A Surveying Method for Major Overhaul of Conventional Railways During Operation Based on Full-Element Track Measurement

By employing a comprehensive track measurement method, combined with lidar and GNSS inertial navigation technology, efficient, safe, and accurate measurements were achieved for the overhaul inspection of conventional railways. This solved the problems of low efficiency and insufficient accuracy in traditional manual measurements and enabled multi-task collaborative inspection.

CN123065417APending Publication Date: 2026-11-10CHINA RAILWAY DESIGN GRP CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611337035.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-31
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

The overhaul and inspection of conventional railways is inefficient and poses high safety risks. Traditional manual measurement methods are difficult to meet the high-precision requirements, and the integration of intelligent equipment is low, with multiple online operations restricting maintenance efficiency.

Method used

By adopting a method based on full-element track measurement, combined with lidar, GNSS and inertial navigation technologies, and through point cloud data processing and combined calculation, high-precision detection of track geometric parameters is achieved, integrating multiple detection tasks and reducing the number of personnel required for on-line operations.

Benefits of technology

It improves the efficiency of overhaul and inspection of conventional railways, meets the requirements of high-precision measurement, reduces the number of personnel on the line, lowers costs, and enables multi-task collaborative inspection.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123065417A_ABST
    Figure CN123065417A_ABST
Patent Text Reader

Abstract

This invention discloses a comprehensive track measurement method for overhauling conventional railways during operation, comprising: S1, acquiring point cloud data of the track section using lidar; obtaining acceleration and angular velocity data using inertial navigation; and acquiring coordinate and time information using GNSS; S2, preprocessing the point cloud data; using a loosely combined inertial navigation and GNSS solution to obtain high-precision attitude and positioning data for a fixed section, converting it into track geometric parameters to obtain the original trajectory line; S3, synchronizing the track geometric parameters with the point cloud data in time and space; S4, processing and analyzing the point cloud data to extract the elevation control line elevation of the contact wire poles; and S5, performing overall adjustment calculations within each segment to obtain the optimal track alignment for each segment, and outputting the track plane and elevation information for the entire line. This method uses elevation constraints to meet the accuracy requirements of overhaul measurement, improves detection efficiency, and reduces detection costs.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of overhaul and inspection technology for conventional railways, specifically to an operational overhaul measurement method for conventional railways based on full-element track measurement. Background Technology

[0002] Major overhauls of conventional railways involve more than twenty specialized surveys, currently relying primarily on manual measurement for each item. The longitudinal and horizontal profile measurements of the line follow the traditional "three-way connection" method (mileage measurement via 100-meter markers, horizontal mapping via RTK, and elevation measurement via leveling). Equipment and facilities are measured manually using steel tapes and distance measuring instruments, resulting in low efficiency, high manpower requirements, significant track maintenance time, and high safety risks. During the overhaul design phase, complex boundary conditions such as catenary conductor pull-out values, bridge eccentricity, and ballast thickness must be comprehensively considered. Traditional design methods heavily rely on manual experience and judgment, leading to low design efficiency and difficulty in quality control.

[0003] Significant progress has been made in railway line overhaul and maintenance technology abroad in recent years, moving towards automated inspection. This technology enables rapid track inspection using precision measuring instruments and intelligent processing methods. For example, European and American countries such as the UK, US, France, and Germany employ multi-sensor integrated track inspection vehicles that can measure track geometry at high speed; or they integrate high-precision 3D scanners to complete tasks such as track cross-section and centerline measurements, and can quickly generate overhaul inspection plans through intelligent system data analysis. While these methods are highly efficient, the integrated systems are generally large, making them difficult to transport and operate.

[0004] Domestic research on key technologies for overhaul inspection is relatively limited, and a systematic inspection method has not yet been formed. Currently, the mainstream operation mode still relies primarily on traditional manual measurement. Although intelligent inspection equipment has been introduced in some scenarios, such as using inertial navigation track inspection instruments to complete track geometry detection and using 3D scanning systems to achieve track cross-section detection, effective integration has not been achieved, requiring multiple online operations and hindering maintenance efficiency. Furthermore, overhaul inspection has certain requirements for the accuracy of track longitudinal profiles, which is difficult to meet directly using track inspection instruments. While leveling measurements at regular intervals can improve accuracy, each inspection requires multiple teams of workers to coordinate, further restricting the overhaul progress. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention provides a method for overhaul measurement of conventional railways during operation based on full-element track measurement, in order to improve the detection efficiency of overhaul measurement of conventional railways.

[0006] Therefore, the present invention adopts the following technical solution:

[0007] A method for overhauling conventional railways during operation based on full-element track measurement includes the following steps:

[0008] S1, point cloud data of the line section with GNSS time information is collected by lidar; continuous acceleration and angular velocity data are obtained by inertial navigation; coordinate and time information are obtained by GNSS; wherein: the point cloud data includes contact wire poles and track line data, and during the collection process, a precision rigid plate is set at the elevation control line of each contact wire pole, the precision rigid plate is perpendicular to the contact wire pole, forming an L-shaped intersection;

[0009] S2, preprocess the point cloud data obtained in S1; use a loosely combined inertial navigation and GNSS combination solution to obtain high-precision attitude and positioning data for a fixed interval; convert the high-precision attitude and positioning data for the fixed interval into orbital geometric parameters to obtain the original trajectory line of the integrated navigation.

[0010] S3, synchronize the orbital geometry parameters obtained in S2 with the preprocessed point cloud data in both time and space;

[0011] S4 processes and analyzes the point cloud data obtained in S3, including feature region segmentation, geometric modeling and intersection point determination, and automatically extracts the elevation control line elevation of each contact wire pole.

[0012] S5 treats the area between two adjacent control points as an independent segment, and constructs the position observation equation, control point strong constraint equation and smoothing constraint equation within this segment. It then performs overall adjustment calculation to obtain the smoothed optimal track alignment of the segment. After obtaining the smoothed optimal track alignment of each segment, it outputs the track plane and elevation information of the entire line to complete the full-element track measurement.

[0013] The specific operation of obtaining high-precision attitude and positioning data for a fixed interval using a loosely combined inertial navigation and GNSS solution in step S2 above is as follows:

[0014] By performing initial alignment and mechanical arrangement on the acceleration, angular velocity and other information collected by the S1 inertial navigation system, the inertial navigation calculation is completed to obtain position, velocity and attitude information;

[0015] By utilizing the satellite carrier phase observations of the base station receiver and the GNSS rover receiver under synchronous observation, the relative position positioning deviation between the two receivers is determined through carrier phase double-difference positioning, which is then converted into the positioning point coordinates of the rover in the local coordinate system to complete the GNSS data processing.

[0016] Based on the position, velocity, and attitude information obtained from the inertial navigation system (INS) calculation and the coordinates of the positioning point obtained from the GNSS data calculation, the state parameters are calculated using Kalman filtering and fed back into the mechanical arrangement of the INS calculation process to correct the INS calculation results. At the same time, reverse smoothing filtering is performed in a fixed interval within two constraint points.

[0017] The two constraint points are the starting point and the ending point of the inertial navigation calculation, respectively, and appropriate constraint points can be selected as needed;

[0018] The coordinates of the original trajectory points included in the original trajectory line are the coordinates of the track center point included in the track geometry parameters.

[0019] In step S3 above: time synchronization is to align to a unified time reference according to GNSS time; spatial synchronization is to transform to a unified spatial coordinate system.

[0020] The specific operation of feature region segmentation in step S4 above is as follows:

[0021] Candidate L-shaped bend regions are extracted from the point cloud data obtained from S3 using region growing segmentation and orientation histogram constraints. These regions include two linear edge regions, denoted as R1 and R2.

[0022] The region growth and segmentation are implemented using a parallel region growth algorithm.

[0023] The specific operation of the orientation histogram constraint is as follows: construct a three-level pyramid point cloud, calculate the main direction of each layer of point cloud, and generate a 32-bin orientation histogram; use the L-shaped orthogonality property to select cluster pairs (R1, R2) with a main direction angle > 80° as candidate corner regions.

[0024] The specific operations for geometric modeling described in step S4 above are as follows:

[0025] The RANSAC algorithm is executed synchronously on the three-level pyramid point cloud. At the low scale, an initial model is constructed, and at the high scale, the parameters of the model are optimized to obtain the fitted linear model of R1 and R2.

[0026] Orthogonality verification of the fitted linear models of R1 and R2: Calculate the dot product of the direction vectors of the two fitted linear models. If the dot product of the direction vectors is less than 0.1, then the two fitted linear models are orthogonal and acceptable.

[0027] In step S4 above: the intersection point q of the two fitted straight line models is solved by the least squares method, the elevation value of the intersection point q is extracted, and the elevation control line elevation of the contact wire pole is obtained. .

[0028] In step S5 above:

[0029] Let the first root and first The center points of the elevation control lines on the contact wire poles are respectively used as control points. and A segment is defined between two control points, and the segmentation criteria are consistent with those of the fixed interval, with each segment corresponding to the other one-to-one. This is recorded as the starting point of the segment. Control points This is denoted as the end point of the segment. ,in, ;

[0030] The trajectory line obtained based on the two control points and the S2 contained between the two control points From the original trajectory points, we can construct the original trajectory point set. ,in, , The trajectory line obtained by S2 between the two control points One original trajectory point;

[0031] Among them, any original trajectory point ,in, East coordinates, North coordinates, This represents the elevation value.

[0032] The coordinates of the control points mentioned in step S5 above are obtained in the following way:

[0033] Let the control point be The control point The east and north coordinates are the coordinates of the constraint points used in the inertial navigation solution obtained in S2; the control points elevation value The elevation control line of the contact wire pole obtained from S4 is converted, and the calculation formula is as follows:

[0034] ,

[0035] In the formula, This indicates that the phase center of the inertial navigation system is high. This indicates that the phase center of the lidar is high.

[0036] In step S5 above:

[0037] Let the trajectory curve model corresponding to the optimized trajectory curve of the current segment be: ,in, These are the curve parameters to be solved. The trajectory curve model is a variable. Can be based on the received and Output a three-dimensional coordinate;

[0038] The position observation equation is:

[0039] ,

[0040] In the formula, In order to be with the first The position residuals corresponding to each original trajectory point;

[0041] The strong constraint equation for the control points is:

[0042] ,

[0043] ,

[0044] In the formula, The starting coordinates are calculated for the trajectory curve model. The endpoint coordinates calculated for the trajectory curve model;

[0045] The smoothing constraint equation is:

[0046] ,

[0047] in, In order to be with the first The positional slope residuals corresponding to each original trajectory point; For and The distance between the coordinates output by the trajectory curve model; For and The distance between the coordinates output by the trajectory curve model.

[0048] In step S5 above:

[0049] The above position observation equations, control point strong constraint equations, and smoothing constraint equations are integrated into a unified least squares adjustment problem, as shown in the following formula:

[0050] ,

[0051] In the formula, For the total weighted sum of squared residuals, In order to be with the first The position observation matrix corresponding to each original trajectory point In order to be with the first The weighted sum of squared position residuals corresponding to each original trajectory point As the global weight of slope, In order to be with the first The weighted sum of squares of the slope residuals corresponding to each original trajectory point;

[0052] To minimize the total weighted sum of squared residuals, complete the overall adjustment and obtain the optimal curve parameters. ;

[0053] Then through Calculate the optimized position corresponding to each original trajectory point in the current segment. A set of points with corrected positions is obtained, and then smoothed using cubic spline curve fitting to obtain the optimal smoothed track shape for that segment, including track plane and elevation information.

[0054] Compared with the prior art, the method of the present invention has the following beneficial effects:

[0055] 1. This invention proposes a method for full-element track measurement to meet the multi-task collaborative detection requirements of railway overhaul measurement. It integrates inertial measurement technology, three-dimensional scanning technology, and GNSS positioning technology, and develops corresponding functions to meet the needs of overhaul measurement tasks.

[0056] 2. This invention utilizes a high-precision inertial navigation and GNSS integrated navigation solution method to rapidly complete track geometry detection, position coordinate measurement, and mileage measurement for conventional railways. Through high-precision lidar and GNSS detection methods, it quickly completes tasks such as catenary guide height and pull-out value detection, elevation control line elevation measurement, clearance detection, shoulder elevation measurement, and equipment and facility surveys during major overhauls of conventional railways. Furthermore, by integrating the two detection methods, multiple overhaul tasks, including track inspection and cross-section measurement, can be completed simultaneously in a single operation, greatly improving detection efficiency, reducing the number of personnel required for on-line operations, and lowering detection costs.

[0057] 3. In horizontal elevation measurement, this invention uses only the horizontal position information provided by GNSS and quickly obtains the elevation through an automatic elevation control line elevation extraction algorithm. It uses lidar to measure line features instead of GNSS to provide elevation information. Compared to existing inertial navigation and GNSS combined navigation methods, lidar elevation measurement is more accurate and meets the elevation measurement accuracy requirements during major overhauls. Furthermore, compared to GNSS horizontal and leveling elevation measurements, at least three fewer personnel can be required for each major overhaul online measurement.

[0058] 4. This invention is applicable to the overhaul measurement of conventional railways under various railway bureaus. It can obtain high-precision track geometric parameters and high-precision cross-sectional point cloud data at the corresponding mileage of the line. The measurement results can provide data reference for overhaul and renovation plans. Attached Figure Description

[0059] Figure 1 This is a flowchart of a measurement method for major overhaul of conventional railways during operation based on full-element track measurement, according to an embodiment of the present invention.

[0060] Figure 2 This is a schematic diagram of the intersection of the L-shaped bend formed by the precision rigid plate and the contact wire rod in an embodiment of the present invention;

[0061] Figure 3 This is a diagram showing the relative positions of the sensors in this invention;

[0062] Figure 4 This is a point cloud diagram of a line cross-section in an embodiment of the present invention;

[0063] Figure 5 This is a schematic diagram of the transformation between two coordinate systems in space according to an embodiment of the present invention;

[0064] Figure 6 This is a schematic diagram of elevation control line elevation measurement in an embodiment of the present invention. Detailed Implementation

[0065] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and embodiments.

[0066] Example

[0067] A comprehensive track measurement method for major overhauls of conventional speed railways during operation, such as... Figure 1 As shown, it includes the following steps:

[0068] S1, Data Acquisition, specifically includes:

[0069] Point cloud data of the track section with GNSS time information was collected using lidar, including data on contact wire poles and track lines. During the data collection process, a precision rigid plate was installed at the elevation control line (red line elevation) of each contact wire pole, perpendicular to the contact wire pole (e.g., ...). Figure 2 As shown in the figure, an L-shaped intersection is formed. Continuous acceleration and angular velocity data are obtained using inertial navigation; coordinate (position) and time information are obtained using GNSS (GNSS rover receiver).

[0070] The relative positions of the various sensors (GNSS, inertial navigation, and lidar) on the platform of the track inspection vehicle are as follows: Figure 3 As shown.

[0071] like Figure 4 The image shown is a point cloud diagram of a line cross-section collected by lidar.

[0072] S2, Data Preprocessing, as follows:

[0073] S21 preprocesses the point cloud data obtained in S1, including multi-scale voxel filtering, statistical outlier removal, and normal estimation optimization. This process also includes scanning noise reduction, unified point cloud density, and optimized feature representation, which is beneficial for subsequent point cloud segmentation.

[0074] The specific operation of the multi-scale voxel filtering is as follows:

[0075] Point cloud data is downsampled using dynamic voxel meshes, with voxel sizes... Adaptive adjustment of voxel size based on local point cloud density P. The adjustment formula is as follows:

[0076] ,

[0077] In this embodiment, the basic dimensions =5mm, density sensitivity coefficient =0.3, For the global maximum density, This represents the current neighborhood density of the voxel.

[0078] The specific operation for removing statistical outliers is as follows:

[0079] Based on the K-nearest neighbor (K=50 neighbors in this embodiment) distance distribution, calculate the local mean. with standard deviation Distance removal The point is denoted by 'a', which is an adaptive threshold that is dynamically adjusted according to the noise level of the point cloud.

[0080] The specific operation of the normal estimation optimization is as follows:

[0081] For each point in the point cloud data, an adaptive radius neighborhood is used (the initial radius is set to 0.1m, and it is expanded to 0.2m if the point cloud is sparse). The normal vector is obtained by decomposing the covariance matrix through PCA. The orientation is then unified to point towards the scanner's viewpoint.

[0082] S22, combined inertial navigation and GNSS solution:

[0083] High-precision fusion positioning is achieved by using a loosely coupled inertial navigation and GNSS combination solution. This enables the acquisition of high-precision attitude and positioning data within a fixed area, including information such as mileage, elevation, and plane coordinates. The specific operation is as follows:

[0084] By performing initial alignment and mechanical arrangement of the acceleration, angular velocity and other information collected by the S1 inertial navigation system, the inertial navigation calculation is completed to obtain position, velocity and attitude information.

[0085] By utilizing the satellite carrier phase observations of the base station receiver and the GNSS rover receiver under synchronous observation, the relative positional deviation between the two receivers is determined through carrier phase double-difference positioning, which is then converted into the positioning point coordinates of the rover in the local coordinate system to complete the GNSS data processing.

[0086] Based on the position, velocity, and attitude information obtained from inertial navigation system (INS) calculations and the coordinates of the positioning point (i.e., the GNSS receiver antenna coordinates) obtained from GNSS data calculations, Kalman filtering is used to calculate the state parameters (position, velocity, and attitude information errors), which are then fed back into the mechanical arrangement of the INS calculation process to correct the INS calculation results. Simultaneously, reverse smoothing filtering is performed in a fixed interval between the two constraint points to improve positioning accuracy. The two constraint points are the start and end points of the INS calculation, respectively. Appropriate constraint points can be selected as needed. In this embodiment, the constraint point (containing both coordinate and elevation information) is based on the center point of the elevation control line on two contact wire poles (elevation information only) and the point obtained from the coordinates of the positioning point.

[0087] The high-precision attitude and positioning data of the fixed interval is converted into track geometric parameters (including track center point and track gauge, etc.), and the track measurement in the overhaul inspection is completed to obtain the original trajectory line of the integrated navigation. The coordinates of the original trajectory points contained in the original trajectory line are the coordinates of the track center point contained in the track geometric parameters.

[0088] The initial alignment requires acquiring initial values ​​for position, velocity, and attitude information. These initial values, along with the coordinates of relatively good constraint points in the signal region acquired during the GNSS data processing, are obtained and provided to the inertial navigation system at regular intervals as new starting points. The initial attitude estimation calculation formula in the initial alignment is as follows:

[0089] ,

[0090] ,

[0091] ,

[0092] In the formula, Represents the navigation coordinate system. Represents the sensor coordinate system. Represents the geocentric coordinate system. Represents an inertial coordinate system. The direction cosine matrix of the attitude angles can be converted into three attitude Euler angles. For the accelerometer output matrix, This is the output matrix for the gyroscope. This is the projection matrix of gravity in the navigation coordinate system. This is the projection matrix of the Earth's rotational angular velocity in the navigation coordinate system. and Represents velocity in different coordinate systems.

[0093] S3 synchronizes the orbital geometry parameters obtained in S2 with the preprocessed point cloud data in both time and space, where:

[0094] Since the centers of the various sensors are not located in the same position, spatial synchronization is required to bring them into the same coordinate system. The center positions of each sensor are located and calibrated, and then the coordinate system is transformed to a unified spatial coordinate system to complete spatial synchronization.

[0095] Each pair of spatial rectangular coordinate systems and The conversion is as follows Figure 5 As shown. The two coordinate axes are not parallel and their scales are inconsistent. First, the two coordinate systems are translated so that the origins coincide, and then three rotations are performed to make them coincide. The transformation relationship between the two parallel coordinate systems is shown in the following equation:

[0096] ,

[0097] In the formula, , , The coordinates before transformation , , The coordinates are the transformed coordinates. , , For translation parameters, As a scale factor, The equation is for the rotation matrix.

[0098] Each sensor uses GNSS time, so the orbital geometry parameter data obtained in S22 also carries GNSS time. This data is aligned with the preprocessed point cloud data obtained in S21 to a unified time reference, thus achieving time synchronization.

[0099] S4, Automatic Elevation Extraction of Contact Network Pole Elevation Control Line:

[0100] The point cloud data after S3 synchronization fusion is in the same coordinate system as the track geometric parameter data. The point cloud data obtained from S3 is processed and analyzed to automatically extract the elevation control line elevation of each contact wire pole, as detailed below:

[0101] S41, Feature Region Segmentation: such as Figure 6 As shown, candidate corner regions for L-shaped bends are extracted from the point cloud data, including two linear edge regions of the L-shape, denoted as R1 and R2, which are specifically achieved through region growth segmentation and orientation histogram constraints.

[0102] The region growth and segmentation is implemented using a parallel region growth algorithm (specifically, OpenMP multi-threaded seed point partitioning), with the following parameter settings:

[0103] Seed point selection: Prioritize points with curvature c < 0.1 (low curvature region) and stable normal direction;

[0104] Growth criteria: the angle between the normals of neighboring points < 5° (smoothness threshold), curvature difference < 0.05;

[0105] Termination condition: Number of cluster points 1000 < N < 10 6 (Filtering out small noise clusters and excessively large background clusters).

[0106] In one embodiment of the present invention, the specific operation of the direction histogram constraint is as follows:

[0107] A three-level pyramid point cloud was constructed (scale factor set to 0.5). The main direction of the point cloud (PCA first feature vector) was calculated for each layer of the point cloud, and a 32-bin orientation histogram was generated. Using the L-shaped orthogonality property, cluster pairs (R1, R2) with a main direction angle > 80° were selected as candidate corner regions.

[0108] S42, Geometric Modeling: Fitting the straight line model of the candidate bend region using the RANSAC algorithm and verifying its orthogonality; wherein, the process of obtaining the fitted straight line models R1 and R2 using the RANSAC algorithm is as follows:

[0109] The RANSAC algorithm is executed synchronously on the three-level pyramid point cloud. An initial model is constructed at a low scale (coarse fitting), and the parameters of the model are optimized at a high scale (fine fitting) to obtain a fitted linear model.

[0110] Orthogonality verification of the fitted linear models of R1 and R2: Calculate the dot product of the direction vectors of the two fitted linear models. If the dot product of the direction vectors is less than 0.1 (corresponding to an angle between the lines greater than 84°, tolerating industrial processing errors), then the two fitted linear models are orthogonal and acceptable.

[0111] S43. Solve for the intersection point q of the two fitted straight line models using the least squares method, extract the elevation value of the intersection point q, and obtain the elevation of the contact wire pole's elevation control line. .

[0112] After completing the elevation measurement of the contact wire poles' elevation control lines, the elevation control line elevation and the relative position of the lidar to the inertial navigation center can be converted into control point elevation information for the next step of data recalculation.

[0113] S5. Data recalculation based on control point elevations: Treating the area between two adjacent control points as an independent segment, and within this segment, constructing position observation equations, strong control point constraint equations, and smoothing constraint equations, performing overall adjustment to obtain the optimal track alignment for this segment. After obtaining the smoothed optimal track alignment for each segment, outputting the track plane and elevation information for the entire line, completing the full-element track measurement. The specific operations are as follows:

[0114] S51, Obtain the coordinates of the control points for each segment and the original trajectory point set:

[0115] Let the first root and first Each of the contact wire poles serves as a control point. and Each segment is defined as a point between points, and the segmentation criteria are consistent with those of the fixed interval, with each segment corresponding to a control point. This is recorded as the starting point of the segment. Control points This is denoted as the end point of the segment. ,in, The trajectory line is obtained based on the two control points and the S2 contained between the two control points. From the original trajectory points, we can construct the original trajectory point set. ,in, , The trajectory line is obtained by considering S2 between the two control points. One original trajectory point.

[0116] Among them, any original trajectory point ,in, East coordinates, North coordinates, This represents the elevation value.

[0117] Control Points and The coordinates are obtained in the following way:

[0118] Let the control point be The control point The east and north coordinates are the coordinates of the constraint points used in the inertial navigation solution obtained in S2; the control points elevation value The elevation of the contact wire pole is derived from the elevation control line obtained in S4, and the calculation formula is as follows:

[0119] ,

[0120] In the formula, This indicates that the phase center of the inertial navigation system is high. This indicates that the phase center of the lidar is high.

[0121] The above operations yield the coordinates of the control points for each segment and the original trajectory point set.

[0122] S52, Construct the position observation equation, the strong constraint equation for control points, and the smoothing constraint equation:

[0123] Let the trajectory curve model corresponding to the optimized trajectory curve of the current segment be: ,in, These are the curve parameters to be solved. The trajectory curve model is a variable. Can follow according to the received and Output a three-dimensional coordinate.

[0124] Establish the position observation equation, the strong constraint equation for the control points, and the smoothing constraint equation, where:

[0125] The position observation equation is:

[0126] ,

[0127] In the formula, In order to be with the first The position residuals corresponding to each original trajectory point. Treating observations as noisy, during the optimization process, minimize... The difference between the three-dimensional coordinates corresponding to the optimized curve output.

[0128] The strong constraint equation for the control points is:

[0129] ,

[0130] ,

[0131] In the formula, The starting coordinates are calculated for the trajectory curve model. The coordinates of the endpoint are calculated for the trajectory curve model. Strong control point constraints require that the start and end points of the optimized trajectory curve completely coincide with the known control points.

[0132] The constructed smooth constraint equation is as follows:

[0133] ,

[0134] in, In order to be with the first The position slope residuals corresponding to the original trajectory points Calculation for trajectory curve model Output coordinates and calculations The distance between the output coordinates is used for trajectory curve model calculation. Output coordinates and calculations The distance between the output coordinates.

[0135] To obtain a high-precision track alignment, geometric smoothness constraints are applied to the longitudinal profile alignment. This involves constraining the elevation gradient changes between adjacent points to ensure a smooth transition, as required by the vertical curve requirements. In other words, the objective during optimization is to make the elevation gradient between adjacent points as smooth as possible. The weighted sum of squares is the smallest.

[0136] S53, Overall Adjustment Solution:

[0137] The above position observation equations, control point strong constraint equations, and smoothing constraint equations are integrated into a unified least squares adjustment problem, as shown in the following formula:

[0138] ,

[0139] In the formula, For the total weighted sum of squared residuals, In order to be with the first The position observation matrix corresponding to each original trajectory point In order to be with the first The weighted sum of squared position residuals corresponding to each original trajectory point As the global weight of slope, In order to be with the first The weighted sum of squares of the slope residuals corresponding to each original trajectory point.

[0140] To minimize the total weighted sum of squared residuals, complete the overall adjustment and obtain the optimal curve parameters. .

[0141] S54, optimal linear output:

[0142] Solve for the optimal curve parameters After that, you can pass Calculate the optimized position corresponding to each original trajectory point in the current segment. A set of points with corrected positions is obtained, and then smoothed using cubic spline curve fitting to obtain the optimal smoothed track shape (including track plane and elevation information) for this segment.

[0143] Through the above operations, the smoothed optimal track alignment of each segment is obtained, and then the track plane and elevation information of the entire line are output, completing the full-element track measurement.

[0144] In existing inertial navigation and GNSS combined navigation orbit measurements, the horizontal accuracy can reach 15mm, and the vertical accuracy can reach 25mm. The horizontal measurement accuracy meets the requirements for overhaul inspection. Using the method of providing vertical constraints for inertial navigation with lidar in this invention, the vertical measurement accuracy is better than 20mm, and both horizontal and vertical measurements meet the requirements for overhaul inspection.

[0145] Based on the S5 integrated navigation and lidar elevation control point data, the track geometry parameters are recalculated to complete the final track measurement. Simultaneously, the point cloud data is automatically processed and analyzed to complete the track cross-section measurement, including catenary guide height and pull-out value detection, elevation control line elevation measurement, clearance detection, shoulder elevation measurement, and equipment and facility measurement. Ultimately, full-element track measurement is achieved, enabling multi-task collaborative operation in a single deployment. Measurement accuracy meets overhaul and inspection requirements, while significantly improving measurement efficiency by reducing the number of personnel and lowering measurement costs.

Claims

1. A method for overhauling conventional-speed railways during operation based on full-element track measurement, characterized in that: Includes the following steps: S1, point cloud data of the line section with GNSS time information is collected by lidar; continuous acceleration and angular velocity data are obtained by inertial navigation; coordinate and time information are obtained by GNSS; wherein: the point cloud data includes contact wire poles and track line data, and during the collection process, a precision rigid plate is set at the elevation control line of each contact wire pole, the precision rigid plate is perpendicular to the contact wire pole, forming an L-shaped intersection; S2, preprocess the point cloud data obtained in S1; use a loosely combined inertial navigation and GNSS combination solution to obtain high-precision attitude and positioning data for a fixed interval; convert the high-precision attitude and positioning data for the fixed interval into orbital geometric parameters to obtain the original trajectory line of the integrated navigation. S3, synchronize the orbital geometry parameters obtained in S2 with the preprocessed point cloud data in both time and space; S4 processes and analyzes the point cloud data obtained in S3, including feature region segmentation, geometric modeling and intersection point determination, and automatically extracts the elevation control line elevation of each contact wire pole. S5 treats the area between two adjacent control points as an independent segment, and constructs the position observation equation, control point strong constraint equation and smoothing constraint equation within this segment. It then performs overall adjustment calculation to obtain the smoothed optimal track alignment of the segment. After obtaining the smoothed optimal track alignment of each segment, it outputs the track plane and elevation information of the entire line to complete the full-element track measurement.

2. The measurement method for major overhaul of conventional-speed railways during operation as described in claim 1, characterized in that, The specific operation for obtaining high-precision attitude and positioning data for a fixed interval using a loosely combined inertial navigation and GNSS solution in S2 is as follows: By performing initial alignment and mechanical arrangement on the acceleration, angular velocity and other information collected by the S1 inertial navigation system, the inertial navigation calculation is completed to obtain position, velocity and attitude information; By utilizing the satellite carrier phase observations of the base station receiver and the GNSS rover receiver under synchronous observation, the relative position positioning deviation between the two receivers is determined through carrier phase double-difference positioning, which is then converted into the positioning point coordinates of the rover in the local coordinate system to complete the GNSS data processing. Based on the position, velocity, and attitude information obtained from the inertial navigation system (INS) calculation and the coordinates of the positioning point obtained from the GNSS data calculation, the state parameters are calculated using Kalman filtering and fed back into the mechanical arrangement of the INS calculation process to correct the INS calculation results. At the same time, reverse smoothing filtering is performed in a fixed interval within two constraint points. The two constraint points are the starting point and the ending point of the inertial navigation calculation, respectively, and appropriate constraint points can be selected as needed; The coordinates of the original trajectory points included in the original trajectory line are the coordinates of the track center point included in the track geometry parameters.

3. The measurement method for major overhaul of conventional-speed railways during operation as described in claim 2, characterized in that, In S3: the time synchronization is to align to a unified time reference according to GNSS time; the spatial synchronization is to transform to a unified spatial coordinate system.

4. The measurement method for major overhaul of conventional-speed railways during operation as described in claim 3, characterized in that, The specific operation for feature region segmentation described in S4 is as follows: Candidate L-shaped bend regions are extracted from the point cloud data obtained from S3 using region growing segmentation and orientation histogram constraints. These regions include two linear edge regions, denoted as R1 and R2. The region growth and segmentation are implemented using a parallel region growth algorithm. The specific operation of the orientation histogram constraint is as follows: construct a three-level pyramid point cloud, calculate the main direction of each layer of point cloud, and generate a 32-bin orientation histogram; use the L-shaped orthogonality property to select cluster pairs (R1, R2) with a main direction angle > 80° as candidate corner regions.

5. The measurement method for major overhaul of conventional-speed railways during operation as described in claim 4, characterized in that, The specific operations for geometric modeling described in S4 are as follows: The RANSAC algorithm is executed synchronously on the three-level pyramid point cloud. At the low scale, an initial model is constructed, and at the high scale, the parameters of the model are optimized to obtain the fitted linear model of R1 and R2. Orthogonality verification of the fitted linear models of R1 and R2: Calculate the dot product of the direction vectors of the two fitted linear models. If the dot product of the direction vectors is less than 0.1, then the two fitted linear models are orthogonal and acceptable.

6. The measurement method for major overhaul of conventional-speed railways during operation as described in claim 5, characterized in that, In S4: The intersection point q of the two fitted straight line models is solved by the least squares method, and the elevation value of the intersection point q is extracted to obtain the elevation control line elevation of the contact wire pole. .

7. The measurement method for major overhaul of conventional-speed railways during operation as described in claim 6, characterized in that, In S5: Let the first root and first The center points of the elevation control lines on the contact wire poles are respectively used as control points. and A segment is defined between two control points, and the segmentation criteria are consistent with those of the fixed interval, with each segment corresponding to the other one-to-one. This is recorded as the starting point of the segment. Control points This is denoted as the end point of the segment. ,in, ; The trajectory line obtained based on the two control points and the S2 contained between the two control points From the original trajectory points, we can construct the original trajectory point set. ,in, , The trajectory line obtained by S2 between the two control points One original trajectory point; Among them, any original trajectory point ,in, East coordinates, North coordinates, This represents the elevation value.

8. The measurement method for major overhaul of conventional-speed railways during operation as described in claim 7, characterized in that, The coordinates of the control points mentioned in S5 are obtained in the following way: Let the control point be The control point The east and north coordinates are the coordinates of the constraint points used in the inertial navigation solution obtained in S2; the control points elevation value The elevation control line of the contact wire pole obtained from S4 is converted, and the calculation formula is as follows: , In the formula, This indicates that the phase center of the inertial navigation system is high. This indicates that the phase center of the lidar is high.

9. The measurement method for major overhaul of conventional-speed railways during operation as described in claim 8, characterized in that, In S5: Let the trajectory curve model corresponding to the optimized trajectory curve of the current segment be: ,in, These are the curve parameters to be solved. The trajectory curve model is a variable. Can be based on the received and Output a three-dimensional coordinate; The position observation equation is: , In the formula, In order to be with the first The position residuals corresponding to each original trajectory point; The strong constraint equation for the control points is: , , In the formula, The starting coordinates are calculated for the trajectory curve model. The endpoint coordinates calculated for the trajectory curve model; The smoothing constraint equation is: , in, In order to be with the first The positional slope residuals corresponding to each original trajectory point; For and The distance between the coordinates output by the trajectory curve model; For and The distance between the coordinates output by the trajectory curve model.

10. The measurement method for major overhaul of conventional-speed railways during operation, as described in claim 9, is characterized in that... In S5: The above position observation equations, control point strong constraint equations, and smoothing constraint equations are integrated into a unified least squares adjustment problem, as shown in the following formula: , In the formula, For the total weighted sum of squared residuals, In order to be with the first The position observation matrix corresponding to each original trajectory point In order to be with the first The weighted sum of squared position residuals corresponding to each original trajectory point As the global weight of slope, In order to be with the first The weighted sum of squares of the slope residuals corresponding to each original trajectory point; To minimize the total weighted sum of squared residuals, complete the overall adjustment and obtain the optimal curve parameters. ; Then through Calculate the optimized position corresponding to each original trajectory point in the current segment. A set of points with corrected positions is obtained, and then smoothed using cubic spline curve fitting to obtain the optimal smoothed track shape for this segment, including track plane and elevation information.