Track plate detection device and method for multi-sensor networking mobile measurement

Abstract

The invention relates to a track board detection device and method for multi-sensor networking mobile measurement, and the device is characterized in that a mobile bearing platform is divided into a front equipment region, a middle operation region and a rear equipment region, and the middle operation region is provided with a centralized cabinet with heat dissipation and shielding functions; the core control and synchronization unit comprises an integrated positioning and attitude determination system comprising a global navigation satellite system receiver antenna and an integrated inertial measurement unit and odometer module; the high-precision synchronous controller is arranged in the centralized cabinet and is used for receiving the space-time reference of the integrated positioning and attitude determination system and distributing a system hardware synchronous trigger signal to the multi-sensor networking module; the multi-sensor networking module comprises a surface appearance acquisition module; an internal state sensing module; a macroscopic geometric measurement module; and the integrated data processing unit is internally provided with a data fusion and intelligent analysis algorithm. The system has the advantages that the comprehensive state of the track plate is comprehensively sensed in high-speed movement, and seamless connection from data acquisition to maintenance decision is realized.

Description

Technical Field

[0001] This invention relates to the field of track infrastructure inspection technology, specifically to a track slab inspection device and method using multi-sensor networked mobile measurement. Background Technology

[0002] Ballastless track systems are core infrastructure ensuring high smoothness and safety of train operation. The track slab, as the main load-bearing and force-transmitting component of the system, directly determines the quality of track operation through its long-term performance. In actual service, track slabs are prone to various types of defects, which exhibit significant correlation and co-evolutionary characteristics: first, apparent defects such as surface cracks and concrete spalling; second, geometric deviations such as flatness deviations and abnormal height values; and third, internal hidden defects such as mortar layer separation and voids beneath the slab. Typically, voids beneath the slab lead to uneven stress on the track slab, inducing surface crack propagation, accompanied by geometric parameter drift. Without coordinated detection and correlation analysis of these three types of defects, scientific maintenance and condition-based repair of the track slab will be difficult to achieve.

[0003] Currently, track inspection technology has evolved from traditional manual static inspection to automated mobile inspection. However, existing mobile inspection solutions still suffer from three major systemic technical defects when dealing with the multi-dimensional collaborative inspection requirements of track slabs' "apparent internal geometry," which severely restricts inspection accuracy and maintenance efficiency: First, multi-dimensional detection data exists in a state of "information silos," making effective integration difficult. Existing mobile inspection equipment often employs a "single-point functional breakthrough" sensor configuration: while machine vision-based inspection devices can efficiently capture images of the track slab surface, they cannot simultaneously acquire three-dimensional geometric topography data of the corresponding area; inspection systems equipped with laser scanning modules can accurately measure track geometric parameters, but lack the ability to detect surface material damage and internal structural defects; even when some devices integrate multiple types of sensors such as vision, laser, and ultrasound, the lack of a unified spatiotemporal reference and synchronous triggering mechanism prevents precise matching of timestamps and spatial coordinates of the sensor data—for example, after detecting a surface crack, it is impossible to immediately correlate it with the void under the slab directly below the crack, nor can it determine whether the geometric shape and position around the crack exceed limits. This results in disease diagnosis remaining only at the "superficial recognition" level, unable to trace the cause of the disease, and unable to support targeted repair work.

[0004] Secondly, the accuracy of defect location is insufficient, making precise data tracing impossible. Existing mobile detection devices primarily rely on two methods for positioning: GNSS positioning, which suffers significant accuracy drops when GNSS signals are blocked or malfunctioning, such as in elevated bridges and tunnels; and odometer-based positioning, which suffers from unavoidable cumulative errors, resulting in a mismatch between the detected defect location and the actual track mileage or design station number. This problem of "identifiable defects but unclear locations" forces maintenance personnel to spend considerable time verifying defect locations, severely reducing the accuracy and efficiency of maintenance and repair work.

[0005] Third, the data-to-decision transformation chain is broken, lacking intelligent analysis capabilities. Existing detection systems primarily output massive amounts of raw data (such as images, point clouds, and waveform signals) or isolated single indicators (such as crack length and geometric exceedances), lacking core processing modules capable of deep fusion, intelligent identification, and correlation analysis of multi-source heterogeneous data. This leads to the interpretation of detection results heavily relying on the experience and judgment of professionals: on the one hand, data processing cycles are long and inefficient; on the other hand, it cannot automatically generate comprehensive health status assessment reports for individual track slabs, nor can it output targeted maintenance decision recommendations, failing to meet the actual needs of high-speed railways for "precise maintenance and efficient repair."

[0006] In summary, existing track slab movement detection technologies have significant technical shortcomings in multi-dimensional data fusion, precise defect location, and intelligent decision support. There is an urgent need to develop a technical solution that can achieve integrated collaborative detection, precise location, and intelligent analysis of both the "appearance and internal geometry" to solve the above-mentioned technical problems. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of the prior art and provide a track slab detection device and method with multi-sensor network mobile measurement, which can synchronously, accurately and comprehensively perceive the overall status of the track slab during high-speed movement, and realize seamless connection from data acquisition to maintenance decision-making, thereby truly promoting the development of railway infrastructure operation and maintenance towards intelligence and precision.

[0008] To achieve the above objectives, a multi-sensor networked mobile measurement track slab detection device is designed, comprising: a mobile support platform equipped with wheel sets for autonomous or passive movement on the track, divided into a front equipment area, a middle operating area, and a rear equipment area; the middle operating area having a centralized cabinet with heat dissipation and shielding functions; a core control and synchronization unit, including: an integrated positioning and attitude determination system, comprising a global navigation satellite system receiver antenna and an integrated inertial measurement unit and odometer module, for generating a high-precision spatiotemporal reference; a high-precision synchronization controller, located in the centralized cabinet, receiving the spatiotemporal reference from the integrated positioning and attitude determination system and distributing a unified hardware synchronization trigger signal to the multi-sensor network module; and a multi-sensor group. The network module includes: a surface appearance acquisition module, located at the bottom of the front equipment area, comprising at least one integrated area array high-definition camera and a line laser scanner; an internal state perception module, located behind the surface appearance acquisition module, comprising at least one array-type ground-penetrating radar; a macroscopic geometric measurement module, located at the top of the rear equipment area, comprising a three-dimensional laser scanner; and an integrated data processing unit, located in the central cabinet, which has built-in data fusion and intelligent analysis algorithms for receiving and processing synchronous data from the multi-sensor network module, and unifying multi-source data into the same spatiotemporal coordinate system based on the position and attitude information of the integrated positioning and attitude system, thereby realizing the fusion analysis and integrated evaluation of the track slab's appearance, internal structure, and geometric information.

[0009] Preferably, the present invention further includes: the high-precision synchronization controller is disposed in the centralized cabinet and receives the spatiotemporal reference of the integrated positioning and attitude determination system, and distributes a unified hardware synchronization trigger signal to all sensors in the multi-sensor networking module; the high-precision synchronization controller adopts a hardware synchronization method, and the accuracy of its synchronization trigger signal is higher than 1 microsecond; the integrated positioning and attitude determination system includes a global navigation satellite system receiver antenna disposed on the top of the mobile carrier platform, and an inertial measurement unit and odometer module integrated near the center of gravity of the platform, used to generate a high-precision spatiotemporal reference; when the global navigation satellite system signal fails, the integrated positioning and attitude determination system maintains a relative positioning accuracy of higher than one-thousandth through a combined navigation algorithm of the inertial measurement unit and the odometer module.

[0010] Preferably, the present invention further includes: the surface appearance acquisition module is mounted on the center line of the bottom of the front equipment area of ​​the mobile bearing platform via a rigidly connected mounting bracket, and is located in front of the foremost wheel of the platform, so that the laser stripes of the line laser scanner and the field of view of the area array high-definition camera can completely cover the track path without interference from the wheel; the line laser scanner and the area array high-definition camera are rigidly connected in space and jointly calibrated to form a structured light three-dimensional measurement subsystem, which is used to directly output the three-dimensional deformation and texture information of the track slab surface with absolute coordinates.

[0011] Preferably, the present invention further includes: the transmitting and receiving antenna array of the radar in the internal state perception module is installed at the bottom of the front equipment area of ​​the mobile carrier platform and arranged immediately behind the surface appearance acquisition module to realize the successive detection of the same track slab area; the array-type ground-penetrating radar adopts a multi-channel antenna array, the width of which is greater than the width of a single track slab, and its transmission and acquisition are controlled by the synchronous trigger signal to ensure that the radar scanning profile corresponds precisely to the absolute mileage position of the track slab.

[0012] Preferably, the present invention further includes: the three-dimensional laser scanner is mounted on the top center of the equipment area at the rear of the mobile support platform via a column to obtain an unobstructed wide field of view for acquiring panoramic three-dimensional point clouds of the track slab and ancillary facilities; the data fusion and intelligent analysis algorithms executed by the integrated data processing unit include: a deep learning-based target detection algorithm for automatically identifying surface defects such as cracks and spalling from the images acquired by the area array high-definition camera; a point cloud processing algorithm for automatically extracting macroscopic geometric parameters such as flatness, superelevation, and track gauge of the track slab from the point cloud of the three-dimensional laser scanner; a radar image processing and inversion algorithm for identifying internal defects such as gaps, voids, and water-bearing anomalies in the track slab from the data of the array-type ground-penetrating radar; and a multi-source information association module for superimposing and correlating the identified surface defect locations, internal defect areas, and macroscopic geometric deformations in a unified spatiotemporal coordinate system to generate a comprehensive track slab health status assessment report.

[0013] This invention also provides a track slab detection method using a multi-sensor networked mobile measurement system. The method includes the following steps: S1: System initialization and self-test. The device is started, the core control and synchronization unit performs a self-test, and the integrated positioning and attitude determination system receives and initializes signals from the Global Navigation Satellite System; S2: Synchronous data acquisition. The mobile carrier platform moves along the track, and the high-precision synchronization controller distributes synchronization trigger signals to all sensors in the multi-sensor network module, controlling the surface appearance acquisition module, internal state perception module, and macroscopic geometric measurement module to perform synchronous data acquisition; S3: Data preprocessing and spatiotemporal alignment. The integrated data processing unit receives the collected data from each module and uses the precise position and attitude data provided by the integrated positioning and attitude system to unify all data into the geodetic coordinate system or the line design coordinate system; S4: Multi-source information fusion and intelligent recognition, performs parallel processing on the spatiotemporally aligned data, and performs surface defect identification, internal defect detection and macroscopic geometric parameter extraction respectively; S5: Comprehensive evaluation and decision support, performs correlation analysis on the various defects and parameters identified in step S4, evaluates the comprehensive health status of a single track slab, and generates a status distribution map of the track slabs along the entire line and graded maintenance and repair suggestions based on the continuous detection results of all track slabs in the line.

[0014] Preferably, the present invention further includes: the spatiotemporal alignment mentioned in step S3 specifically includes: using the position, attitude, and velocity information provided by the integrated positioning and attitude determination system at each moment, and through coordinate transformation, accurately assigning absolute geographic coordinates to the image center of the area array high-definition camera, the scanning line of the line laser scanner, the detection profile of the ground-penetrating radar, and the single-frame point cloud of the three-dimensional laser scanner.

[0015] Preferably, the present invention further includes: the correlation analysis mentioned in step S5 specifically includes: when a surface crack is identified at a certain absolute coordinate position, the three-dimensional contour data corresponding to that position is retrieved simultaneously to analyze whether there is deformation, and the ground-penetrating radar profile data directly below that position is retrieved to analyze whether there is internal separation, thereby comprehensively judging the severity level and cause of the disease.

[0016] Preferably, the present invention further includes: the system initialization and self-test in step S1 further includes: the global navigation satellite system receiver in the integrated positioning and attitude determination system searches for and tracks satellite signals, the inertial measurement unit warms up and performs initial alignment, the tightly combined Kalman filter inside the system is started, the high-precision synchronization controller generates test synchronization pulses, the hardware trigger link with all sensors is verified, each sensor in the multi-sensor networking module reports its own status, the integrated data processing unit starts the server, and loads the data acquisition software, the track slab design model and the pre-trained deep learning model.

[0017] Preferably, the present invention further includes: in step S2, the mobile carrier platform travels along the track at a constant speed, the high-precision synchronous controller uses its internal high-stability crystal oscillator as the frequency reference, receives the 1-second pulse signal from the integrated positioning and attitude determination system for synchronous correction, generates a hardware trigger pulse at a fixed frequency of 100Hz, the rise time jitter of the pulse is better than 1 microsecond, and simultaneously sends it to each sensor through a network of shielded cables of equal length to achieve synchronous data acquisition, and at each synchronous trigger moment, the integrated positioning and attitude determination system will output the corresponding position and attitude data packet, and the data acquisition system will stamp a unified timestamp on the data packet received by each sensor.

[0018] Compared with the prior art, the advantages of this invention are: 1. Multi-sensor networking combined with high-precision synchronous control enables synchronous acquisition of multi-dimensional data on the appearance, interior, and geometry of the track slab. Relying on a unified spatiotemporal reference to link data, it breaks the traditional "information silo" problem, enabling the tracing of internal defects and geometric deformations from surface defects, and significantly improving the accuracy of defect diagnosis. 2. An integrated positioning and attitude determination system combining GNSS, inertial measurement unit, and odometer can still accurately locate objects even in scenarios where GNSS signals fail, such as tunnels and elevated structures. This ensures that the location of the defect matches the actual track slab station number, eliminating the need for manual verification and improving maintenance efficiency. 3. Built-in intelligent data processing unit can automatically complete feature extraction and correlation analysis of multi-source data, replace manual experience interpretation, quickly generate track slab health reports and maintenance suggestions, shorten the data processing cycle, and avoid subjective bias to ensure the objectivity of results. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the overall device structure of the present invention; Figure 2 This is a schematic diagram of the bottom of the mobile support platform of the present invention; Figure 3 This is a schematic diagram of the connection relationship in this invention; Figure 4 This is a flowchart of the multi-sensor networking and data processing of the present invention; Figure 5 This is a schematic diagram of the multi-source information correlation analysis of the present invention; Figure 6 This is a diagram illustrating the execution steps of the detection method of the present invention.

[0020] In the diagram: 1. Mobile support platform; 2. Surface appearance acquisition module; 3. Internal state perception module; 4. Macroscopic geometric measurement module; 5. Central cabinet; 6. GNSS receiver antenna; 7. Inertial measurement unit and odometer module. Detailed Implementation

[0021] To make the purpose, principle and structure of the present invention clearer, the following description is provided in conjunction with the accompanying drawings and specific embodiments.

[0022] This invention provides a track slab detection device and method for multi-sensor network motion measurement.

[0023] In a first aspect, a track slab detection device for multi-sensor networked mobile measurement includes: a mobile carrier platform, a core control and synchronization unit, a multi-sensor network module, and an integrated data processing unit; the mobile carrier platform has wheels for autonomous or passive movement on the track, and its structure defines a front equipment area at the front of the direction of travel, a middle operating area, and a rear equipment area, wherein the middle operating area is equipped with a centralized cabinet with heat dissipation and shielding functions; the core control and synchronization unit includes a high-precision synchronization controller and an integrated positioning and attitude determination system; the integrated positioning and attitude determination system includes a GNSS receiver antenna mounted on the top of the mobile carrier platform, and an integrated inertial measurement unit and odometer module located near the center of gravity of the platform, for generating a high-precision spatiotemporal reference; the high-precision synchronization controller is located in the centralized cabinet, and after receiving the spatiotemporal reference from the integrated positioning and attitude determination system, it distributes a unified hardware synchronization trigger signal to all sensors in the multi-sensor network module; the multi-sensor network module includes: a surface appearance acquisition module, which includes at least one integrated area array high-definition camera and a line laser scanner, the module being rigidly connected The mounting brackets are positioned on the centerline of the bottom of the front equipment area of ​​the mobile platform, in front of the foremost wheels, ensuring that the laser stripes of the line laser scanner and the field of view of the area array high-definition camera completely cover the track path without interference from the wheels. The internal state perception module includes at least one array-type ground-penetrating radar. The transmitting and receiving antenna arrays of this radar are mounted at the bottom of the front equipment area of ​​the mobile platform, immediately behind the surface appearance acquisition module, to enable sequential detection of the same track slab area. The macroscopic geometric measurement module includes a 3D laser scanner, which is mounted on the top center of the rear equipment area of ​​the mobile platform via a column to obtain an unobstructed, wide field of view for acquiring panoramic 3D point clouds of the track slab and its ancillary facilities. The integrated data processing unit is located in a centralized cabinet and incorporates data fusion and intelligent analysis algorithms. It receives and processes synchronous data from the multi-sensor network module and, based on the high-precision position and attitude information provided by the integrated positioning and attitude system, unifies multi-source data into the same spatiotemporal coordinate system, achieving fusion analysis and integrated evaluation of the track slab's appearance, internal structure, and geometric information. Preferably, the high-precision synchronization controller adopts a hardware synchronization method, and the accuracy of the synchronization trigger signal is better than 1 microsecond; when the GNSS signal fails, the integrated positioning and attitude determination system, based on the combined navigation algorithm of the inertial measurement unit and the odometer module, can still maintain a relative positioning accuracy better than one-thousandth. Preferably, the line laser scanner and the area array high-definition camera in the surface appearance acquisition module are rigidly connected in space and jointly calibrated to form a structured light three-dimensional measurement subsystem, which can directly output three-dimensional deformation and texture information of the track slab surface with absolute coordinates.Preferably, the array-type ground-penetrating radar employs a multi-channel antenna array, the width of which is greater than the width of a single track slab. Transmission and acquisition are controlled by synchronous trigger signals to ensure precise correspondence between the radar scan profile and the absolute mileage position of the track slab. Preferably, the data fusion and intelligent analysis algorithms executed by the integrated data processing unit include: a deep learning-based target detection algorithm for automatically identifying surface defects such as cracks and spalling from images acquired by the area array high-definition camera; a point cloud processing algorithm for automatically extracting macroscopic geometric parameters such as flatness, superelevation, and gauge of the track slab from the point cloud of the 3D laser scanner; a radar image processing and inversion algorithm for identifying internal gaps, voids, and water-bearing anomalies within the track slab from the array-type ground-penetrating radar data; and a multi-source information association module for superimposing and correlating the identified surface defect locations, internal defect areas, and macroscopic geometric deformations in a unified spatiotemporal coordinate system to generate a comprehensive track slab health status assessment report.

[0024] Secondly, a track slab detection method using multi-sensor networked mobile measurement, based on the track slab detection device using multi-sensor networked mobile measurement described in the first aspect, includes the following steps: S1: System initialization and self-test: After starting the device, the core control and synchronization unit performs a self-test, and the integrated positioning and attitude determination system receives GNSS signals (GNSS refers to the Global Navigation Satellite System) and completes initialization; S2: Synchronous data acquisition: As the mobile carrier platform moves along the track, the high-precision synchronization controller distributes synchronization trigger signals to all sensors in the multi-sensor network module, controlling the surface appearance acquisition module, internal state perception module, and macroscopic geometric measurement module to perform synchronous data acquisition; S3: Data preprocessing and spatiotemporal alignment: The integrated data processing unit receives the acquired data from each module and utilizes the precise POS data (POS data refers to Position and Orientation System) provided by the integrated positioning and attitude determination system. S3: Unify all data into the geodetic coordinate system or the line design coordinate system; S4: Multi-source information fusion and intelligent identification: Perform parallel processing on the spatiotemporally aligned data, and perform surface defect identification, internal defect detection and macroscopic geometric parameter extraction respectively; S5: Comprehensive evaluation and decision support: Perform correlation analysis on the various defects and parameters identified in step S4, evaluate the comprehensive health status of a single track slab, and generate a status distribution map of the entire track slab and graded maintenance and repair suggestions based on the continuous detection results of all track slabs in the line. Preferably, in step S3, the spatiotemporal alignment specifically involves: using the position, attitude and velocity information provided by the integrated positioning and attitude determination system at each moment, through coordinate transformation, accurately assigning absolute geographic coordinates to the image center of the area array high-definition camera, the scanning line of the line laser scanner, the detection profile of the ground-penetrating radar and the single-frame point cloud of the 3D laser scanner. Preferably, in step S5, the correlation analysis specifically involves: when a surface crack is identified at a certain absolute coordinate position, simultaneously retrieving the three-dimensional contour data corresponding to that position to analyze whether there is deformation, and simultaneously retrieving the ground radar profile data directly below that position to analyze whether there is internal separation, thereby comprehensively judging the severity level and cause of the disease.

[0025] Example 1: Structural composition of a track slab detection device for multi-sensor networked motion measurement.

[0026] Please see Figures 1 to 6The present invention discloses a track slab detection device and method using multi-sensor network mobile measurement. The device integrates a multi-sensor network module through a mobile carrier platform, and achieves high-precision hardware synchronization and unified spatiotemporal reference for all sensors through a core control and synchronization unit. Finally, the integrated data processing unit performs fusion analysis and integrated evaluation of multi-source synchronized data, thereby achieving comprehensive, accurate and efficient detection of track slab appearance, internal and geometric information.

[0027] I. Device Structure and Installation.

[0028] The entity of this invention is a highly integrated track slab inspection system, the core of which is a dedicated inspection vehicle built on a mobile support platform 1. This platform is specifically designed and partitioned structurally to achieve optimal spatial layout of sensors and maximize system functionality.

[0029] 1. Platform partitioning and core unit installation.

[0030] The mobile platform 1 is physically divided into three functional areas, based on the coordination requirements of sensor functions and detection logic: Front Equipment Area: Located at the front of the vehicle, its bottom structure has been reinforced and leveled. The core function of this area is to house sensors that perform "near-field" and "contact" sensing of the track slab, ensuring that the original state of the track slab is accurately collected before the vehicle body rolls over the measured area.

[0031] Central Operation Area: Located in the middle of the vehicle, this is the system's "command center." This area houses a centrally located cabinet 5 with protective heat dissipation. This cabinet provides a stable and reliable operating environment for all precision electronic control and computing units.

[0032] Rear Equipment Area: Located at the rear of the vehicle, this area has a flat and open roof. Its core function is to provide an unobstructed macroscopic field of view for mounting sensors that perform "far-field" measurements of the track and surrounding environment.

[0033] The installation of the core control and synchronization unit is as follows: High-precision synchronous controller: As a dedicated timing and triggering board, it is installed in a dedicated slot in the central cabinet 5, closely adjacent to the integrated data processing unit, to minimize the transmission delay of control signals.

[0034] Integrated positioning and attitude determination system: GNSS receiver antenna 6 (GNSS stands for Global Navigation Satellite System): This high-performance dual-frequency measurement antenna is mounted on the mobile platform 1. Its position ensures that the antenna maintains maximum visibility into the sky and maximizes satellite signal reception even when the vehicle is moving or turning.

[0035] The coordinates of the antenna phase center in the vehicle coordinate system are expressed by the following formula: [x gnss ,y gnss ,z gnss ] T ; Where T represents the transpose symbol, which converts a row vector into a column vector, and is a standard for vector representation; x gnss This represents the coordinate value of the GNSS antenna phase center on the x-axis (direction of travel) of the vehicle coordinate system; y gnss This represents the coordinate value of the GNSS antenna phase center on the y-axis (to the right of the track) in the vehicle coordinate system; z gnss This indicates the coordinates of the GNSS antenna phase center on the z-axis (vertical to the sky) of the vehicle coordinate system, which have been precisely measured using a total station.

[0036] Inertial Measurement Unit and Odometer Module 7: A tactical-grade IMU (Inertial Measurement Unit) is employed. Its housing is rigidly fixed to the vehicle chassis structure via high-performance vibration isolators, and positioned as close as possible to the center of gravity of the mobile platform 1. The IMU's sensing axes (x-forward, y-right, z-upward) are strictly aligned with the coordinate system of the mobile platform 1. Odometer signals are derived from a high-precision photoelectric encoder in the vehicle's transmission system, with pulse signals directly input to the integrated processing circuitry within the IMU.

[0037] 2. Installation and spatial calibration of multi-sensor networking modules.

[0038] a) Installation and calibration of surface appearance acquisition module 2.

[0039] Mechanical Installation: The area array high-definition camera and line laser scanner are rigidly connected via a high-precision aluminum alloy clamp, forming a robust measurement unit. This unit is then mounted vertically downwards on the longitudinal centerline of the front equipment area base plate via a dedicated bracket equipped with polyurethane vibration dampers. The installation position has been precisely calculated to ensure that the module's detection end face is positioned in three-dimensional space in front of the projected contour of the vehicle's foremost wheel. The engineering purpose of this layout is to physically isolate the camera lens and laser from dust and water mist kicked up by the wheels.

[0040] Electrical and optical calibration: Camera intrinsic parameter calibration: Using a high-precision checkerboard calibration board, multiple images were acquired at different distances and angles. The camera's intrinsic parameter matrix k and lens distortion coefficient vector d were then calculated using the Zhang Zhengyou calibration method (Zhang Zhengyou calibration method is used for intrinsic parameter calibration of the area array high-definition camera in the surface appearance acquisition module: by acquiring multiple images of the high-precision checkerboard calibration board at different distances and angles, the camera intrinsic parameter matrix (including x / y focal length and principal point coordinates) and lens distortion coefficient vector are solved, providing the camera parameter basis for subsequent calculation of the three-dimensional coordinates of the track board surface from pixel coordinates).

[0041] in: ; Among them, f x ,f y c is the focal length in the x and y directions. x ,c y The coordinates of the main point.

[0042] Laser plane calibration: The calibration board is placed in front of the laser line and moved to multiple poses. Images of the laser line on the calibration board are captured by a camera. Using the calibrated camera parameters, the equation of the light plane formed by the laser beam in the camera coordinate system is fitted using the least squares method. π: aX c +bY c +cZ c +d=0; Where π represents the light plane, i.e., the plane formed by the laser beam projected by the line laser scanner; a, b, and c represent the unit normal vector components of the light plane in the camera coordinate system, determining the spatial orientation of the light plane; X c Y c Z c d represents the three-dimensional coordinates of any point in the camera coordinate system; d represents the distance from the origin of the camera coordinate system to the light plane.

[0043] At this point, the surface appearance acquisition module 2 is calibrated as a structured light 3D measurement subsystem capable of directly calculating the 3D coordinates of an object from pixel coordinates (u,v): P c =[X c ,Y c Z c ] T ; Where Pc represents the three-dimensional coordinate vector of the object in the camera coordinate system, which is the final result describing the spatial position of a point on the track surface; X c This represents the coordinates of that point on the object along the X-axis of the camera coordinate system (the X-axis corresponds to the horizontal direction of the camera); Y... c Z represents the coordinate value of that point on the object along the Y-axis in the camera coordinate system (the Y-axis corresponds to the vertical direction of the camera); cThis represents the coordinate value of that point on the object along the Z-axis in the camera coordinate system (the Z-axis corresponds to the optical axis of the camera lens, i.e., the direction perpendicular to the image plane); T represents the transpose symbol, which transposes the horizontally arranged [X... c ,Y c Z c Converting a row vector to a column vector is the mathematically standard form of vector representation, without changing the meaning of the coordinates themselves.

[0044] b) Installation and calibration of internal state sensing module 3.

[0045] Mechanical Installation: The transmitting and receiving antenna arrays of the array-type ground-penetrating radar are integrated into a low-profile protective housing made of fiberglass-reinforced plastic. The internal state perception module 3 is directly bolted to the rear of the surface appearance acquisition module 2 mounting bracket, ensuring that the radar detection profile and the line laser scanning line are spatially adjacent, enabling successive detection of the same track slab area.

[0046] System calibration: In the laboratory, the radar is placed on a concrete slab with a known thickness and dielectric constant for detection. By measuring the round-trip time of the electromagnetic wave, the average propagation velocity v of the wave in typical media (concrete, asphalt mortar) of the track slab is calibrated. This parameter is used to convert the two-way travel time t in the radar image into the depth axis d, with the conversion relationship: d = v × t / 2.

[0047] c) Installation and calibration of macroscopic geometric measurement module 4.

[0048] Mechanical Installation: A multi-beam 3D laser scanner is mounted on the center of the roof in the rear equipment area via a solid aluminum alloy column. This ensures that the scanner's laser beam can "cross" the roof and the driver's cab, providing unobstructed 360° panoramic coverage of the track area already traversed by the vehicle, the slopes on both sides, cable trenches, and contact wire poles. Although the platform itself may obscure a very small area directly below the scanner, the core task of this module is to acquire the continuous macroscopic geometry of the track, while the detailed surface information of the obscured area is fully acquired by the front surface appearance acquisition module 2, thus the two modules complement each other functionally.

[0049] System calibration: Using precision measuring instruments such as a total station, accurately determine the three-dimensional coordinates of the scanning center of the 3D laser scanner in the vehicle coordinate system and its installation deflection angle.

[0050] The three-dimensional coordinates are represented by the following formula: p platform scanner =[x s ,y s ,z s ] T ; Where p represents a "coordinate point," a core parameter describing the spatial location of the scanning center of the 3D laser scanner; the superscript "scanner" indicates that the object corresponding to this coordinate point is the scanning center of the "3D laser scanner," avoiding confusion with the coordinates of other components; the subscript "platform" indicates that the reference coordinate system for this coordinate is the "vehicle coordinate system" (i.e., the coordinate system of the mobile platform itself); X s : The coordinates of the scan center in the vehicle coordinate system along the x-axis (corresponding to the platform's direction of travel); Y s This represents the coordinate value of the scan center in the y-axis direction of the vehicle coordinate system (corresponding to the right side of the track perpendicular to the direction of travel); Z s This represents the coordinates of the scanning center in the z-axis direction of the vehicle coordinate system (corresponding to the upward direction perpendicular to the track plane); T represents the transpose symbol, which converts the horizontally arranged coordinate parameters into a vertical column vector. This is a standard vector representation and does not change the meaning of the coordinates.

[0051] The installation deflection angle is expressed by the following formula: ψ platform scanner =[φs,θs,ψs] T ; ψ represents the "attitude angle," specifically referring to the installation deflection angle of the 3D laser scanner, describing the scanner's attitude offset relative to the vehicle coordinate system. The superscript "scanner" indicates that the object corresponding to this attitude angle is a "3D laser scanner." The subscript "platform" indicates that the reference coordinate system for defining the attitude angle is the "vehicle coordinate system" (the coordinate system of the mobile platform itself). φs represents the scanner's "roll angle" deflection angle (rotation offset around the vehicle's x-axis) in the vehicle coordinate system. θs represents the scanner's "pitch angle" deflection angle (rotation offset around the vehicle's y-axis) in the vehicle coordinate system. ψs represents the scanner's "heading angle" deflection angle (rotation offset around the vehicle's z-axis) in the vehicle coordinate system. T represents the transpose symbol, converting the horizontally arranged angle parameters into a vertical column vector, which is a standard vector representation and does not change the meaning of the angle.

[0052] Using these parameters, the fixed rigid body transformation matrix from the scanner coordinate system to the vehicle coordinate system can be calculated: T platform scanner ; Where T represents the "transformation matrix," a mathematical tool for converting point coordinates between different coordinate systems; the superscript "scanner" defines the "source coordinate system" of the matrix as the "local coordinate system of the 3D laser scanner itself," i.e., the starting coordinate system of the matrix transformation; the subscript "platform" defines the "target coordinate system" of the matrix as the "coordinate system of the mobile carrier platform (vehicle)," i.e., the ending coordinate system of the matrix transformation; the overall matrix is ​​used to describe the fixed position and orientation relationship between the scanner's local coordinate system and the vehicle coordinate system. The parameters are calculated from the scanner's installation coordinates and deflection angle, and can transform the point cloud data collected by the scanner from its local coordinate system to the vehicle coordinate system.

[0053] This matrix serves as the basis for unifying all laser point cloud data into a global coordinate system.

[0054] 3. Deployment of integrated data processing units.

[0055] The integrated data processing unit, an industrial server equipped with a multi-core CPU and a high-performance GPU, is installed in the central operating area's centralized rack 5. It shares the rack's cooling duct and power bus with the high-precision synchronous controller. The server is pre-installed with a complete software system for data acquisition, real-time processing, intelligent analysis, and report generation, as well as a trained deep learning model, forming the "data brain" of the entire device.

[0056] II. Implementation of testing methods.

[0057] The detection method of the present invention is a highly coordinated, multi-stage data acquisition and intelligent processing flow, which relies entirely on the aforementioned precisely integrated device structure.

[0058] 1. System initialization and self-test.

[0059] Before the testing operation begins, the system performs a rigorous initialization and self-test process to establish a reliable benchmark for dynamic measurements.

[0060] 1) Platform power-on and unit self-test: Start the power supply system of the mobile carrier platform 1, and all units are powered on in sequence and execute the built-in self-test program.

[0061] Integrated positioning and attitude determination system: The GNSS receiver searches for and tracks satellite signals, the inertial measurement unit (IMU) warms up and performs initial alignment. The tightly combined Kalman filter inside the system starts, and its state vector X... k At time k, it is defined as: ; in: p=[x,y,z] T This represents the platform's three-dimensional position in the geodetic coordinate system. v=[v x,v y ,v z ] T Let v represent the three-dimensional velocity of the platform in the geodetic coordinate system; where v represents "velocity", referring to the three-dimensional velocity vector of the mobile platform. x "x" represents the x-axis direction in the geodetic coordinate system, indicating the platform's velocity component in the x-axis direction; v y "y" represents the y-axis direction in the geodetic coordinate system, indicating the platform's velocity component in the y-axis direction; v z "z" represents the z-axis direction in the geodetic coordinate system (usually the vertical direction towards the sky), indicating the platform's velocity component along the z-axis. "T" represents the transpose, converting the horizontally arranged velocity components into a vertical column vector; this is a standard vector representation and does not change the meaning of the velocity. ψ=[φ,θ,ψ] T The attitude angles of the platform (roll, pitch, yaw).

[0062] b g =[b gx ,b gy ,b gz ] T This represents the zero bias of the gyroscope across three axes; where b represents "zero bias," which is the inherent error of the device when it is static; the subscript g indicates that the object is a "gyroscope"; b gx This represents the zero bias value of the gyroscope on the x-axis (corresponding to the platform's travel direction); b gy This represents the zero bias value of the gyroscope on the y-axis (corresponding to the right side of the track); b gz This represents the zero bias value of the gyroscope on the z-axis (corresponding to the vertical sky direction); T represents the transpose symbol, which converts the horizontal parameter into a vertical column vector, which is a normalized vector representation.

[0063] b a =[b ax ,b ay ,b az ] T This represents the zero bias of the accelerometer across its three axes; where b represents "zero bias," i.e., the inherent error of the device when it is static; the subscript a indicates that the object is defined as an "accelerometer"; b ax This indicates the zero bias value of the accelerometer on the x-axis (corresponding to the platform's direction of travel); b ay This indicates the zero bias value of the accelerometer on the y-axis (corresponding to the right side of the track); b az This indicates the zero bias value of the accelerometer on the z-axis (corresponding to the vertical sky direction); T represents the transpose symbol, which converts the horizontal parameter into a vertical column vector, which is a normalized vector representation.

[0064] δscale represents the odometer scale error.

[0065] High-precision synchronization controller: generates test synchronization pulses to verify the hardware triggering link with all sensors.

[0066] Multi-sensor networking module: Area array high-definition camera reports focal length and aperture status; line laser scanner provides feedback on laser power; 3D laser scanner starts scanning motor; array-type ground-penetrating radar performs internal circuit self-diagnosis.

[0067] Integrated data processing unit: Starts the server and loads the data acquisition software, track slab design model, and pre-trained deep learning model.

[0068] 2) Spatiotemporal reference establishment and filter convergence: The integrated positioning and attitude determination system enters dynamic operating mode. The filter is recursively estimated through the state equation (based on the discretization of the inertial navigation mechanics equation) and the observation equation (GNSS pseudorange, carrier phase, and odometer increment). After a brief convergence process, the system stably outputs a smooth, continuous, high-precision POS data stream at a frequency of 200Hz, which serves as the unified spatiotemporal reference for the entire system.

[0069] The formula for representing POS data stream is: {p(t),v(t),ψ(t)}; Wherein POS stands for "Position and Orientation System," referring to the dynamic data sequence output by this system; p(t): "p" represents "Position," and "t" represents "time," collectively referring to the three-dimensional position data (x, y, z coordinates in the geodetic coordinate system or the route design coordinate system) of the mobile platform at time t; v(t): "v" represents "Velocity," and "t" represents "time," collectively referring to the three-dimensional velocity data of the mobile platform at time t (the velocity components in the x, y, and z directions in the corresponding position coordinate system); ψ(t): "ψ" represents "Attitude," and "t" represents "time," collectively referring to the three-dimensional attitude data (roll angle φ, pitch angle θ, and heading angle ψ) of the mobile platform at time t.

[0070] 2. Synchronous data acquisition.

[0071] The mobile carrier platform 1 travels along the track at a constant speed (e.g., 80 km / h) and enters the core data acquisition phase.

[0072] High-precision synchronous triggering: The high-precision synchronous controller uses its internal high-stability crystal oscillator as the frequency reference, receives the 1PPS signal (1PPS signal refers to a one-second pulse signal) from the integrated positioning and attitude determination system for synchronization correction, and generates a hardware trigger pulse at a fixed frequency of 100Hz. The rise time jitter of this pulse is better than 1 microsecond, and it is delivered simultaneously through a network of shielded cables of equal length.

[0073] Surface appearance acquisition module 2: The trigger line laser scanner projects a laser line and precisely synchronously triggers the area array high-definition camera for exposure, capturing an image I(t) containing laser stripes and natural textures. i ).

[0074] Internal state perception module 3: triggers the array-type ground-penetrating radar to emit electromagnetic pulses and simultaneously records the reflected echoes, forming an AScan data vector.

[0075] Macroscopic geometric measurement module 4: Triggers the 3D laser scanner to record a single frame of point cloud at the current moment: {P scanner local (t i )}; Where P represents "Point," referring here to the coordinates of a single spatial point acquired by a 3D laser scanner, which together constitute a point cloud; the superscript "local" indicates that the coordinate reference is defined as a "local coordinate system," that is, the local coordinate system of the 3D laser scanner itself; the subscript "scanner" indicates that the acquisition device for this point cloud is a "3D laser scanner"; t i The parameter indicates that "t" represents "time" and "i" represents the i-th synchronization trigger moment, meaning that the single frame of point cloud was acquired at the i-th synchronization moment; the overall parameter means "a frame of point cloud data acquired by the 3D laser scanner at the i-th synchronization moment in its own local coordinate system".

[0076] Data is bound to a spatiotemporal reference: at each synchronization trigger time t i Integrated positioning and attitude determination systems will output a corresponding POS data packet: {p(t i ), ψ(t) i )}: Where p represents "Position", corresponding to the platform's three-dimensional position data (x, y, z coordinates) in the geodetic coordinate system in the data packet, describing the platform's spatial position; v represents "Velocity", corresponding to the platform's three-dimensional velocity data (x, y, z direction velocity components) in the geodetic coordinate system in the data packet, describing the platform's speed; ψ represents "Attitude", corresponding to the platform's three-dimensional attitude data (roll angle φ, pitch angle θ, yaw angle ψ) in the data packet, describing the platform's spatial attitude; t represents "Time", corresponding to the timestamp of each position, velocity, and attitude information in the data packet, used to achieve data time synchronization.

[0077] The data acquisition system adds this unique timestamp t to every data packet (image, radar AScan, laser point cloud frame) received from the sensor. i This mechanism fundamentally ensures that multi-source heterogeneous data are precisely aligned to the same point in time.

[0078] 3. Data preprocessing and spatiotemporal alignment.

[0079] The integrated data processing unit receives and caches all the raw data with timestamps, and then performs key preprocessing and coordinate unification to map all the data to the same spatial reference.

[0080] Surface 3D Reconstruction and Coordinate Transformation: Image preprocessing and laser line extraction: For image I(t) i Noise reduction was performed, and lens distortion correction was performed using the calibrated distortion coefficient d. Subsequently, subpixel algorithms such as Steger were used to accurately extract the center line of the laser stripes from the image, obtaining a series of pixel coordinates (u, v).

[0081] 3D coordinate calculation (camera coordinate system): For each laser point pixel (u, v), its distortion-corrected normalized coordinates (X, V) are calculated using the calibrated camera intrinsic parameter matrix K. n Y n ).

[0082] ; Where: f x and f y c represents the equivalent focal length of the camera in the x and y directions (in pixels); x and c y The coordinates of the camera principal point in the image coordinate system.

[0083] Combined with the pre-calibrated laser plane equation: π: aX c +bY c +cZc +d=0; Where [a, b, c] T Let be the unit normal vector of the laser plane in the camera coordinate system, and d be the distance from the origin to the plane. Thus, the surface appearance acquisition module 2 is calibrated as a module capable of directly calculating the object's 3D coordinates P from the pixel coordinates (u, v). c =[X c ,Y c Z c ] T The structured light 3D measurement subsystem, wherein: Z c =(-d) / (a×x) n +b×y n +c)),X c =(x n ×Z c ),Y c =y n ×Z c； Among them, X c The "X-axis component" (horizontal direction, such as left and right) represents the three-dimensional coordinates; the Y-axis component represents the horizontal component. c The Y-axis component (vertical direction, such as up and down) represents the three-dimensional coordinate system; Z... c The "Z-axis component" (depth direction, such as distance to the camera / sensor) represents the 3D coordinates. n The "X-axis component" represents the horizontal direction of a two-dimensional plane coordinate system (e.g., pixel column number related values); y n The "Y-axis component" represents the two-dimensional plane coordinates (vertical direction of the image, such as pixel row number related values). a, b, and c are coefficients of the plane projection equation (such as describing the projection relationship between the image plane and the spatial plane); d is a scaling factor or offset factor (used to adjust the calculation scale of depth Zc).

[0084] Coordinate transformation to global coordinate system: transform point P in the camera coordinate system c Through a fixed installation matrix T platform laser and t i Dynamic POS transformation matrix T at time t global platform (t) i Transform to global coordinate system P global surface : P global surface =T global platform (t i )×T global platform ×Pc ; Among them, T global platform (t i ) is composed of p(t) i ) and ψ(t i The 4x4 homogeneous transformation matrix is ​​formed by .

[0085] Radar data positioning: Each detection profile of an array-type ground-penetrating radar is located through its mounting matrix T. platform radar and dynamic POS transformation matrix T global platform (t i Position its center point to the global coordinate system P. global radar : P global radar =T global platform (t i )×T global radar c ; Laser point cloud localization: This involves locating a single frame of point cloud from a 3D laser scanner (P...) scanner local (t i )}, by installing calibration matrix T platform scanner And dynamic POS transformation, uniformly transformed to global coordinate system P global pointcloud : P global pointcloud =T global platform (t i )×T platform scanner ×P scanner loacal (t i ); At this point, all sensor data have achieved spatiotemporal alignment under a unified high-precision global coordinate system, laying a solid foundation for subsequent multi-source information fusion.

[0086] 4. Multi-source information fusion and intelligent recognition.

[0087] Within a unified spatiotemporal framework, data from various dimensions are processed and features are extracted in parallel.

[0088] Intelligent identification of surface defects: The preprocessed high-resolution image I(t) iThe input is fed into a pre-trained deep learning semantic segmentation network (such as UNet). This network outputs a segmentation map S(t). i Each pixel is categorized (e.g., "crack", "peeling", "background").

[0089] Information overlay: The segmentation result S(t) is overlaid. i ) and the refined 3D point cloud {P} that has been transformed to the global coordinate system global pointcloud (t i The data is overlaid. For each pixel identified as a defect, the system can immediately obtain its corresponding absolute geographic coordinates P. defect And three-dimensional geometric information (such as length, width, and depth).

[0090] Automatic extraction of macroscopic geometric parameters: Data processing is performed on the panoramic 3D point cloud generated by the macroscopic geometric measurement module 4.

[0091] Track gauge and superelevation calculation: Using region growing and the RANSAC algorithm, the rail surface points of the left and right rails are accurately separated from the panoramic 3D point cloud generated by the macroscopic geometric measurement module 4 and fitted to a plane. The horizontal distance between the centerlines of the two rail surfaces is calculated to obtain the track gauge G, and the elevation difference between the two rail surfaces is calculated to obtain the superelevation H.

[0092] Track slab flatness assessment: Point cloud of a single track slab surface {P i Perform plane fitting and calculate the value of each point {P}. i The distance d to the fitted plane i Then, the standard deviation σ of all distances is calculated as an evaluation index for flatness: ; Where, d i Let be the distance from the i-th point to the fitted plane; For all d i The mean of the points; N is the number of points in the point cloud.

[0093] Internal defect detection: The raw echo signal from the ground-penetrating radar is preprocessed, including gain compensation and background denoising, to form a radar image (BScan).

[0094] The Kirchhoff migration algorithm is used to process radar images, converging the reflected signals to their true spatial locations. Using the calibrated wave velocity v, the two-way travel time t is converted to depth d: d = (v × t) / 2, thereby identifying anomalous areas such as gaps and voids inside the track slab.

[0095] 5. Comprehensive assessment and decision support This is a key step in transforming data into knowledge, aiming to generate decision support reports that can directly guide maintenance and repair by associating information from multiple sources.

[0096] Multi-source information correlation analysis: The system uses a single track slab as the basic evaluation unit and initiates the multi-source information correlation module. When at absolute coordinate P... crack When a surface crack is identified, the module automatically performs a spatial query: In P crack Query nearby detailed 3D point clouds to analyze whether there are local depressions or deformations.

[0097] In P crack Directly below, query the ground-penetrating radar profile data to analyze whether there is a strong reflective interface (internal fracture) at the corresponding depth.

[0098] Comprehensive Status Assessment and Report Generation: Based on the above correlation analysis results, a comprehensive diagnosis is performed on each track slab.

[0099] Scenario A: Only surface cracks exist; no internal defects or geometric deformation. Conclusion: Surface damage; low maintenance priority.

[0100] Scenario B: Surface cracks and internal gaps appear simultaneously, but the geometric deformation is within acceptable limits. Conclusion: In the early stages of structural damage, planned maintenance is required.

[0101] Scenario C: Surface cracks, internal gaps, and localized geometric deformations (such as settlement) occur simultaneously. Conclusion: Severe structural damage; high maintenance priority.

[0102] Ultimately, the system automatically generates a status distribution map of the entire track slab, with different colors indicating the health level, and includes a detailed list of test data and a graded maintenance recommendation report, achieving seamless integration of testing and maintenance decisions.

[0103] Example 2: How to use a track slab detection device for multi-sensor networked motion measurement.

[0104] This embodiment provides a method for using the track slab detection device based on the multi-sensor network motion measurement described in Embodiment 1. The specific implementation method is as follows: Step S1: The track slab inspection vehicle, equipped with a multi-sensor networked mobile measurement system, is driven to the starting point of the line to be inspected.

[0105] Step S2: Perform initial positioning calibration using an integrated positioning and attitude determination system.

[0106] The GNSS receiver antenna receives satellite signals, the inertial measurement unit synchronously collects platform attitude data, and the odometer module records the initial position reference.

[0107] Step S3: Start the core control and synchronization unit to trigger the multi-sensor networking module to enter the working state.

[0108] The high-precision synchronous controller distributes synchronous trigger signals (with an accuracy better than 1 microsecond) to the surface appearance acquisition module, the internal state perception module, and the macroscopic geometric measurement module.

[0109] Step S4: The surface appearance acquisition module completes the field of view alignment and scanning preparation.

[0110] The field of view of the area array high-definition camera covers the surface of the track slab, and the line laser scanner emits laser stripes and simultaneously collects reflected signals. The two are kept in a fixed spatial relative position by a rigid bracket.

[0111] In step S5, the mobile support platform moves at a constant speed along the track, and each sensor synchronously performs data acquisition.

[0112] The array-type ground-penetrating radar's antenna array scans closely against the bottom of the track slab, while the three-dimensional laser scanner continuously collects panoramic cloud data from an unobstructed rear view. All data is transmitted in real time to the integrated data processing unit.

[0113] Step S6: The integrated data processing unit performs spatiotemporal alignment and fusion analysis on the collected data.

[0114] Based on the POS data provided by the integrated positioning and attitude determination system, image texture, laser point cloud, and radar profile are unified into the line design coordinate system. Deep learning algorithm automatically identifies surface cracks, and radar inversion algorithm locates the internal void area.

[0115] Step S7: After the inspection vehicle reaches the end of the route, the system generates a comprehensive evaluation report.

[0116] The multi-source information association module associates and superimposes the surface defects, internal flaws and geometric position data of a single track slab, outputs the health status level and maintenance suggestions, and then the inspection vehicle leaves the inspection area to prepare for the next section inspection operation.

[0117] The above description is merely a specific embodiment of the invention, but the scope of protection of the invention is not limited thereto. Any equivalent substitutions or changes made by those skilled in the art within the technical scope disclosed in the invention, based on the technical solution and concept of the invention, should be covered within the scope of protection of the invention.

Claims

1. A track slab detection device for multi-sensor networked motion measurement, characterized in that, include: A mobile carrying platform is equipped with wheel sets for autonomous or passive movement on a track and is divided into a front equipment area, a middle operating area and a rear equipment area. The middle operating area is equipped with a centralized cabinet with heat dissipation and shielding functions. A core control and synchronization unit includes: an integrated positioning and attitude determination system, comprising a global navigation satellite system receiver antenna and an integrated inertial measurement unit and odometer module, used to generate a high-precision spatiotemporal reference; and a high-precision synchronization controller, located in the central cabinet, which receives the spatiotemporal reference of the integrated positioning and attitude determination system and distributes a unified hardware synchronization trigger signal to the multi-sensor networking module. A multi-sensor networking module includes: a surface appearance acquisition module, located at the bottom of the front equipment area, including at least one integrated area array high-definition camera and line laser scanner; an internal state perception module, located behind the surface appearance acquisition module, including at least one array-type ground-penetrating radar; and a macroscopic geometric measurement module, located at the top of the rear equipment area, including a three-dimensional laser scanner. An integrated data processing unit, located in the central cabinet, has built-in data fusion and intelligent analysis algorithms. It is used to receive and process the synchronous data of the multi-sensor networking module, and to unify the multi-source data into the same spatiotemporal coordinate system based on the position and attitude information of the integrated positioning and attitude system, so as to realize the fusion analysis and integrated evaluation of the appearance, internal and geometric information of the track slab.

2. The track slab detection device for multi-sensor networked motion measurement as described in claim 1, characterized in that, The high-precision synchronization controller is located in the central cabinet and receives the spatiotemporal reference of the integrated positioning and attitude determination system. It distributes a unified hardware synchronization trigger signal to all sensors in the multi-sensor networking module. The high-precision synchronization controller adopts a hardware synchronization method, and the accuracy of its synchronization trigger signal is higher than 1 microsecond. The integrated positioning and attitude determination system includes a global navigation satellite system receiver antenna mounted on the top of the mobile carrier platform, and an inertial measurement unit and odometer module integrated near the platform's center of gravity. It is used to generate a high-precision spatiotemporal reference. When the global navigation satellite system signal fails, the integrated positioning and attitude determination system maintains a relative positioning accuracy of more than one-thousandth through a combined navigation algorithm of the inertial measurement unit and the odometer module.

3. The track slab detection device for multi-sensor networked motion measurement as described in claim 1, characterized in that, The surface appearance acquisition module is mounted on the center line of the bottom of the front equipment area of ​​the mobile carrier platform through a rigidly connected mounting bracket, and is located in front of the wheel at the foremost end of the platform, so that the laser stripes of the line laser scanner and the field of view of the area array high-definition camera can completely cover the track path without being disturbed by the wheel. The line laser scanner and the area array high-definition camera are rigidly connected in space and jointly calibrated to form a structured light three-dimensional measurement subsystem, which is used to directly output three-dimensional deformation and texture information of the track slab surface with absolute coordinates.

4. The track slab detection device for multi-sensor networked motion measurement as described in claim 1, characterized in that, The radar's transmitting and receiving antenna array in the internal state perception module is installed at the bottom of the front equipment area of ​​the mobile carrier platform and is arranged immediately behind the surface appearance acquisition module to achieve successive detection of the same track plate area. The array-type ground-penetrating radar uses a multi-channel antenna array, the width of which is greater than the width of a single track slab. Its transmission and acquisition are controlled by the synchronous trigger signal to ensure that the radar scanning profile corresponds precisely to the absolute mileage position of the track slab.

5. The track slab detection device for multi-sensor networked motion measurement as described in claim 1, characterized in that, The 3D laser scanner is mounted on the top center of the equipment area at the rear of the mobile support platform via a column to obtain an unobstructed wide field of view for acquiring panoramic 3D point clouds of the track slab and ancillary facilities. The data fusion and intelligent analysis algorithms executed by the integrated data processing unit include: A deep learning-based target detection algorithm is used to automatically identify surface defects such as cracks and spalling from images acquired by the area array high-definition camera; a point cloud processing algorithm is used to automatically extract macroscopic geometric parameters such as flatness, superelevation, and gauge of the track slab from the point cloud of the 3D laser scanner; a radar image processing and inversion algorithm is used to identify internal defects such as gaps, voids, and water-bearing anomalies in the track slab from the data of the array-type ground-penetrating radar; and a multi-source information association module is used to overlay and correlate the identified surface defect locations, internal defect areas, and macroscopic geometric deformations in a unified spatiotemporal coordinate system to generate a comprehensive track slab health status assessment report.

6. A method for detecting track slabs using multi-sensor network motion measurement, comprising using the track slab detection device for multi-sensor network motion measurement as described in any one of claims 1-5, characterized in that, Includes the following steps: S1: System initialization and self-test, start-up device, core control and synchronization unit perform self-test, integrated positioning and attitude determination system receives global navigation satellite system signals and initializes; S2: Synchronous data acquisition. The mobile carrier platform moves along the track. The high-precision synchronous controller distributes synchronous trigger signals to all sensors in the multi-sensor networking module to control the surface appearance acquisition module, internal state perception module and macroscopic geometric measurement module to perform synchronous data acquisition. S3: Data preprocessing and spatiotemporal alignment. The integrated data processing unit receives the collected data from each module and uses the precise position and attitude data provided by the integrated positioning and attitude system to unify all data into the geodetic coordinate system or the route design coordinate system. S4: Multi-source information fusion and intelligent recognition, which performs parallel processing on spatiotemporally aligned data, and performs surface disease identification, internal defect detection and macroscopic geometric parameter extraction respectively; S5: Comprehensive assessment and decision support. This involves correlation analysis of the various defects and parameters identified in step S4, assessment of the overall health status of a single track slab, and generation of a status distribution map of all track slabs along the line and graded maintenance recommendations based on continuous inspection results of all track slabs in the line.

7. The track slab detection method for multi-sensor networked motion measurement as described in claim 6, characterized in that, The spatiotemporal alignment mentioned in step S3 specifically includes: By utilizing the position, attitude, and velocity information provided by the integrated positioning and attitude determination system at every moment, and through coordinate transformation, the image center of the area array high-definition camera, the scanning line of the line laser scanner, the detection profile of the ground-penetrating radar, and the single-frame point cloud of the 3D laser scanner are accurately assigned absolute geographic coordinates.

8. The track slab detection method for multi-sensor networked motion measurement as described in claim 6, characterized in that, The association analysis described in step S5 specifically includes: When a surface crack is identified at a certain absolute coordinate position, the corresponding three-dimensional contour data is retrieved simultaneously to analyze whether there is deformation, and the ground-penetrating radar profile data directly below the position is retrieved to analyze whether there is internal separation. In this way, the severity and cause of the disease can be comprehensively judged.

9. The track slab detection method for multi-sensor networked motion measurement as described in claim 6, characterized in that, The system initialization and self-test in step S1 also include: In the integrated positioning and attitude determination system, the global navigation satellite system receiver searches for and tracks satellite signals, the inertial measurement unit warms up and performs initial alignment, the tightly combined Kalman filter inside the system is activated, the high-precision synchronization controller generates test synchronization pulses, and the hardware trigger links with all sensors are verified. In the multi-sensor networking module, each sensor reports its own status, the integrated data processing unit starts the server, and loads the data acquisition software, the track slab design model, and the pre-trained deep learning model.

10. The track slab detection method for multi-sensor networked motion measurement as described in claim 6, characterized in that, In step S2, the mobile carrier platform travels along the track at a constant speed. The high-precision synchronous controller uses its internal high-stability crystal oscillator as the frequency reference and receives the 1-second pulse signal from the integrated positioning and attitude determination system for synchronous correction. It generates a hardware trigger pulse at a fixed frequency of 100Hz. The rise time jitter of this pulse is better than 1 microsecond. The pulse is simultaneously sent to each sensor through a network of shielded cables of equal length to achieve synchronous data acquisition. At each synchronous trigger moment, the integrated positioning and attitude determination system outputs the corresponding position and attitude data packet. The data acquisition system adds a unified timestamp to the data packet received by each sensor.

Images