Unmanned aerial vehicle point cloud defect identification method for contact net of rail transit rotating device

Abstract

The present application provides a unmanned aerial vehicle point cloud defect identification method for rotating equipment of a catenary of rail transit, and relates to the technical field of rail transit facility state detection. The present application collects original three-dimensional point cloud data, flight attitude, spatial position information and equipment design parameters of rotating equipment, constructs a rotating constraint coordinate system, realizes rotating state normalization, forms a spatial geometric residual field, fuses rotating axis offset, component curvature change and rotating attitude deviation, calculates rotating geometric equivalent deviation coefficients, and judges the rotating geometric state. For abnormal candidate samples, a point cloud structure credibility evaluation mechanism is introduced to suppress false deformation and position distortion caused by point cloud quality fluctuations. The rotating safety margin decay coefficient is calculated to realize the hierarchical evaluation of the running safety state of the rotating equipment, generate operation and maintenance strategies such as speed limit operation, priority repair or immediate maintenance, thereby improving the accuracy of the defect identification of the catenary rotating equipment and the running safety guarantee capability.

Description

Technical Field

[0001] This invention relates to the field of rail transit facility condition monitoring technology, specifically a method for identifying defects in the contact network of rotating rail transit equipment using unmanned aerial vehicles (UAVs) point clouds. Background Technology

[0002] As rail transit systems develop towards higher speeds, denser systems, and unmanned operation, the overhead contact system and its constituent rotating equipment, such as pulleys, compensation devices, and rotating connecting components, are susceptible to the combined effects of vibration loads, environmental corrosion, and fluctuations in operating conditions during long-term service. Changes in their geometric state and structural safety directly impact the stable operation of the power supply system and train safety. Therefore, conducting high-precision, non-contact condition monitoring and defect identification of rotating equipment has become a crucial technical requirement in the rail transit operation and maintenance field.

[0003] In recent years, 3D point cloud detection technology based on UAVs equipped with lidar or multi-sensor fusion has been gradually applied to the inspection of overhead contact line equipment. By acquiring high-density point cloud data of the equipment surface, structural morphology reconstruction and anomaly identification can be achieved without stopping operations, offering advantages such as wide coverage and high efficiency. However, in practical engineering applications, the acquired point cloud data often suffers from problems such as uneven density, local missing data, noise superposition, and spatial distribution distortion due to the influence of UAV flight attitude changes, scanning angle limitations, lighting and environmental interference, and the equipment's own rotation characteristics.

[0004] In existing technologies, most point cloud defect identification methods focus on threshold judgment of single geometric indicators or local anomalies, such as defect determination based on local curvature abrupt changes, point spacing anomalies, or single-frame geometric deviations. These methods typically assume that the point cloud data quality is reliable and do not systematically evaluate the credibility of the overall point cloud structure. When the point cloud acquisition quality fluctuates, geometric anomalies caused by acquisition noise or structural incompleteness are easily misjudged as real defects, leading to amplification of the anomaly amplitude and spatial drift of the anomaly location, affecting the accuracy of subsequent safety assessment results.

[0005] Furthermore, existing methods for safety status assessment of rotating equipment often rely on static criteria such as "whether geometric deviation exceeds limits," lacking joint modeling of point cloud data quality, structural consistency, and the reliability of geometric anomalies. This makes it difficult to reflect the true changes in the safety margin of rotating structures under complex operating conditions. Especially when point cloud quality is unstable, generating maintenance decisions solely based on geometric deviation indicators can easily lead to over-maintenance or missed risk assessments, failing to meet the actual needs of refined maintenance and tiered decision-making. Summary of the Invention

[0006] The purpose of this invention is to provide a method for identifying defects in the point cloud of overhead contact lines of rotating rail transit equipment using unmanned aerial vehicles (UAVs), in order to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A method for identifying defects in the overhead contact system of rotating equipment in rail transit using unmanned aerial vehicles (UAVs), comprising the following specific steps:

[0009] Step 1: Collect raw 3D point cloud data with spatial coordinate information; simultaneously collect UAV flight attitude parameters, spatial position information and timestamps, and combine them with the design parameter data of the inspected rotating equipment stored in the contact network equipment management system;

[0010] Step 2: Denoise, resample, and perform attitude compensation on the original point cloud to obtain a standardized point cloud with consistent attitude; extract the point cloud of rotating equipment and identify its axial distribution characteristics, fit the actual rotation axis and construct a rotation constraint coordinate system, and normalize the rotation state of the point cloud; extract the rotation parameter distribution and calculate the deviation between it and the predefined rotation structure reference, and construct a spatial geometric residual field.

[0011] Step 3: Calculate and obtain the rotational geometric equivalent deviation coefficient, and compare it with the rotational geometric equivalent deviation threshold to make an overall judgment on the rotational geometric state of the rotating component. When the deviation exceeds the limit, mark the corresponding rotational state normalized point cloud as an abnormal candidate sample and trigger the quality reliability assessment process.

[0012] Step 4: Calculate and obtain the point cloud structure credibility coefficient, and compare it with the point cloud structure credibility threshold to determine the reliability of the point cloud data quality corresponding to the current rotating device. When the credibility is insufficient, trigger an early warning and perform resampling, completion smoothing and weakening processing under the structural consistency constraint on the point cloud to suppress the risk of false deformation and position distortion.

[0013] Step 5: Calculate the rotational safety margin attenuation coefficient and compare it with the rotational safety margin attenuation threshold to determine whether the rotating equipment is within the safe operating margin range. When the safety margin is insufficient, trigger an early warning and generate operation and maintenance strategies such as speed-limited operation, priority maintenance, or immediate maintenance.

[0014] Further, step one includes:

[0015] S11. Using drones equipped with lidar and imaging devices, a preset route is used for inspection flights above the rail transit line along the contact wire. This allows the drones to complete multi-angle and multi-height scanning coverage within the space area where rotating contact wire equipment is located, and ensures that the overall structure of the rotating equipment is within the effective detection range.

[0016] S12. Using a laser radar scanning device installed on the UAV body, continuously scan the contact wire rotating equipment, collect the original three-dimensional point cloud data of the rotating equipment, and record the spatial coordinate value corresponding to each laser echo point.

[0017] S13. During the lidar scanning process, the flight attitude data of the UAV, including pitch angle, roll angle and heading angle, are collected synchronously by the inertial measurement equipment and attitude sensing equipment installed on the UAV to form scanning attitude parameters.

[0018] S14. During the point cloud acquisition process, the spatial location information and corresponding timestamp of the UAV are simultaneously obtained through the satellite positioning module and system clock module carried on the UAV.

[0019] S15. By accessing the overhead contact line equipment management system, obtain the design parameter data of the rotating equipment being inspected, including the theoretical rotation axis position, rotation center coordinates, design rotation angle range, and nominal dimension parameters of the components.

[0020] Furthermore, step two includes:

[0021] S21. Isolated noise points in the original 3D point cloud data are removed by using a neighborhood statistics-based outlier detection method; density equalization of the point cloud is performed through resampling to ensure consistent point cloud distribution at different scanning distances and angles; attitude compensation and coordinate transformation are performed on the point cloud based on scanning attitude parameters to unify point clouds collected with different flight attitudes, different spatial location information, and corresponding timestamps into the same reference coordinate system; a standardized point cloud with consistent attitude and balanced density is obtained.

[0022] S22. Based on standardized point clouds, a point cloud segmentation method based on spatial position constraints is adopted to extract the point cloud regions corresponding to rotating equipment in the contact network, forming a subset of equipment point clouds; and in the subset of equipment point clouds, axial projection and continuity analysis are performed on the structural features of rotating equipment components that are continuously distributed along the axial direction to obtain the distribution characteristics of the point cloud along the potential rotation direction.

[0023] S23. Based on the equipment point cloud subset and the distribution characteristics of the point cloud along the potential rotation direction, a point cloud geometric fitting method is used to perform axis fitting processing on the key components of rotating equipment: the main direction vector of the equipment point cloud is calculated by the least squares fitting analysis method; under the main direction constraint, the position and direction parameters of the actual rotation axis are determined to obtain the actual rotation axis parameters; at the same time, the corresponding rotation center position is calculated.

[0024] S24. Based on the theoretical rotation axis position, rotation center coordinates, design rotation angle range, and nominal dimension parameters of the component, and combined with the actual rotation axis parameters, construct a rotation constraint coordinate system with the rotation axis as the reference; under the current rotation constraint coordinate system, perform coordinate mapping and rotation normalization processing on the standardized point cloud, and uniformly map the point clouds of different scanning postures and different rotation states to the equivalent reference rotation state to obtain the rotation state normalized point cloud;

[0025] S25. Based on the normalized point cloud of the rotation state, extract the rotation angle distribution, radial offset distribution and axial deformation distribution under rotation constraints to establish the rotation parameter set of the equipment; under the premise of eliminating the influence of normal rotation attitude, calculate the deviation between the normalized point cloud and the predefined rotation structure reference to construct the spatial geometric residual field.

[0026] Furthermore, step three includes:

[0027] S31. Based on the actual rotation axis parameters and the theoretical rotation axis position, calculate the minimum spatial distance between them in the rotation constraint coordinate system to obtain the rotation axis offset; based on the nominal size parameters of the component, determine the reference normalized size of the rotation axis offset to obtain the nominal structural size parameters.

[0028] S32. Based on the normalized point cloud of the rotation state, the point cloud is segmented axially along the rotation axis, and local geometric fitting is performed on each axial segment to extract the local curvature value corresponding to each segment; the curvature change of the rotating component along the axis is obtained based on the difference in curvature values ​​of adjacent axial segments; and the curvature reference value of the rotating component under normal conditions is obtained through the statistical results of the point cloud of rotating equipment under historical normal working conditions, and the curvature reference parameter is determined.

[0029] S33. Based on the set of rotation parameters, calculate the actual rotation attitude parameters under the current rotation state; based on the design rotation angle range, determine the corresponding design rotation attitude reference; calculate the angle difference between the actual rotation attitude parameters and the design rotation attitude reference to obtain the rotation attitude angle deviation parameters; based on the design rotation angle range, determine the reference standard parameters for the rotation attitude angle deviation.

[0030] Furthermore, step three also includes:

[0031] S34. After dimensionless processing of the obtained rotation axis offset, nominal structural dimension parameters, axial curvature change of rotating components, curvature reference parameters, rotation attitude angle deviation parameters, and reference standard parameters of rotation attitude angle deviation, the rotation geometric equivalent deviation coefficient is calculated.

[0032] Furthermore, step three also includes:

[0033] S35. By setting a preset rotational geometric equivalent deviation threshold, and comparing and analyzing the rotational geometric equivalent deviation coefficient with the rotational geometric equivalent deviation threshold, the first evaluation result is obtained, including:

[0034] When the rotational geometric equivalent deviation coefficient is less than or equal to the rotational geometric equivalent deviation threshold, it indicates that the rotational geometry of the rotating component is qualified and should be continuously monitored.

[0035] When the rotational geometric equivalent deviation coefficient is greater than the rotational geometric equivalent deviation threshold, it indicates that the rotational geometry of the rotating component is unqualified and that the rotating component has spatial geometric residual field characteristics that cannot be explained by the rotational constraints. This triggers the first warning instruction and generates the first strategy: to mark the current rotational geometry as a potential deformation risk state, and to use the corresponding rotational state normalized point cloud as anomaly candidate sample, and to start the quality reliability assessment mechanism.

[0036] Furthermore, step four includes:

[0037] S41. Initiate the quality reliability assessment mechanism, and based on the rotation state normalized point cloud in the abnormal candidate samples, use the point cloud neighborhood consistency analysis method to perform axial continuity and structural symmetry constraint analysis on the point cloud of rotating equipment, count the number of point cloud points that meet the axial continuity and structural symmetry constraint conditions, and obtain the number of valid point cloud points; count the total number of point cloud points corresponding to the current rotating equipment, and obtain the total number of point cloud points.

[0038] S42. Based on the rotation state normalized point cloud in the abnormal candidate samples, the statistical analysis method of the distance between adjacent points is used to statistically analyze the changes in the distance between points in the local neighborhood of the point cloud, calculate the degree of fluctuation of the local density of the point cloud, and obtain the local density fluctuation coefficient of the point cloud.

[0039] S43. Based on the rotational state normalized point cloud in the abnormal candidate samples, a statistical constraint analysis method based on the consistency of point cloud spatial distribution is adopted to comprehensively evaluate the continuity and consistency of the spatial distribution of point cloud in the axial and radial directions under the rotational constraint coordinate system, and obtain the structural self-consistency score parameters.

[0040] Furthermore, step four also includes:

[0041] S44. After dimensionless processing of the obtained effective point cloud points, total point cloud points, point cloud local density fluctuation coefficient, and structure self-consistency scoring parameters, the point cloud structure reliability coefficient is calculated.

[0042] S45. By setting a preset point cloud structure credibility threshold, and comparing the point cloud structure credibility coefficient with the point cloud structure credibility threshold, the second evaluation result is obtained, including:

[0043] When the point cloud structure reliability coefficient is greater than or equal to the point cloud structure reliability threshold, it indicates that the point cloud data quality corresponding to the current rotating device is reliable and should be continuously monitored.

[0044] When the point cloud structure reliability coefficient is less than the point cloud structure reliability threshold, it indicates that the point cloud data quality corresponding to the current rotating device is unreliable. The spatial geometric residual field has the risk of false deformation, abnormal amplitude amplification, and abnormal spatial position distortion caused by fluctuations in point cloud acquisition quality. This triggers a second warning instruction and generates a second strategy: for the rotation state normalized point cloud in the abnormal candidate samples, perform a reprocessing operation oriented towards overall structural consistency; perform resampling processing under structural consistency constraints on the overall distribution of the point cloud; perform completion or smoothing constraint processing on the continuous interrupted regions in the spatial geometric residual field; and perform overall weakening processing on the point cloud regions that cannot meet the requirements of rotation constraints and structural consistency.

[0045] Furthermore, step five includes:

[0046] S51. The rotational geometric equivalent deviation coefficient and point cloud structure reliability coefficient obtained through calculation are processed without dimensions, and a weighted fusion calculation method is used to calculate the rotational safety margin attenuation coefficient.

[0047] 10. The method for identifying UAV point cloud defects in the overhead contact system of rotating rail transit equipment according to claim 9, characterized in that: step five further includes:

[0048] S52. By setting a preset rotational safety margin attenuation threshold and comparing the rotational safety margin attenuation coefficient with the rotational safety margin attenuation threshold, the third evaluation results are obtained, including:

[0049] When the rotational safety margin attenuation coefficient is less than or equal to the rotational safety margin attenuation threshold, it indicates that the rotating equipment is still within the safe operating margin range under the combined influence of geometric state deviation and point cloud quality reliability. The rotating equipment is judged to be in a safe operating state and is continuously monitored.

[0050] When the rotational safety margin attenuation coefficient exceeds the rotational safety margin attenuation threshold, it indicates that the geometric deviation and point cloud quality factors of the rotating equipment have had a cumulative weakening effect on the operational safety margin. It is determined that the current operational safety margin of the rotating equipment is insufficient and there is an operational risk. The third warning instruction is triggered, and the third strategy is generated: the current rotating equipment is marked as a key focus object; the speed limit operation instruction, priority maintenance instruction, or immediate maintenance instruction is triggered; and the corresponding rotational state normalized point cloud, spatial geometric residual field, and various evaluation coefficients are archived as the basis for operation and maintenance decision-making.

[0051] Compared with the prior art, the beneficial effects of the present invention are:

[0052] This invention eliminates the interference of UAV flight attitude changes and normal equipment rotation on the point cloud geometry by constructing a rotation-constrained coordinate system and performing rotation state normalization processing on the point cloud. On this basis, a spatial geometric residual field is constructed to characterize geometric anomalies that cannot be explained by normal rotation in three-dimensional space. This enables rotation-related defects such as axis offset, curvature anomaly, and attitude deviation to be quantitatively identified, avoiding the problem that traditional two-dimensional image or unconstrained point cloud methods cannot distinguish between "normal rotation" and "real deformation", and significantly improving the accuracy of defect identification.

[0053] This invention also addresses the common problems of occlusion, sparsity, uneven density, and noise interference in UAV inspections. After determining geometric anomalies, this invention further calculates the point cloud structure credibility coefficient. It comprehensively evaluates the point cloud quality by considering the proportion of effective points, local density stability, and spatial distribution consistency. When the credibility is insufficient, it automatically triggers resampling, completion smoothing, and weakening processing under structural consistency constraints. This mechanism suppresses the risks of false deformation, amplified abnormal amplitude, and spatial position distortion, avoids low-quality point clouds from directly participating in defect determination, and improves the overall system's engineering robustness and reliability.

[0054] This invention also constructs a rotational safety margin attenuation coefficient by integrating the rotational geometric equivalent deviation coefficient and the point cloud structure credibility coefficient, and generates differentiated operation and maintenance strategies such as speed-limited operation, priority inspection, or immediate maintenance based on threshold judgment. This realizes a complete closed loop from "point cloud acquisition - defect identification - risk assessment - operation and maintenance decision-making". It not only avoids the isolated output of defect results at the detection level, but also provides a quantitative, interpretable, and executable intelligent operation and maintenance basis for rotating equipment of rail transit catenary, effectively improving the safety of line operation and maintenance efficiency. Attached Figure Description

[0055] Figure 1 This is a schematic diagram of the overall method steps of the present invention. Detailed Implementation

[0056] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0057] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0058] Example 1

[0059] Please see Figure 1 The present invention provides a technical solution: the specific steps include:

[0060] Step 1: Collect raw 3D point cloud data with spatial coordinate information; simultaneously collect UAV flight attitude parameters, spatial position information and timestamps, and combine them with the design parameter data of the inspected rotating equipment stored in the contact network equipment management system;

[0061] Step 2: Denoise, resample, and perform attitude compensation on the original point cloud to obtain a standardized point cloud with consistent attitude; extract the point cloud of rotating equipment and identify its axial distribution characteristics, fit the actual rotation axis and construct a rotation constraint coordinate system, and normalize the rotation state of the point cloud; extract the rotation parameter distribution and calculate the deviation between it and the predefined rotation structure reference, and construct a spatial geometric residual field.

[0062] Step 3: Calculate and obtain the rotational geometric equivalent deviation coefficient, and compare it with the rotational geometric equivalent deviation threshold to make an overall judgment on the rotational geometric state of the rotating component. When the deviation exceeds the limit, mark the corresponding rotational state normalized point cloud as an abnormal candidate sample and trigger the quality reliability assessment process.

[0063] Step 4: Calculate and obtain the point cloud structure credibility coefficient, and compare it with the point cloud structure credibility threshold to determine the reliability of the point cloud data quality corresponding to the current rotating device. When the credibility is insufficient, trigger an early warning and perform resampling, completion smoothing and weakening processing under the structural consistency constraint on the point cloud to suppress the risk of false deformation and position distortion.

[0064] Step 5: Calculate the rotational safety margin attenuation coefficient and compare it with the rotational safety margin attenuation threshold to determine whether the rotating equipment is within the safe operating margin range. When the safety margin is insufficient, trigger an early warning and generate operation and maintenance strategies such as speed-limited operation, priority maintenance, or immediate maintenance.

[0065] In this embodiment, by introducing a comprehensive judgment mechanism of point cloud structure reliability assessment and rotational safety margin attenuation based on rotational geometric deviation judgment, the geometric anomaly identification, data quality reliability judgment and operation safety assessment of UAV point clouds are coupled in a layered manner. This effectively avoids misjudgment problems caused by point cloud quality fluctuations or normal rotational attitude changes, enabling the defect identification results of rotating equipment to directly support operation and maintenance decisions such as speed-limited operation, priority inspection or immediate maintenance, thereby significantly improving the reliability and engineering application value of the inspection results of rotating equipment in rail transit catenary.

[0066] Example 2

[0067] Please see Figure 1 In this embodiment, as explained in Embodiment 1, specifically, step one includes:

[0068] S11. Using drones equipped with lidar and imaging devices, a preset route is used for inspection flights above the rail transit line along the contact wire. This allows the drones to complete multi-angle and multi-height scanning coverage within the space area where rotating contact wire equipment is located, and ensures that the overall structure of the rotating equipment is within the effective detection range.

[0069] S12. Using a laser radar scanning device installed on the UAV body, continuously scan the contact wire rotating equipment, collect the original three-dimensional point cloud data forming the rotating equipment, denoted as Praw, and record the spatial coordinate value corresponding to each laser echo point.

[0070] S13. During the lidar scanning process, the flight attitude data of the UAV, including pitch angle, roll angle and heading angle, are collected synchronously by the inertial measurement equipment and attitude sensing equipment installed on the UAV to form scanning attitude parameters, denoted as θf.

[0071] S14. During the point cloud acquisition process, the spatial location information and corresponding timestamp of the UAV are simultaneously obtained through the satellite positioning module and system clock module carried on the UAV.

[0072] S15. By accessing the overhead contact line equipment management system, obtain the design parameter data of the rotating equipment being inspected, including the theoretical rotation axis position, rotation center coordinates, design rotation angle range, and nominal dimension parameters of the components.

[0073] In this embodiment, by simultaneously acquiring flight attitude parameters, spatial location information, timestamps, and design parameter data of rotating equipment during the UAV point cloud acquisition process, the original 3D point cloud has complete spatiotemporal reference and design constraints at the acquisition stage. This provides a reliable basis for subsequent attitude compensation, rotation axis fitting, and rotation constraint coordinate system construction, thereby effectively reducing the accumulation of spatial deviations caused by multi-angle and multi-height scanning and improving the overall integrity and usability of the point cloud data of the overhead contact line rotating equipment.

[0074] Example 3

[0075] Please see Figure 1 In the explanation of Example 2, this embodiment specifically includes the following steps:

[0076] S21. Isolated noise points in the original 3D point cloud data Praw are removed using a neighborhood statistics-based outlier detection method. Density equalization of the point cloud is achieved through resampling to ensure consistent point cloud distribution at different scanning distances and angles. Attitude compensation and coordinate transformation are performed on the point cloud based on the scanning attitude parameter θf, unifying point clouds collected with different flight attitudes, spatial locations, and corresponding timestamps to the same reference coordinate system. A standardized point cloud with consistent attitude and balanced density is obtained, denoted as P1.

[0077] S22. Based on the standardized point cloud P1, a point cloud segmentation method based on spatial position constraints is used to extract the point cloud region corresponding to rotating equipment in the contact network to form a subset of equipment point clouds. In the subset of equipment point clouds, axial projection and continuity analysis are performed on the structural features of rotating equipment components that are continuously distributed along the axial direction to obtain the distribution characteristics of the point cloud along the potential rotation direction.

[0078] S23. Based on the equipment point cloud subset and the distribution characteristics of the point cloud along the potential rotation direction, the point cloud geometric fitting method is used to perform axis fitting processing on the key components of rotating equipment: the main direction vector of the equipment point cloud is calculated by the least squares fitting analysis method; under the constraint of the main direction, the position and direction parameters of the actual rotation axis are determined, and the actual rotation axis parameters are obtained, denoted as Ar; at the same time, the corresponding rotation center position is calculated.

[0079] S24. Based on the theoretical rotation axis position, rotation center coordinates, design rotation angle range, and nominal dimension parameters of the component, and combined with the actual rotation axis parameter Ar, construct a rotation constraint coordinate system with the rotation axis as the reference. Under the current rotation constraint coordinate system, perform coordinate mapping and rotation normalization processing on the standardized point cloud P1, and uniformly map the point clouds under different scanning postures and different rotation states to the equivalent reference rotation state to obtain the rotation state normalized point cloud, denoted as P2.

[0080] S25. Based on the normalized point cloud P2 of the rotation state, extract the rotation angle distribution, radial offset distribution and axial deformation distribution under rotation constraints to establish the rotation parameter set of the equipment; under the premise of eliminating the influence of normal rotation attitude, calculate the deviation between the normalized point cloud and the predefined rotation structure reference to construct the spatial geometric residual field.

[0081] In this embodiment, by performing attitude compensation, density equalization and rotation state normalization on the original point cloud, and constructing a spatial geometric residual field under the rotation constraint coordinate system, the point cloud data collected under different scanning attitudes and different rotation states are comparable. This enables the accurate characterization of the true geometric deviation of rotating equipment on the basis of eliminating the influence of normal rotation attitude, thereby improving the stability and discrimination accuracy of defect identification of rotating contact network equipment.

[0082] Example 4

[0083] Please see Figure 1 In the explanation of Example 3, this embodiment specifically includes the following steps:

[0084] S31. Based on the actual rotation axis parameter Ar and the theoretical rotation axis position, calculate the minimum spatial distance between them in the rotation constraint coordinate system to obtain the rotation axis offset, denoted as daxis; based on the nominal size parameters of the component, determine the reference normalized size of the rotation axis offset to obtain the nominal structural size parameter, denoted as Dref;

[0085] S32. Based on the normalized point cloud P2 of the rotation state, the point cloud is segmented axially along the rotation axis, and local geometric fitting is performed on each axial segment to extract the local curvature value corresponding to each segment; based on the difference in curvature values ​​of adjacent axial segments, the curvature change of the rotating component along the axis is calculated and denoted as... Furthermore, by utilizing the point cloud statistics of rotating equipment under historical normal operating conditions, the curvature reference value of the rotating component under normal conditions is obtained, and the curvature reference parameter is determined, denoted as... ;

[0086] S33. Based on the set of rotation parameters, calculate the actual rotation attitude parameters under the current rotation state; based on the design rotation angle range, determine the corresponding design rotation attitude reference; calculate the angle difference between the actual rotation attitude parameters and the design rotation attitude reference, and obtain the rotation attitude angle deviation parameter, denoted as... Based on the designed rotation angle range, a reference standard parameter for the rotation attitude angle deviation is determined, denoted as . .

[0087] In this embodiment, the rotational state of rotating equipment is quantitatively characterized from three independent geometric dimensions: rotation axis offset, axial curvature change, and rotational attitude angle deviation. The corresponding structural dimensions and historical normal operating conditions are introduced as reference benchmarks, so that the geometric anomalies of rotating components can be uniformly mapped into comparable parameter indicators. This avoids misjudgment caused by relying on a single feature and improves the comprehensiveness and reliability of rotational geometric state assessment.

[0088] Example 5

[0089] Please see Figure 1 In the explanation of Example 4, specifically, step three further includes:

[0090] S34. The obtained rotation axis offset daxis, nominal structural dimension parameter Dref, and curvature change of the rotating component along the axial direction. Curvature reference parameters Rotational attitude angle deviation parameters Reference standard parameters for rotational attitude angle deviation After dimensionless processing, the rotational geometric equivalent deviation coefficient, denoted as XDP, is calculated and obtained, as shown in the following formula:

[0091]

[0092] In the formula, w1, w2 and w3 represent weighting coefficients.

[0093] The rotation axis offset represents the weight of the rotational geometric equivalent deviation, and has the highest weight. The rotation axis offset directly reflects the degree of deviation between the overall geometric center of the rotating component and the theoretical rotation center. It is a key factor affecting rotational balance, load distribution uniformity and long-term fatigue accumulation, and plays a fundamental and dominant role in rotational geometric stability.

[0094] The axial curvature variation of a rotating component has the second highest weight in the rotational geometric equivalent deviation. Axial curvature anomaly reflects local or overall bending, deflection, or non-uniform material deformation of the component. It is an important manifestation of the failure of geometric continuity of the rotating component and has a significant impact on stress concentration and vibration response during rotation.

[0095] : Characterizes the influence weight of rotational attitude angle deviation in rotational geometric equivalent deviation, accounting for a relatively low weight; rotational attitude angle deviation mainly reflects the attitude execution error within the rotation angle range, usually affected by installation errors or short-term operating condition disturbances. Its impact on structural geometric integrity is relatively indirect, but it still has important supplementary significance for rotational constraint consistency.

[0096] The rotational geometric equivalent deviation coefficient (XDP) normalizes and weights geometric deviation parameters from different sources and with different dimensions, allowing various geometric deviations to be superimposed and expressed on a unified scale. This forms a rotational geometric equivalent deviation index that can be used for threshold determination and state assessment, avoiding the one-sidedness of judging the overall rotational state by a single geometric parameter.

[0097] In this embodiment, the rotation axis offset, axial curvature change and rotation attitude angle deviation are normalized in a dimensionless manner, and a weighted fusion method is used to construct the rotational geometric equivalent deviation coefficient. This enables geometric deviations from different physical dimensions and scales to be comprehensively characterized under a unified evaluation framework, thereby achieving a quantitative assessment of the overall rotational geometric state of rotating equipment and improving the stability and comparability of geometric anomaly determination.

[0098] Example 6

[0099] Please see Figure 1 In the explanation of Example 5, specifically, step three further includes:

[0100] S35. By setting a preset rotational geometric equivalent deviation threshold, denoted as Xth, and comparing the rotational geometric equivalent deviation coefficient XDP with the rotational geometric equivalent deviation threshold Xth, the first evaluation result is obtained, including:

[0101] When the rotational geometric equivalent deviation coefficient XDP ≤ rotational geometric equivalent deviation threshold Xth, it indicates that the rotational geometry of the rotating component is qualified and should be continuously monitored.

[0102] When the rotational geometric equivalent deviation coefficient XDP > the rotational geometric equivalent deviation threshold Xth, it indicates that the rotational geometry of the rotating component is unqualified, and the rotating component has spatial geometric residual field characteristics that cannot be explained by the rotational constraint conditions. This triggers the first warning instruction and generates the first strategy: to mark the current rotational geometry as a potential deformation risk state, and to use the corresponding rotational state normalized point cloud P2 as an abnormal candidate sample, and to start the quality reliability assessment mechanism.

[0103] The method for obtaining the rotational geometric equivalent deviation threshold Xth is as follows: By statistically analyzing point cloud data and operating parameters obtained from a large number of rotating equipment under normal operating conditions, slight off-center load conditions, and abnormal operating conditions, the comprehensive distribution characteristics of geometric deviation indicators such as rotation axis offset, axial curvature change, and rotational attitude angle deviation are extracted. Combined with equipment structural design tolerances, assembly allowable errors, and geometric stability requirements under rotational constraints, the reasonable critical range of rotational geometric equivalent deviation is determined. At the same time, referring to relevant mechanical structure geometric accuracy specifications, installation and operation deviation limits given by rotating equipment manufacturers, and integrating long-term engineering operation experience, the rotational geometric equivalent deviation threshold Xth is formulated to accurately reflect the acceptable degree of geometric deviation of rotating components under normal rotational constraints and to promptly identify abnormal spatial geometric residual risks that cannot be explained by the ideal rotation model.

[0104] In this embodiment, by introducing the rotational geometric equivalent deviation threshold Xth in step three and quantitatively comparing it with the rotational geometric equivalent deviation coefficient XDP, an objective determination of the rotational geometric state of the rotating component is achieved. This not only avoids unnecessary warnings and interventions when the rotational geometric state is normal, but also promptly identifies potential deformation risks when spatial geometric residual field characteristics that exceed the interpretable range of rotational constraints appear. At the same time, by automatically including the normalized point cloud P2 of the rotational state corresponding to the anomaly into the anomaly candidate sample and triggering the quality reliability assessment mechanism, subsequent analysis and decision-making are based on highly reliable data, thereby significantly improving the accuracy of early identification of deformation of rotating components, the pertinence of warnings, and the reliability of overall geometric state monitoring.

[0105] Example 7

[0106] Please see Figure 1In the explanation of Example Six, this embodiment specifically includes step four:

[0107] S41. Initiate the quality reliability assessment mechanism, and based on the rotation state normalized point cloud P2 in the abnormal candidate samples, use the point cloud neighborhood consistency analysis method to perform axial continuity and structural symmetry constraint analysis on the point cloud of rotating equipment, count the number of point cloud points that satisfy the axial continuity and structural symmetry constraint conditions, obtain the effective point cloud point count, denoted as Neff; count the total number of point cloud points corresponding to the current rotating equipment, obtain the total number of point cloud points, denoted as Ntotal;

[0108] S42. Based on the rotation-normalized point cloud P2 from the abnormal candidate samples, the statistical analysis method of adjacent point spacing is used to statistically analyze the changes in point spacing within the local neighborhood of the point cloud, calculate the fluctuation degree of the local density of the point cloud, and obtain the local density fluctuation coefficient of the point cloud, denoted as... ;

[0109] S43. Based on the rotational state normalized point cloud P2 in the abnormal candidate samples, the statistical constraint analysis method based on the spatial distribution consistency of the point cloud is adopted to comprehensively evaluate the spatial distribution continuity and distribution consistency of the point cloud in the axial and radial directions under the rotational constraint coordinate system, and obtain the structural self-consistency score parameter, denoted as Scon.

[0110] In this embodiment, a quality reliability assessment mechanism based on the normalized point cloud P2 of the rotation state of abnormal candidate samples is introduced in step four. This mechanism comprehensively utilizes the effective point cloud number Neff and the local density fluctuation coefficient of the point cloud under the constraints of axial continuity and structural symmetry. In addition, the structural self-consistency scoring parameter Scon is used to conduct multi-dimensional quantitative evaluation of the structural reliability and data stability of point clouds of rotating equipment. It can effectively distinguish between rotational geometric deviations caused by real structural anomalies and false anomalies caused by point cloud noise, occlusion or uneven sampling, thereby significantly reducing the probability of false alarms, improving the credibility and robustness of anomaly judgment results, and providing a stable and reliable data foundation for subsequent risk decisions and status assessments.

[0111] Example 8

[0112] Please see Figure 1 In the explanation of Example 7, specifically, step four further includes:

[0113] S44. Obtain the effective point cloud point count Neff, the total point cloud point count Ntotal, and the local density fluctuation coefficient of the point cloud. The structural self-consistency scoring parameter Scon, after dimensionless processing, is used to calculate the point cloud structure confidence coefficient, denoted as DJK, as follows:

[0114]

[0115] In the formula, a1, a2, and a3 represent weighting coefficients;

[0116] The effective point cloud point ratio represents the weight of the point cloud structure credibility, and has the highest weight. The effective point cloud point ratio directly reflects the available information density of the point cloud under rotation constraints and structural consistency conditions, and is a basic indicator of whether the point cloud can stably support geometric analysis.

[0117] The weight of the influence of the degree of local density fluctuation in point cloud on the reliability of point cloud structure is: local density fluctuation reflects the scanning unevenness, occlusion or noise interference in the point cloud acquisition process, and has a significant impact on the stability of local geometric feature extraction.

[0118] : Characterizes the influence weight of structural self-consistency score in the credibility of point cloud structure; structural self-consistency is used to measure the overall continuity and symmetry of point cloud in axial and radial spatial distribution, and is an important supplementary indicator for judging whether point cloud truly reflects the structural morphology of rotating components.

[0119] The DJK point cloud structure credibility coefficient integrates multi-dimensional quality features such as the number of points, spatial distribution stability, and structural consistency to construct a comprehensive quantitative evaluation of the "interpretability" and "credibility" of point clouds, avoiding the risk of introducing false geometric residuals due to local acquisition anomalies or noise amplification.

[0120] S45. By setting a preset point cloud structure credibility threshold, denoted as Dth, and comparing the point cloud structure credibility coefficient DJK with the point cloud structure credibility threshold Dth, the second evaluation results are obtained, including:

[0121] When the point cloud structure reliability coefficient DJK ≥ the point cloud structure reliability threshold Dth, it indicates that the point cloud data quality corresponding to the current rotating device is reliable and should be continuously monitored.

[0122] When the point cloud structure reliability coefficient DJK < the point cloud structure reliability threshold Dth, it indicates that the point cloud data quality corresponding to the current rotating device is unreliable. The spatial geometric residual field has the risk of false deformation, abnormal amplitude amplification, and abnormal spatial position distortion caused by fluctuations in point cloud acquisition quality. This triggers a second warning instruction and generates a second strategy: For the rotation state normalized point cloud P2 in the abnormal candidate samples, perform a reprocessing operation oriented towards overall structural consistency; perform resampling processing under structural consistency constraints on the overall distribution of the point cloud; perform completion or smoothing constraint processing on the continuous interrupted regions in the spatial geometric residual field; and perform overall weakening processing on the point cloud regions that cannot meet the requirements of rotation constraints and structural consistency.

[0123] The point cloud structure reliability threshold Dth is obtained by: comparing and analyzing the quality of point cloud data generated by multiple batches of rotating devices under different acquisition distances, viewing angles, rotation speeds, and environmental conditions; statistically analyzing the percentage of valid point cloud points, the local density fluctuation characteristics of the point cloud, and the distribution range of the structure self-consistency score under reliable acquisition conditions; and combining the nominal accuracy, scanning resolution, and noise characteristics of the point cloud acquisition system to determine the baseline level of point cloud structure reliability. Simultaneously, referring to relevant technical standards for 3D point cloud quality assessment, data reliability indicators provided by sensor manufacturers, and incorporating the experience of engineering technicians in assessing point cloud availability, a point cloud structure reliability threshold Dth is established to objectively distinguish between true geometric features of the structure and false anomalies caused by acquisition quality fluctuations, ensuring the reliability of subsequent geometric analysis and state assessment.

[0124] In this embodiment, by constructing a point cloud structure credibility coefficient DJK that integrates the effective point cloud ratio, local density stability, and structural self-consistency characteristics, and combining it with a preset credibility threshold for hierarchical evaluation and strategy triggering, this invention can perform prior judgment and dynamic correction of point cloud data quality during rotational geometric anomaly analysis. This effectively suppresses the risk of false deformation and anomaly amplification effect caused by acquisition noise, occlusion interference, or uneven sampling. Under the premise of ensuring that the real structural anomalies are not masked, it significantly improves the stability and reliability of spatial geometric residual field analysis results, thereby enhancing the accuracy and engineering applicability of overall rotating equipment status assessment and early warning decision-making.

[0125] Example 9

[0126] Please see Figure 1 In the explanation of Embodiment Eight, specifically, step five includes:

[0127] S51. The rotational geometric equivalent deviation coefficient XDP and the point cloud structure reliability coefficient DJK, obtained through calculation, are processed dimensionlessly and then weighted fusion is used to calculate the rotational safety margin attenuation coefficient, denoted as ASX, as follows:

[0128]

[0129] In the formula, s1 and s2 represent weighting coefficients.

[0130] The weight representing the impact of rotational geometric equivalent deviation on safety margin decay is dominant; rotational geometric deviation directly affects load distribution, dynamic response, and structural stress level during equipment operation, and is the main source of safety margin decay.

[0131] The weight of the impact of insufficient point cloud structure credibility on the decay of security margin is represented; the decrease in point cloud credibility will weaken the reliability of geometric evaluation results, make the identification of potential risks uncertain, and have an indirect but not negligible impact on security decisions.

[0132] The Rotational Safety Margin Attenuation Factor (ASX) weights and superimposes the "degree of geometric anomaly" and the "degree of insufficient data credibility," enabling the safety margin assessment to simultaneously reflect both the actual structural risk and the perceived uncertainty risk. This avoids making one-sided judgments based solely on geometric deviations or data quality, thereby achieving a more robust and conservative comprehensive assessment of the operational risks of rotating equipment.

[0133] In this embodiment, a rotational geometric equivalent deviation coefficient (XDP), which characterizes the degree of rotational geometric deviation, and a point cloud structure reliability coefficient (DJK), which reflects the reliability of the point cloud data structure, are weighted and fused to construct a rotational safety margin decay coefficient (ASX). This invention can simultaneously characterize the superimposed impact of the actual geometric degradation trend and data quality uncertainty on the safety margin within the same evaluation framework, avoiding misjudgments caused by relying solely on geometric deviation or point cloud quality as a single factor. This transforms the safety status assessment of rotating equipment from "whether it is abnormal" to a quantitative expression of "the degree of continuous decay of the safety margin," thereby providing a more stable, interpretable, and engineering decision-making valuable basis for subsequent graded early warning and intervention strategies.

[0134] Example 10

[0135] Please see Figure 1 In the explanation of Embodiment Nine, specifically, step five further includes:

[0136] S52. By setting a preset rotational safety margin attenuation threshold, denoted as Ath, and comparing the rotational safety margin attenuation coefficient ASX with the rotational safety margin attenuation threshold Ath, the third evaluation results are obtained, including:

[0137] When the rotational safety margin attenuation coefficient ASX ≤ the rotational safety margin attenuation threshold Ath, it indicates that the rotating equipment is still within the safe operating margin range under the combined influence of geometric state deviation and point cloud quality reliability. The rotating equipment is judged to be in a safe operating state and is continuously monitored.

[0138] When the rotational safety margin attenuation coefficient ASX > the rotational safety margin attenuation threshold Ath, it indicates that the geometric deviation and point cloud quality factors of the rotating equipment have had a cumulative weakening effect on the operational safety margin. It is determined that the current operational safety margin of the rotating equipment is insufficient and there is an operational risk. The third warning instruction is triggered, and the third strategy is generated: the current rotating equipment is marked as a key focus object; the speed limit operation instruction, priority maintenance instruction, or immediate maintenance instruction is triggered; and the corresponding rotational state normalized point cloud, spatial geometric residual field, and various evaluation coefficients are archived as the basis for operation and maintenance decision-making.

[0139] The method for obtaining the rotational safety margin attenuation threshold Ath is as follows: Through comprehensive statistical analysis of the relationship between the geometric state evolution, point cloud data quality changes, and actual operational safety events of rotating equipment during long-term operation, the joint distribution characteristics of the rotational geometric equivalent deviation coefficient and the point cloud structure reliability coefficient under safe and risky operation states are extracted. Combined with the equipment design safety margin, allowable degradation range, and operational reliability requirements, the critical judgment interval for rotational safety margin attenuation is determined. At the same time, referring to relevant rotating machinery safety operation standards, safety margin and degradation tolerance indicators given by equipment manufacturers, and integrating expert experience assessments of multi-factor coupled risks, the rotational safety margin attenuation threshold Ath is formulated to accurately characterize the degree of weakening of the equipment's safe operation capability under the superposition of geometric deviation and data uncertainty, and to achieve graded identification and proactive intervention of operational risks of rotating equipment.

[0140] In this embodiment, by introducing a rotational safety margin attenuation threshold Ath and classifying the safe operating status of rotating equipment based on the rotational safety margin attenuation coefficient ASX, the present invention can transform the combined impact of the cumulative effect of geometric deviation and the uncertainty of point cloud quality on operational safety into a quantifiable and comparable safety margin attenuation criterion, realizing the transformation from post-fault identification to pre-risk warning; when the safety margin is insufficient, it triggers differentiated operation and maintenance strategies such as speed limiting, inspection, or immediate maintenance, and simultaneously solidifies key point cloud data, spatial geometric residual field, and evaluation coefficients as decision-making basis, thereby significantly improving the foresight of operational risk identification of rotating equipment, the pertinence of operation and maintenance decisions, and the traceability of full life cycle safety management.

[0141] It should be noted that all calculation formulas in this application employ regression analysis, including but not limited to machine learning algorithms, to deeply analyze the collected parameters and identify their natural trends and interrelationships. Specialized software, such as Python's Scikit-learn library or the R language, is used to automatically generate mathematical models that match the data. Then, cross-validation and other methods are used to objectively evaluate the model performance, and continuous feedback and optimization are combined to ensure that the created formulas truly reflect the inherent laws of the data, thereby guaranteeing their effectiveness and accuracy. In all calculation formulas in this application, the parameters in each formula undergo dimensionless processing within a consistent range to ensure that different physical quantities are compared on the same scale; dimensionless processing techniques include, but are not limited to, min-max-normalization and Z-score standardization.

[0142] The algorithm of this invention is implemented as a Python script. Before executing the core logic, the program first executes a data loading module (e.g., using the widely used pandas library in Python) configured to read the aforementioned spreadsheet file and load its contents into the program's working memory (e.g., a DataFrame data structure). Subsequent algorithm steps will directly query and retrieve the required configuration parameters from this in-memory data structure.

[0143] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for identifying defects in the contact wire of rotating equipment in rail transit using unmanned aerial vehicles (UAVs), characterized in that, The specific steps include: Step 1: Collect raw 3D point cloud data with spatial coordinate information; simultaneously collect UAV flight attitude parameters, spatial position information and timestamps, and combine them with the design parameter data of the inspected rotating equipment stored in the contact network equipment management system; Step 2: Denoise, resample, and perform attitude compensation on the original point cloud to obtain a standardized point cloud with consistent attitude; extract the point cloud of rotating equipment and identify its axial distribution characteristics, fit the actual rotation axis and construct a rotation constraint coordinate system, and normalize the rotation state of the point cloud; extract the rotation parameter distribution and calculate the deviation between it and the predefined rotation structure reference, and construct a spatial geometric residual field. Step 3: Calculate and obtain the rotational geometric equivalent deviation coefficient, and compare it with the rotational geometric equivalent deviation threshold to make an overall judgment on the rotational geometric state of the rotating component. When the deviation exceeds the limit, mark the corresponding rotational state normalized point cloud as an abnormal candidate sample and trigger the quality reliability assessment process. Step 4: Calculate and obtain the point cloud structure credibility coefficient, and compare it with the point cloud structure credibility threshold to determine the reliability of the point cloud data quality corresponding to the current rotating device. When the credibility is insufficient, trigger an early warning and perform resampling, completion smoothing and weakening processing under the structural consistency constraint on the point cloud to suppress the risk of false deformation and position distortion. Step 5: Calculate the rotational safety margin attenuation coefficient and compare it with the rotational safety margin attenuation threshold to determine whether the rotating equipment is within the safe operating margin range. When the safety margin is insufficient, trigger an early warning and generate operation and maintenance strategies such as speed-limited operation, priority maintenance, or immediate maintenance.

2. The method for identifying UAV point cloud defects in the overhead contact system of rotating rail transit equipment according to claim 1, characterized in that: Step one includes: S11. Using drones equipped with lidar and imaging devices, a preset route is used for inspection flights above the rail transit line along the contact wire. This allows the drones to complete multi-angle and multi-height scanning coverage within the space area where rotating contact wire equipment is located, and ensures that the overall structure of the rotating equipment is within the effective detection range. S12. Using a laser radar scanning device installed on the UAV body, continuously scan the contact wire rotating equipment, collect the original three-dimensional point cloud data of the rotating equipment, and record the spatial coordinate value corresponding to each laser echo point. S13. During the lidar scanning process, the flight attitude data of the UAV, including pitch angle, roll angle and heading angle, are collected synchronously by the inertial measurement equipment and attitude sensing equipment installed on the UAV to form scanning attitude parameters. S14. During the point cloud acquisition process, the spatial location information and corresponding timestamp of the UAV are simultaneously obtained through the satellite positioning module and system clock module carried on the UAV. S15. By accessing the overhead contact line equipment management system, obtain the design parameter data of the rotating equipment being inspected, including the theoretical rotation axis position, rotation center coordinates, design rotation angle range, and nominal dimension parameters of the components.

3. The method for identifying UAV point cloud defects in the overhead contact system of rotating rail transit equipment according to claim 2, characterized in that: Step two includes: S21. Isolated noise points in the original 3D point cloud data are removed by using a neighborhood statistics-based outlier detection method; density equalization of the point cloud is performed through resampling to ensure consistent point cloud distribution at different scanning distances and angles; attitude compensation and coordinate transformation are performed on the point cloud based on scanning attitude parameters to unify point clouds collected with different flight attitudes, different spatial location information, and corresponding timestamps into the same reference coordinate system; a standardized point cloud with consistent attitude and balanced density is obtained. S22. Based on standardized point clouds, a point cloud segmentation method based on spatial position constraints is adopted to extract the point cloud regions corresponding to rotating equipment in the contact network, forming a subset of equipment point clouds; and in the subset of equipment point clouds, axial projection and continuity analysis are performed on the structural features of rotating equipment components that are continuously distributed along the axial direction to obtain the distribution characteristics of the point cloud along the potential rotation direction. S23. Based on the equipment point cloud subset and the distribution characteristics of the point cloud along the potential rotation direction, a point cloud geometric fitting method is used to perform axis fitting processing on the key components of rotating equipment: the main direction vector of the equipment point cloud is calculated by the least squares fitting analysis method; under the main direction constraint, the position and direction parameters of the actual rotation axis are determined to obtain the actual rotation axis parameters; at the same time, the corresponding rotation center position is calculated. S24. Based on the theoretical rotation axis position, rotation center coordinates, design rotation angle range, and nominal dimension parameters of the component, and combined with the actual rotation axis parameters, construct a rotation constraint coordinate system with the rotation axis as the reference; under the current rotation constraint coordinate system, perform coordinate mapping and rotation normalization processing on the standardized point cloud, and uniformly map the point clouds of different scanning postures and different rotation states to the equivalent reference rotation state to obtain the rotation state normalized point cloud; S25. Based on the normalized point cloud of the rotation state, extract the rotation angle distribution, radial offset distribution and axial deformation distribution under rotation constraints to establish the rotation parameter set of the equipment; under the premise of eliminating the influence of normal rotation attitude, calculate the deviation between the normalized point cloud and the predefined rotation structure reference to construct the spatial geometric residual field.

4. The method for identifying UAV point cloud defects in the overhead contact system of rotating rail transit equipment according to claim 3, characterized in that: Step three includes: S31. Based on the actual rotation axis parameters and the theoretical rotation axis position, calculate the minimum spatial distance between them in the rotation constraint coordinate system to obtain the rotation axis offset; based on the nominal size parameters of the component, determine the reference normalized size of the rotation axis offset to obtain the nominal structural size parameters. S32. Based on the normalized point cloud of the rotation state, the point cloud is segmented axially along the rotation axis, and local geometric fitting is performed on each axial segment to extract the local curvature value corresponding to each segment; the curvature change of the rotating component along the axis is obtained based on the difference in curvature values ​​of adjacent axial segments; and the curvature reference value of the rotating component under normal conditions is obtained through the statistical results of the point cloud of rotating equipment under historical normal working conditions, and the curvature reference parameter is determined. S33. Based on the set of rotation parameters, calculate the actual rotation attitude parameters under the current rotation state; based on the design rotation angle range, determine the corresponding design rotation attitude reference; calculate the angle difference between the actual rotation attitude parameters and the design rotation attitude reference to obtain the rotation attitude angle deviation parameters; based on the design rotation angle range, determine the reference standard parameters for the rotation attitude angle deviation.

5. The method for identifying UAV point cloud defects in the overhead contact system of rotating rail transit equipment according to claim 4, characterized in that: Step three also includes: S34. After dimensionless processing of the obtained rotation axis offset, nominal structural dimension parameters, axial curvature change of rotating components, curvature reference parameters, rotation attitude angle deviation parameters, and reference standard parameters of rotation attitude angle deviation, the rotation geometric equivalent deviation coefficient is calculated.

6. The method for identifying UAV point cloud defects in the overhead contact system of rotating rail transit equipment according to claim 5, characterized in that: Step three also includes: S35. By setting a preset rotational geometric equivalent deviation threshold, and comparing and analyzing the rotational geometric equivalent deviation coefficient with the rotational geometric equivalent deviation threshold, the first evaluation result is obtained, including: When the rotational geometric equivalent deviation coefficient is less than or equal to the rotational geometric equivalent deviation threshold, it indicates that the rotational geometry of the rotating component is qualified and should be continuously monitored. When the rotational geometric equivalent deviation coefficient is greater than the rotational geometric equivalent deviation threshold, it indicates that the rotational geometry of the rotating component is unqualified and that the rotating component has spatial geometric residual field characteristics that cannot be explained by the rotational constraints. This triggers the first warning instruction and generates the first strategy: to mark the current rotational geometry as a potential deformation risk state, and to use the corresponding rotational state normalized point cloud as anomaly candidate sample, and to start the quality reliability assessment mechanism.

7. The method for identifying UAV point cloud defects in the overhead contact system of rotating rail transit equipment according to claim 6, characterized in that: Step four includes: S41. Initiate the quality reliability assessment mechanism, and based on the rotation state normalized point cloud in the abnormal candidate samples, use the point cloud neighborhood consistency analysis method to perform axial continuity and structural symmetry constraint analysis on the point cloud of rotating equipment, count the number of point cloud points that meet the axial continuity and structural symmetry constraint conditions, and obtain the number of valid point cloud points; count the total number of point cloud points corresponding to the current rotating equipment, and obtain the total number of point cloud points. S42. Based on the rotation state normalized point cloud in the abnormal candidate samples, the statistical analysis method of the distance between adjacent points is used to statistically analyze the changes in the distance between points in the local neighborhood of the point cloud, calculate the degree of fluctuation of the local density of the point cloud, and obtain the local density fluctuation coefficient of the point cloud. S43. Based on the rotational state normalized point cloud in the abnormal candidate samples, a statistical constraint analysis method based on the consistency of point cloud spatial distribution is adopted to comprehensively evaluate the continuity and consistency of the spatial distribution of point cloud in the axial and radial directions under the rotational constraint coordinate system, and obtain the structural self-consistency score parameters.

8. The method for identifying UAV point cloud defects in the overhead contact system of rotating rail transit equipment according to claim 7, characterized in that: Step four also includes: S44. After dimensionless processing of the obtained effective point cloud points, total point cloud points, point cloud local density fluctuation coefficient, and structure self-consistency scoring parameters, the point cloud structure reliability coefficient is calculated. S45. By setting a preset point cloud structure credibility threshold, and comparing the point cloud structure credibility coefficient with the point cloud structure credibility threshold, the second evaluation result is obtained, including: When the point cloud structure reliability coefficient is greater than or equal to the point cloud structure reliability threshold, it indicates that the point cloud data quality corresponding to the current rotating device is reliable and should be continuously monitored. When the point cloud structure reliability coefficient is less than the point cloud structure reliability threshold, it indicates that the point cloud data quality corresponding to the current rotating device is unreliable. The spatial geometric residual field has the risk of false deformation, abnormal amplitude amplification, and abnormal spatial position distortion caused by fluctuations in point cloud acquisition quality. This triggers a second warning instruction and generates a second strategy: for the rotation state normalized point cloud in the abnormal candidate samples, perform a reprocessing operation oriented towards overall structural consistency; perform resampling processing under structural consistency constraints on the overall distribution of the point cloud; perform completion or smoothing constraint processing on the continuous interrupted regions in the spatial geometric residual field; and perform overall weakening processing on the point cloud regions that cannot meet the requirements of rotation constraints and structural consistency.

9. The method for identifying UAV point cloud defects in the overhead contact system of rotating rail transit equipment according to claim 8, characterized in that: Step five includes: S51. The rotational geometric equivalent deviation coefficient and point cloud structure reliability coefficient obtained through calculation are processed without dimensions, and a weighted fusion calculation method is used to calculate the rotational safety margin attenuation coefficient.

10. The method for identifying UAV point cloud defects in the overhead contact system of rotating rail transit equipment according to claim 9, characterized in that: Step five also includes: S52. By setting a preset rotational safety margin attenuation threshold and comparing the rotational safety margin attenuation coefficient with the rotational safety margin attenuation threshold, the third evaluation results are obtained, including: When the rotational safety margin attenuation coefficient is less than or equal to the rotational safety margin attenuation threshold, it indicates that the rotating equipment is still within the safe operating margin range under the combined influence of geometric state deviation and point cloud quality reliability. The rotating equipment is judged to be in a safe operating state and is continuously monitored. When the rotational safety margin attenuation coefficient exceeds the rotational safety margin attenuation threshold, it indicates that the geometric deviation and point cloud quality factors of the rotating equipment have had a cumulative weakening effect on the operational safety margin. It is determined that the current operational safety margin of the rotating equipment is insufficient and there is an operational risk. The third warning instruction is triggered, and the third strategy is generated: the current rotating equipment is marked as a key focus object; the speed limit operation instruction, priority maintenance instruction, or immediate maintenance instruction is triggered; and the corresponding rotational state normalized point cloud, spatial geometric residual field, and various evaluation coefficients are archived as the basis for operation and maintenance decision-making.

Images