain the set of connected anomalous edge points;

[0147] In practical applications, connectivity analysis refers to examining the spatial adjacency of anomalous edge points. Isolated anomalous edge points are usually caused by noise or measurement errors and do not represent the true defect boundaries. By removing these isolated points, interference can be effectively eliminated, ensuring that the remaining set of anomalous edge points is continuous and represents the true defect boundaries.

[0148] Based on the preserved set of abnormal edge points, points on the spatial boundary of the defect candidate region are extracted.

[0149] Therefore, the set of preserved connected anomaly edge points constitutes the precise spatial boundary of the defect candidate region. These points are extracted as part of the set of geometric boundary feature points for newly discovered defects.

[0150] Optionally, the step of performing local surface reconstruction on the point cloud data within the defect candidate region to obtain a local surface mesh includes:

[0151] Spatial neighborhood search is performed on the point cloud data within the defect candidate region to determine the neighborhood point set of each point;

[0152] Calculate the local weighted average normal direction and local density of each point based on the neighborhood point set of each point;

[0153] Based on the local weighted average normal direction and local density, points in the unevenly distributed areas within the defect candidate region are interpolated to generate a supplementary point cloud.

[0154] The original point cloud data and supplementary point cloud data within the defect candidate region are fused to form enhanced point cloud data;

[0155] Based on the enhanced point cloud data, local surface reconstruction is performed to obtain a local surface mesh.

[0156] Specifically, spatial neighborhood search is performed on the point cloud data within the defect candidate region to provide foundational data for subsequent local geometric feature calculations. By determining the set of neighboring points for each point, spatial information around that point can be obtained, laying the foundation for accurate calculation of local features. Spatial neighborhood search can employ methods such as K-nearest neighbor (K-NN) search or radius search.

[0157] Furthermore, based on the neighborhood set of each point, the local weighted average normal direction and local density are calculated for each point. The local weighted average normal direction reflects the orientation of the local surface of the point cloud, while the local density characterizes the density of the point cloud in that region. These parameters are key indicators for evaluating the quality and uniformity of point cloud data, aiming to identify unevenly distributed regions in the point cloud data. For example, regions with low local density may indicate sparse data or the presence of holes.

[0158] Building upon this, points in unevenly distributed areas within the defect candidate region are interpolated based on the locally weighted average normal direction and local density to generate supplementary point clouds. This step aims to fill sparse regions or holes in the point cloud data, generating new points through interpolation to make the point cloud distribution more uniform and complete. The interpolation method can be adaptively adjusted based on local geometric features (such as normal direction and curvature) to ensure the rationality of the supplementary points.

[0159] Subsequently, the original point cloud data and supplementary point cloud data within the defect candidate region are fused to form enhanced point cloud data. This fusion process combines the originally acquired point cloud data with the supplementary point cloud generated through interpolation, resulting in a point cloud dataset with higher density, more uniform distribution, and more complete coverage.

[0160] Finally, based on the enhanced point cloud data, local surface reconstruction is performed to obtain a local surface mesh. By using enhanced, more complete, and uniform point cloud data for surface reconstruction, the accuracy and integrity of the reconstructed local surface mesh can be significantly improved, while reducing voids and discontinuities.

[0161] In some preferred embodiments, a specific example is given below. Suppose that when laser scanning is performed on a defective area of ​​a rail to obtain point cloud data, due to the complex shape of the defect or the limited scanning angle, the point cloud data within the obtained defect candidate area is sparsely distributed in some parts, or even contains small voids.

[0162] To perform accurate local surface reconstruction, a spatial neighborhood search is first performed on the point cloud data within the candidate defect region; for example, finding the 10 nearest neighbors for each point. Based on these neighbors, the local weighted average normal direction and local density are calculated for each point. For example, in sparse point cloud regions, the local density of points will be significantly lower than in dense regions.

[0163] Next, the system identifies these unevenly distributed regions with low local density or drastic changes in normal direction (indicating surface discontinuity). For these regions, a series of supplementary point clouds are generated based on local geometric features of surrounding points (such as normal direction and curvature) using methods such as radial basis function (RBF) interpolation or moving least squares (MLS) interpolation to fill in these voids or sparse areas.

[0164] Subsequently, the original point cloud data and these newly generated supplementary point clouds are fused to form an enhanced point cloud data with higher density, more uniform distribution, and more complete coverage. For example, the fused point cloud data becomes denser in previously sparse areas, and the surface transitions are smoother.

[0165] Finally, based on this enhanced point cloud data, algorithms such as Poisson Reconstruction or Triangulation are used to reconstruct the local surface, resulting in a high-quality, void-free local surface mesh that accurately reflects the three-dimensional morphology of the defect. This accurate surface mesh provides a reliable foundation for subsequent identification of defect edge points, ensuring the accuracy of defect boundary extraction.

[0166] Optionally, the step of performing a spatial neighborhood search on the point cloud data within the defect candidate region to determine the neighborhood point set of each point includes:

[0167] Local density estimation is performed on the point cloud data within the defect candidate region to obtain the local point density of each point;

[0168] The parameters of the spatial neighborhood search are dynamically adjusted based on the local point density.

[0169] Based on the adjusted parameters of the spatial neighborhood search, a spatial neighborhood search is performed on each point to determine the set of neighboring points for each point.

[0170] Specifically, local density estimation of point cloud data within a defect candidate region refers to quantifying the point cloud density of the area where a point is located by calculating the number of points or the distance between points within its local region. For example, the K-Nearest Neighbors (K-NN) algorithm can be used to characterize the local density by calculating the distance from each point to its Kth nearest neighbor; or the Radius Search algorithm can be used to estimate the local density by counting the number of points within a given radius. The purpose is to obtain local feature information of the point cloud distribution. Dynamically adjusting the parameters of the spatial neighborhood search based on the local point density means adaptively changing the range of the spatial neighborhood search or the number of neighboring points based on the estimated local point density value. For example, for areas with low local point density, the search radius or K value can be appropriately increased to ensure that a sufficient number of neighboring points are captured; while for areas with high local point density, the search radius or K value can be decreased to avoid including too many redundant points and improve computational efficiency. Based on the adjusted parameters of the spatial neighborhood search, a spatial neighborhood search is performed on each point to determine the neighborhood point set of each point. This means that by using dynamically adjusted parameters, a neighborhood search operation is performed on each point within the defect candidate region, thereby obtaining a neighborhood point set for each point that can more accurately reflect its local geometric features.

[0171] In some preferred embodiments, a specific example is given below. Suppose that within a candidate region for a rail defect, there exists a sparsely distributed concave region and a densely distributed convex region. When estimating the local density of the point cloud data, the system identifies that the concave region has a lower local point density, while the convex region has a higher local point density. Based on this, when performing a spatial neighborhood search on points in the concave region, the search radius is dynamically adjusted to a larger value to ensure that enough points are included to accurately describe the geometry of the concave region. For example, if the default search radius is 5 mm, it might be adjusted to 8 mm in the sparse region. Conversely, when performing a spatial neighborhood search on points in the convex region, the search radius is dynamically adjusted to a smaller value, such as from 5 mm to 3 mm, to avoid including too many irrelevant points in the neighborhood, thereby more accurately capturing the sharp features of the convex region. In this way, regardless of changes in the point cloud distribution, the neighborhood set of each point can be optimized, providing high-quality input for subsequent local surface reconstruction, ultimately making the extraction of the geometric boundary feature point set of the rail defect more accurate.

[0172] Optionally, the step of estimating the local density of the point cloud data within the defect candidate region to obtain the local point density of each point includes:

[0173] The point cloud data within the defect candidate area is scanned to identify blind spots or occluded areas.

[0174] Local anomaly detection is performed on the non-blind area point cloud data surrounding areas with scanning blind spots or occlusions to obtain anomalies;

[0175] Remove outliers to obtain the surrounding non-blind zone point cloud data after removing the outliers;

[0176] Local smoothing is performed on the surrounding non-blind zone point cloud data after removing outliers to obtain smoothed surrounding non-blind zone point cloud data;

[0177] Based on the smoothed surrounding non-blind area point cloud data, supplementary point clouds are generated in the corresponding scanning blind area or occluded area according to the interpolation rules of local geometric features.

[0178] The original point cloud data and supplementary point cloud data within the defect candidate region are merged to form complete point cloud data;

[0179] Based on complete point cloud data, local density estimation is performed on each point to obtain the local point density of each point.

[0180] Specifically, before performing local density estimation on point cloud data within the defect candidate region, it is necessary to first assess and optimize the completeness and quality of the point cloud data. Among these, blind spot identification refers to identifying regions in the point cloud data that are missing or sparse due to limitations of the scanning equipment, object occlusion, or other reasons by analyzing information such as the distribution, density, and scanning path of the point cloud data. These regions may not provide complete geometric information, thus affecting subsequent density estimation.

[0181] Furthermore, for areas with scanning blind spots or occlusions, the surrounding non-blind spot point cloud data may contain some outliers. Local outlier detection aims to identify and remove these points that deviate from the normal data distribution, such as outliers caused by sensor noise, measurement errors, or environmental interference. Commonly used detection methods include statistical methods (such as Z-score, LOF local outlier factor), distance-based methods (such as K-nearest neighbor distance), or clustering-based methods. Removing these outliers helps improve the purity of the data.

[0182] After removing outliers, to further improve data quality, local smoothing is performed on the remaining surrounding non-blind zone point cloud data. Smoothing reduces random noise in the data, making the point cloud surface more continuous and uniform, thus providing a more reliable basis for subsequent interpolation and density estimation. Smoothing methods can include moving average, Gaussian filtering, or bilateral filtering.

[0183] Subsequently, based on the smoothed surrounding non-blind area point cloud data, supplementary point clouds are generated within the previously identified scanning blind areas or occluded regions according to interpolation rules based on local geometric features. The interpolation rules can predict the location and attributes of points in the missing region based on the geometric information of neighboring points (such as normal direction, curvature, local plane fitting, etc.) to fill data gaps. For example, radial basis function interpolation, kriging interpolation, or Poisson reconstruction-based interpolation methods can be used.

[0184] Finally, the original point cloud data within the defect candidate region is fused with the generated supplementary point cloud to form a more complete, continuous, and high-quality complete point cloud dataset. Based on this complete point cloud dataset, local density estimation is then performed for each point. Local density estimation can employ various methods, such as calculating the number of points within a specific radius neighborhood for each point, or calculating the reciprocal of the average distance from each point to its K nearest neighbors. By using complete point cloud data for estimation, more accurate and representative local point density information can be obtained.

[0185] In some preferred embodiments, a specific example is given below. Suppose that during point cloud scanning of candidate areas for rail defects, blind spots are generated in certain recessed or narrow areas due to the complexity of the rail geometry or the limitations of the scanner's viewing angle. Simultaneously, due to environmental vibrations or momentary sensor malfunctions, some outliers are mixed into the non-blind spot point cloud data.

[0186] First, the system analyzes the raw point cloud data within the defect candidate region, identifying these scanning blind spots through voxel-based density analysis or ray casting. For example, if the number of points within a voxel is significantly lower than the average of its surrounding voxels, that region may be marked as a blind spot.

[0187] Next, for the non-blind zone point cloud data surrounding the blind zone, a statistical outlier detection method (such as the SOR filter based on statistical principles) is used to detect local outliers. For example, for each point, the average distance and standard deviation of its K nearest neighbors are calculated. If the distance from the point to its neighbors exceeds a preset statistical threshold, it is identified as an outlier and removed.

[0188] Subsequently, local smoothing processing is performed on the surrounding non-blind zone point cloud data after removing outliers. For example, moving least squares method is used to fit the point cloud to eliminate residual noise and make the surface smoother.

[0189] Based on this, the system generates supplementary point clouds within the previously identified scanning blind zones using smoothed surrounding non-blind zone point cloud data. Specifically, an interpolation method based on radial basis functions (RBF) can be used to predict and generate points within the blind zone using the coordinates and normal information of known surrounding points. For example, an implicit surface can be constructed, and new points can be generated by sampling on that surface.

[0190] Finally, the original point cloud data is fused with these newly generated supplementary point clouds to form a more spatially continuous and complete point cloud dataset. Based on this complete point cloud data, local density estimation is performed for each point. For example, a spherical neighborhood search can be used to count the number of points within a sphere of a given radius for each point, and this number can be used as its local point density. In this way, even if the original data has defects, accurate local point density can be obtained, providing a reliable basis for subsequent dynamic adjustment of spatial neighborhood search parameters.

[0191] This application proposes a rail defect laser additive manufacturing management system for performing rail defect laser additive manufacturing management, combined with... Figure 3 As shown, the rail defect laser additive management system 1 includes:

[0192] The spatial anchor point determination module 11 is used to establish a regional health history for the repaired area of ​​the rail after the first laser additive repair of the rail is completed, and to determine the three-dimensional spatial anchor points of the regional health history.

[0193] The regional history storage module 12 is used to store the repair information of the first laser additive repair as the initial content of the regional health history.

[0194] The location information acquisition module 13 is used to identify newly discovered defects in the rail and acquire the three-dimensional location information of the newly discovered defects;

[0195] The spatial overlap judgment module 14 is used to determine whether there is spatial overlap between the three-dimensional location information and the three-dimensional spatial anchor points in the regional health record, and to obtain the spatial overlap judgment result.

[0196] The supplementary content module 15 is used to supplement the newly discovered defect repair information as supplementary content to the regional health record if the spatial overlap judgment result indicates that spatial overlap exists.

[0197] The spatial anchor point determination module can be implemented as a standalone software program unit, for example, as a service module in a backend data management platform, or as a processing unit embedded in the repair equipment control system. Its function is to receive the signal and related data upon completion of the first repair, and automatically generate or guide the user to establish a regional health history based on preset rules or algorithms, while simultaneously calculating and determining the corresponding three-dimensional spatial anchor points. For example, this module can determine the three-dimensional spatial anchor points based on the geometric dimensions and shape of the repair area or point cloud data of the repair area obtained through 3D scanning, using geometric modeling algorithms (such as the minimum envelope box algorithm or ellipsoid fitting algorithm). In some implementations, this module can also determine the three-dimensional spatial anchor points through manual input or selection of a predefined regional template. The specific operational procedures for establishing a regional health history and determining three-dimensional spatial anchor points have been described in detail in the above implementations and will not be repeated here.

[0198] The regional history storage module can be a database management system or a file storage interface, configured to receive regional health history information generated by the spatial anchor point determination module and repair information from the first laser additive repair, and store this information as initial content in a persistent storage medium in a structured or unstructured format. For example, this information can be stored in a relational database, NoSQL database, or distributed file system and associated with the corresponding regional health history. The specific operational procedures for storing the first laser additive repair information have been described in detail in the above embodiments and will not be repeated here.

[0199] The location information acquisition module can be a data acquisition and processing unit configured to identify newly discovered defects on the rail using various detection methods (e.g., high-precision 3D scanning equipment, machine vision systems, ultrasonic testing equipment, etc.), and extract or calculate the precise 3D location information of the newly discovered defects. For example, this module can receive point cloud data from a 3D scanner and use point cloud processing algorithms (such as segmentation and feature extraction) to identify defects and obtain their 3D coordinates, size, and shape information. In some embodiments, this module can also obtain the approximate location information of the defects through manual input or image recognition technology. The specific operational procedures for identifying newly discovered defects on the rail and obtaining their 3D location information have been described in detail in the above embodiments and will not be repeated here.

[0200] The spatial overlap determination module can be a computational analysis unit configured to receive the 3D location information of newly discovered defects provided by the location information acquisition module, as well as the 3D spatial anchor point information stored in the regional health record, and execute a spatial geometric determination algorithm to determine whether there is a spatial intersection or overlap between the two. For example, this module can use algorithms such as bounding box collision detection, geometric intersection testing, or point cloud overlap analysis to complete the determination. The determination result can be a Boolean value (overlap exists / does not exist) or an overlap metric. The specific operation process for determining whether there is spatial overlap between the 3D location information and the 3D spatial anchor points has been described in detail in the above embodiments and will not be repeated here.

[0201] The supplementary content module can be a data update and management unit. It is configured to receive newly discovered defect repair information as supplementary content when the spatial overlap judgment module determines that spatial overlap exists, and update or append it to the corresponding regional health history. For example, this module can generate a version data package from the new repair information and add it to the repair history list of the regional health history, while simultaneously updating the latest version identifier and last update time of the history. This module ensures that multiple repair histories of the same region can be completely recorded and traced. The specific operational procedures for supplementing the regional health history with newly discovered defect repair information have been described in detail in the above embodiments and will not be repeated here.

[0202] The above are merely embodiments of this application and are not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for laser additive manufacturing management of rail defects, characterized in that, include: After the first laser additive repair of the rail is completed, a regional health history is established for the repaired area of ​​the rail, and the three-dimensional spatial anchor points of the regional health history are determined. The repair information from the first laser additive repair is stored as the initial content of the region's health record; Identify newly discovered defects in the rails and obtain the three-dimensional location information of the newly discovered defects; Determine whether the three-dimensional location information overlaps with the three-dimensional spatial anchor points in the regional health record, and obtain the spatial overlap determination result; If the spatial overlap determination result indicates that spatial overlap exists, the repair information of the newly discovered defect will be added as supplementary content to the regional health record. The step of determining whether the three-dimensional location information overlaps with the three-dimensional spatial anchor points in the regional health record, and obtaining the spatial overlap determination result, includes: Obtain the laser output power, laser scanning speed, powder feeding rate, and preheating temperature for the first laser additive repair. Based on the laser output power, laser scanning speed, powder feeding rate, and preheating temperature of the first laser additive repair, the three-dimensional ellipsoidal spatial anchor point is calculated and determined as the three-dimensional spatial anchor point of the health history of the region. Determine whether the three-dimensional position information overlaps with the spatial anchor point of the three-dimensional ellipsoid to obtain the spatial overlap determination result; The three-dimensional ellipsoidal spatial anchor point, as a geometric model, has a clear mapping relationship between its major axis, minor axis, and center position and the physical control parameters of laser additive repair.

2. The method for laser additive manufacturing management of rail defects according to claim 1, characterized in that, The step of determining whether the three-dimensional position information overlaps with the spatial anchor point of the three-dimensional ellipsoid and obtaining the spatial overlap determination result includes: Based on the three-dimensional location information, obtain the three-dimensional point cloud data of the newly discovered defect; Local geometric feature analysis is performed on the three-dimensional point cloud data to obtain the set of geometric boundary feature points of the newly discovered defect; Based on the set of geometric boundary feature points, construct the minimum envelope geometry of the newly discovered defect; Determine whether there is spatial overlap between the minimum envelope geometry and the spatial anchor point of the three-dimensional ellipsoid, and obtain the spatial overlap determination result.

3. The method for laser additive manufacturing management of rail defects according to claim 2, characterized in that, The step of performing local geometric feature analysis on the three-dimensional point cloud data to obtain the set of geometric boundary feature points of the newly discovered defect includes: The three-dimensional point cloud data is divided into local regions to obtain several local point cloud regions; Calculate the local geometric feature parameters of each point within each of the local point cloud regions; Based on the local geometric feature parameters, abnormal points in the local point cloud region that differ from the preset point cloud features of a healthy rail surface are identified. Spatial connectivity analysis is performed on the anomalies, and spatially connected anomalies are clustered into defect candidate regions. Extract points on the spatial boundary of each of the defect candidate regions to obtain the set of geometric boundary feature points of the newly discovered defect.

4. The method for managing rail defects using laser additive manufacturing according to claim 1, characterized in that, The step of adding the repair information of the newly discovered defects as supplementary content to the regional health record includes: Collect repair process data for the newly discovered defects; the repair process data includes three-dimensional morphology data, laser power, scanning speed, powder feeding rate, preheating temperature, and quality inspection results after repair. Based on the repair process data, a repair version data package corresponding to the newly discovered defect is generated, and a unique version number and timestamp are assigned to the repair version data package. The repaired version data package will be added as additional content to the repair history list of the corresponding region's health record; Based on the unique version number and timestamp of the repaired version data package, update the latest version identifier and last update time of the regional health record.

5. The method for laser additive manufacturing management of rail defects according to claim 3, characterized in that, The step of extracting points on the spatial boundary of each of the defect candidate regions includes: Local surface reconstruction is performed on the point cloud data within the defect candidate region to obtain a local surface mesh; On the local surface mesh, edge points where the curvature change exceeds a preset threshold or where there is an abrupt change in the normal direction are identified and recorded as abnormal edge points; Perform connectivity analysis on the abnormal edge points, remove isolated abnormal edge points, and retain the set of connected abnormal edge points; Based on the retained set of abnormal edge points, points on the spatial boundary of the defect candidate region are extracted.

6. The method for laser additive manufacturing management of rail defects according to claim 5, characterized in that, The step of performing local surface reconstruction on the point cloud data within the defect candidate region to obtain a local surface mesh includes: A spatial neighborhood search is performed on the point cloud data within the defect candidate region to determine the neighborhood point set of each point; Calculate the local weighted average normal direction and local density of each point based on the neighborhood point set of each point; Based on the local weighted average normal direction and the local density, points in the unevenly distributed areas within the defect candidate region are interpolated to generate a supplementary point cloud. The original point cloud data within the defect candidate region and the supplementary point cloud data are fused to form enhanced point cloud data; Based on the enhanced point cloud data, local surface reconstruction is performed to obtain a local surface mesh.

7. The method for laser additive manufacturing management of rail defects according to claim 6, characterized in that, The step of performing a spatial neighborhood search on the point cloud data within the defect candidate region to determine the neighborhood point set of each point includes: Local density estimation is performed on the point cloud data within the defect candidate region to obtain the local point density of each point; The parameters of the spatial neighborhood search are dynamically adjusted based on the local point density. Based on the adjusted parameters of the spatial neighborhood search, a spatial neighborhood search is performed on each point to determine the set of neighboring points for each point.

8. The method for laser additive manufacturing management of rail defects according to claim 7, characterized in that, The step of estimating the local density of the point cloud data within the defect candidate region to obtain the local point density of each point includes: The point cloud data within the defect candidate region is scanned to identify blind spots and determine areas with scan blind spots or occlusions. Local anomaly detection is performed on the non-blind area point cloud data surrounding areas with scanning blind spots or occlusions to obtain anomalies; Remove the abnormal points to obtain the surrounding non-blind area point cloud data after removing the abnormal points; Local smoothing is performed on the surrounding non-blind zone point cloud data after removing outliers to obtain smoothed surrounding non-blind zone point cloud data; Based on the smoothed surrounding non-blind area point cloud data, supplementary point clouds are generated in the corresponding scanning blind area or occluded area according to the interpolation rules of local geometric features. The original point cloud data within the defect candidate region and the supplementary point cloud data are fused to form complete point cloud data; Based on the complete point cloud data, local density estimation is performed on each point to obtain the local point density of each point.

9. A rail defect laser additive manufacturing management system, used for performing rail defect laser additive manufacturing management, characterized in that, include: The spatial anchor point determination module is used to establish a regional health history for the repaired area of ​​the rail after the first laser additive repair is completed, and to determine the three-dimensional spatial anchor points of the regional health history. The regional history storage module is used to store the repair information of the first laser additive repair as the initial content of the regional health history; The location information acquisition module is used to identify newly discovered defects in the rail and acquire the three-dimensional location information of the newly discovered defects; The spatial overlap judgment module is used to determine whether there is spatial overlap between the three-dimensional location information and the three-dimensional spatial anchor points in the regional health history, and to obtain the spatial overlap judgment result. The supplementary content module is used to supplement the newly discovered defect repair information as supplementary content of the regional health history if the spatial overlap judgment result indicates that spatial overlap exists. The spatial overlap determination module is also used for: Obtain the laser output power, laser scanning speed, powder feeding rate, and preheating temperature for the first laser additive repair. Based on the laser output power, laser scanning speed, powder feeding rate, and preheating temperature of the first laser additive repair, the three-dimensional ellipsoidal spatial anchor point is calculated and determined as the three-dimensional spatial anchor point of the health history of the region. Determine whether the three-dimensional position information overlaps with the spatial anchor point of the three-dimensional ellipsoid to obtain the spatial overlap determination result; The three-dimensional ellipsoidal spatial anchor point, as a geometric model, has a clear mapping relationship between its major axis, minor axis, and center position and the physical control parameters of laser additive repair.

Images