A locomotive external track geometry shape detection device and detection method

Abstract

The present application belongs to the technical field of traffic track detection, and particularly relates to a locomotive external track geometric position detection device and a detection method. The method comprises the following steps: real-time acquisition of the lateral acceleration and vertical acceleration of the vehicle and comparison with the corresponding preset threshold values, respectively, when any acceleration exceeds the threshold value, determination of excessive vibration interference; when the vibration interference is excessive, acquisition of the geometric appearance of the track, the distance information of the track relative to the vehicle, the position information of the vehicle and the motion state of the vehicle; based on the geometric appearance of the track, the distance information of the track relative to the vehicle, the position information of the vehicle and the motion state of the vehicle, real-time track gauge calculation, track horizontal calculation, track height calculation and triangular pit calculation of the vehicle are carried out to obtain the dynamic geometric position parameters of the track. The method realizes on-demand detection, avoids data redundancy, improves detection pertinence and resource utilization, and is suitable for accurate identification and normal operation and maintenance of track disease sections.

Description

Technical Field

[0001] This invention belongs to the field of transportation track inspection technology, specifically relating to a locomotive external track geometry and position detection device and method. Background Technology

[0002] With the continuous expansion of rail transit networks and the increasing demands for operational safety and maintenance efficiency, dynamic monitoring of track geometry parameters (such as gauge, level, elevation, and triangular grooves) is a core element in ensuring safe railway operation. Traditional inspection methods mainly rely on dedicated inspection vehicles. While these vehicles offer high accuracy, they suffer from high operating costs, low inspection frequency, and poor deployment flexibility, making it difficult to meet the routine inspection needs of large-scale rail networks. With technological advancements, external dynamic inspection devices, which combine low-cost deployment, all-encompassing environmental adaptability, high-precision inspection, and rapid response capabilities, have become the mainstream tool in the field of rail operation and maintenance technology.

[0003] Existing external detection devices mostly employ a fusion scheme of vision and inertial measurement units (IMUs), attempting to achieve dynamic detection by attaching them to ordinary locomotives. However, these devices often suffer from excessive detection deviations in practical applications, failing to meet the accuracy requirements of millimeter-level detection. The most significant accuracy issue stems from the interference of mechanical vibrations: due to terrain changes or natural disasters, track irregularities or defective sections cause multi-frequency vibrations during locomotive operation, leading to image blurring and IMU data drift. This results in the acquired data containing significant noise or irregular variations, leading to significant errors in parameter calculation. Insufficient detection accuracy results in excessive deviations in track geometry parameters, potentially causing safety accidents such as train derailment and accelerated component wear. Summary of the Invention

[0004] To address the problems existing in the prior art, the purpose of this invention is to provide a locomotive-mounted track geometry and position detection device and method, enabling high-precision, high-accuracy, and highly adaptable dynamic monitoring of geometry and position under non-dedicated inspection vehicle conditions.

[0005] The technical solution of this invention is:

[0006] A method for detecting the geometric shape and position of an external track on a locomotive includes the following steps:

[0007] Real-time acquisition of the vehicle's lateral and vertical acceleration;

[0008] The lateral acceleration and vertical acceleration are compared with their respective preset thresholds. When either acceleration exceeds the preset threshold, the vibration interference is determined to be excessive.

[0009] When vibration interference exceeds the standard, the geometric appearance of the track, the distance information of the track relative to the vehicle, the position information of the vehicle, and the motion state of the vehicle are obtained respectively; wherein, the motion state of the vehicle includes the roll angle, pitch angle, and yaw angle of the vehicle.

[0010] Based on the track's geometric appearance, the track's distance from the vehicle, the vehicle's position information, and the vehicle's roll, pitch, and yaw angles, the system performs real-time track gauge calculation, track level calculation, track elevation calculation, and triangular pit calculation to obtain the track's dynamic geometric position parameters.

[0011] Preferably, the track gauge is determined according to the following formula:

[0012] ,

[0013] In the formula: For calibration of track gauge; This is the original track gauge; The vehicle's roll angle;

[0014] Among them, the original track gauge Determined according to the following formula:

[0015] ,

[0016] In the formula: u and v are the image coordinates of feature points at the end face of the track; The rotation matrix represents the vehicle's attitude. The vehicle's pitch angle; The vehicle's yaw angle; , , X, Y, Z are the world coordinates of the camera; X, Y, Z are the world coordinates of feature points in the geometric appearance image of the train; K is the camera intrinsic parameter matrix; K 33 is a constant, representing the element in the third row and third column of the camera intrinsic parameter matrix; and These are the coordinates of the camera's principal point.

[0017] Preferably, the orbital level calculation is determined according to the following formula:

[0018] ,

[0019] in, , ,

[0020] In the formula, S is the original horizontal value of the track; This is the vertical distance from the top of the left rail; This is the vertical distance from the top of the right rail; This is the absolute elevation of the distance measuring point; The vehicle's pitch angle; This is the absolute elevation of the top of the left rail; This is the absolute elevation of the top of the right rail.

[0021] Preferably, the orbital elevation calculation is determined according to the following formula:

[0022] ,

[0023] in, ,

[0024] , , ,

[0025] In the formula, Orbital elevation and elevation values; X represents the rail top elevation of the data collection point; X represents the mileage coordinates; L represents the length of the selected track segment; X i H represents the mileage coordinates of the i-th data collection point; i Let be the top elevation of the track at the i-th acquisition point, which is the absolute vertical height of the top of the track at the i-th acquisition point. This is the sampling interval distance; The reference elevation line is obtained by fitting the track segment data using the least squares method; A is the slope of the reference elevation line, which reflects the average trend of the elevation change of the track segment; B is the intercept parameter.

[0026] Preferably, the triangular pit solution is determined according to the following formula:

[0027] ,

[0028] in, ,

[0029] In the formula, T is the triangular pit value; X i and X j Both S represent the mileage of the vehicle at corresponding positions i and j on the track; i For X i The track level value corresponding to the mileage; S j For X j The track level value corresponding to the mileage.

[0030] Preferably, a locomotive external track geometry and position detection device is used to implement any of the detection methods described above, comprising:

[0031] The first acquisition unit is used to acquire the vehicle's lateral acceleration and vertical acceleration in real time;

[0032] The first processing unit is used to compare the lateral acceleration and vertical acceleration with preset thresholds respectively;

[0033] The second acquisition unit is used to acquire the geometric appearance of the track, the distance information of the track relative to the vehicle, the position information of the vehicle, and the motion state of the vehicle when the lateral acceleration or vertical acceleration is greater than a preset threshold.

[0034] The second processing unit is used to perform parameter calculations based on vehicle acceleration, track geometry, track distance relative to vehicle, vehicle position, and vehicle motion state to obtain dynamic geometric position parameters of the track.

[0035] Preferably, the first acquisition unit includes a triaxial accelerometer and a mounting plate. The mounting plate is detachably connected to the vehicle's chassis and is located in the middle of the vehicle's chassis. The triaxial accelerometer is fixed to the mounting plate and is used to simultaneously detect the lateral acceleration and vertical acceleration of the vehicle.

[0036] Preferably, the second acquisition unit includes a track shape acquisition module, a laser ranging module, a vehicle positioning module, and an inertial reference module. The track shape acquisition module is used to acquire the geometric appearance of the track, the laser ranging module is used to acquire the distance information of the track relative to the vehicle, the vehicle positioning module is used to acquire the position information of the vehicle, and the inertial reference module is used to acquire the motion state of the vehicle.

[0037] Preferably, the track shape acquisition module includes a laser positioning device and at least four high-speed industrial cameras. The high-speed industrial cameras are respectively mounted on both sides of the rear vehicle in the horizontal direction via a crossbeam. There are at least two laser positioning devices, which are respectively installed at both ends of the crossbeam for calibrating the lenses of the high-speed industrial cameras.

[0038] Compared with the prior art, the locomotive external track geometry detection device and detection method of the present invention have the following advantages:

[0039] This invention monitors the lateral and vertical acceleration of vehicles in real time and intelligently compares it with preset thresholds. When vibration interference exceeds the limit, a high-precision data acquisition process is automatically triggered, enabling on-demand detection and effectively avoiding data redundancy and resource waste caused by continuous sampling across the entire track. Furthermore, during parameter calculation, vehicle motion state information such as roll angle, pitch angle, and yaw angle are incorporated to dynamically compensate for the collected track geometry and distance data. This effectively suppresses image distortion, data drift, and noise interference caused by mechanical vibration, significantly improving the calculation accuracy of key parameters such as track gauge, level, elevation, and triangular pits to the millimeter level. This solves the problem of excessive detection deviation caused by vibration interference, improves measurement stability and accuracy in complex environments, enhances the reliability and safety of track condition monitoring, and reduces maintenance costs. Attached Figure Description

[0040] Figure 1 This is a schematic diagram showing the overall position of the detection device in an embodiment of the present invention.

[0041] Figure 2 This is a schematic diagram showing the installation position of the track state adaptive online evaluation component in an embodiment of the present invention.

[0042] Figure 3 This is a schematic diagram of the structure of the adaptive online evaluation component for track status in an embodiment of the present invention.

[0043] Figure 4 This is a schematic diagram of the track geometry information acquisition unit in an embodiment of the present invention.

[0044] Figure 5 This is a cross-sectional view of the beam device structure in an embodiment of the present invention.

[0045] Explanation of reference numerals in the attached figures:

[0046] 1. Track State Adaptive Online Evaluation Component: 1a. Vehicle Dynamic Response Acquisition Module: 1a1. Triaxial Accelerometer: 1a2. Mounting Plate: 1b. State Analysis and Evaluation Module: 1b1. Calculation Chip: 2. Track Geometry Information Precision Acquisition Component: 2a. Mounting Platform: 2a1. Crossbeam: 2a2. Quick-Release Locking Structure: 2a3. Positioning Pin: 2b. Track Shape Acquisition Module: 2b1. High-Speed ​​Industrial Camera: 2b2. Laser Positioner: 2c. Laser Ranging Module: 3. Dynamic Spatial Reference Stabilization Component: 3a. Vehicle Positioning Module: 3a1. GNSS Receiver: 3a2. Communication Module: 3a3. CAN Bus: 3b. Inertial Reference Module: 3b1. IMU Unit: 4. Track Geometry and Position Parameter Solving Module: 4a. Master-Slave Chip Module: 4a1. Main Processing Chip: 4a2. Coprocessing Chip: 4b. Algorithm Processing Module: 5. Wireless Data Transmission and Data Storage Unit: 5a. Wireless Data Transmission Module: 5b. Data Storage Module: Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0048] Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this invention.

[0049] Furthermore, the technical solutions of the various embodiments of the present invention can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

[0050] Traditional track geometry and position detection relies heavily on dedicated inspection vehicles. These vehicles are costly to purchase and maintain, and their operation is constrained by schedules, making it difficult to achieve high-frequency, comprehensive, and routine monitoring. While existing external inspection devices can be deployed on ordinary locomotives, they generally suffer from fixed detection strategies and an inability to dynamically adjust to actual track conditions. This results in the continuous collection of large amounts of redundant data in sections with good track conditions, while in deteriorated sections, insufficient sampling may lead to missed detection of critical defects. Furthermore, vibration interference, changes in lighting, and spatial reference drift under complex operating environments severely impact detection accuracy and stability. These issues collectively hinder the realization of efficient and reliable dynamic monitoring of track geometry.

[0051] The present invention proposes the following:

[0052] A locomotive-mounted track geometry and position detection device includes a track state adaptive online evaluation component 1, a second acquisition unit, and a second processing unit. The track state adaptive online evaluation component 1 includes a first acquisition unit and a first processing unit. The second acquisition unit includes a track shape acquisition module 2b, a laser ranging module 2c, a vehicle positioning module 3a, and an inertial reference module 3b. Through the acquisition and fusion of multi-source data, it acquires the track geometry, track distance relative to the vehicle, vehicle position information, and vehicle motion state when the lateral or vertical acceleration exceeds a preset threshold.

[0053] The first acquisition unit includes a vehicle dynamic response acquisition module 1a, and the first processing unit includes a state analysis and evaluation module 1b. The vehicle dynamic response acquisition module 1a is fixed to the underframe of the front-end vehicle to detect and transmit the vehicle's acceleration to the state analysis and evaluation module 1b. The acceleration includes the lateral and vertical acceleration of the vehicle. The state analysis and evaluation module 1b performs threshold evaluation on the acquired acceleration. Specifically, the purpose of the threshold evaluation is to determine the track degradation level, i.e., based on the stability of the vehicle's on-track operation, in order to provide an accurate operating environment for subsequent data acquisition. The specific evaluation logic is as follows:

[0054] Lateral acceleration ≤ 0.6 m / s² and vertical acceleration ≤ 1.0 m / s², with good track smoothness. The precise acquisition component 2 for track geometry information is not activated (industrial camera and laser ranging module 2c remain in sleep mode), and the dynamic spatial reference stabilization component 3 is not activated (GNSS and IMU remain in sleep mode), continuously storing acceleration data.

[0055] If the lateral acceleration is >0.6 m / s² or the vertical acceleration is >1.0 m / s², the track has defects, and the second acquisition unit is activated.

[0056] Specifically, the track shape acquisition module 2b and the laser ranging module 2c together form the track geometry information precision acquisition component 2, which is installed on the underside of the rear vehicle. The track shape acquisition module 2b is used to acquire the geometric appearance of the track, and the laser ranging module 2c is used to acquire the distance information of the track relative to the vehicle.

[0057] The vehicle positioning module 3a and the inertial reference module 3b together form a dynamic spatial reference stabilization component 3, which is installed on the front or rear vehicle. The vehicle positioning module 3a is used to obtain the vehicle's position information, and the inertial reference module 3b is used to obtain the vehicle's motion state.

[0058] The second processing unit includes a track geometry parameter calculation module 4, which is used to collect vehicle acceleration, track geometry, track distance relative to vehicle, vehicle position information, and vehicle motion state to obtain track dynamic geometry data.

[0059] The system comprises a track state adaptive online evaluation component 1, a track geometry information precision acquisition component 2, a dynamic spatial reference stabilization component 3, and a second processing unit. The second processing unit includes a track geometry parameter calculation module 4 and a wireless data transmission and storage module 5b. The track state adaptive online evaluation component 1 controls the activation of the track geometry information precision acquisition component 2 and the dynamic spatial reference stabilization component 3 via the wireless data transmission module 5a. The communication module 3a2 transmits the acquired information to the track geometry parameter calculation module via the CAN bus 3a3. The track geometry parameter calculation module calculates the dynamic geometry data of the track gauge, track level, track elevation, and triangular pits. The track geometry parameter calculation module stores the calculated dynamic geometry data in the data storage module 5b via the wireless transmission module. The modules achieve rapid signal and power connection via wireless transmission, resulting in a high overall protection level and suitability for complex outdoor testing environments.

[0060] This detection device integrates four functional units: track condition perception, geometric information acquisition, spatial benchmark maintenance, and data fusion processing. It constructs a non-intrusive dynamic monitoring system for track geometry and position that can be deployed on locomotives in daily operation. The track condition adaptive online assessment component 1 is located on the front underframe of the train. It uses the dynamic response caused by track irregularities during train operation as input signals, and analyzes and judges the track quality deterioration trend by real-time acquisition of lateral and vertical accelerations combined with preset thresholds. The track geometry information precision acquisition component 2 is located at the bottom of the rear vehicle, away from the traction power source and the core vibration area of ​​the bogie, effectively reducing the impact of mechanical disturbances and ensuring image acquisition stability. It captures the rail cross-sectional contour shape through the track shape acquisition module 2b, and simultaneously uses the laser ranging module 2c to determine the relative distance between the track structure and the onboard equipment, forming high-resolution spatial point cloud data. The dynamic spatial reference stabilization component 3 can be flexibly configured on the front or rear of the vehicle according to actual deployment requirements. It provides absolute geographic coordinates and a time reference through the vehicle positioning module 3a (such as GNSS), and works in conjunction with the inertial reference module 3b (Inertial Measurement Unit, IMU) to continuously output motion parameters such as vehicle attitude angles, angular velocities, and specific forces, jointly establishing a stable measurement reference system unaffected by vehicle vibration. The second processing unit serves as the control and computational center of the entire system, receiving data streams from various units, performing multi-source information synchronization, coordinate transformation, and fusion calculations, and ultimately outputting key track geometry parameters such as track gauge, level, elevation, and triangular crater.

[0061] The aforementioned units do not operate in isolation, but rather are closely linked based on a collaborative logic of "trigger-response-correction-solution". The track state adaptive online evaluation component 1, acting as a front-end sensing node, directly determines whether the track geometry information precision acquisition component 2 is activated, thus achieving on-demand acquisition and avoiding the accumulation of invalid data. The timing of acquisition takes into account the time delay between preceding and following vehicles, ensuring that the same track section can be accurately captured by the back-end acquisition unit after degradation assessment. The position and attitude information provided by the dynamic spatial reference stabilization component 3 permeates the entire data link, serving both spatial registration at the acquisition moment and supporting vibration compensation and coordinate transformation in subsequent parameter calculations. The second processing unit not only undertakes the data aggregation task but also reconstructs the true geometric shape of the track under dynamic conditions through spatiotemporal alignment of information with different physical dimensions and model-driven calculations.

[0062] The above technical solution enables intelligent, adaptive, and high-precision dynamic monitoring of track geometry. Due to the pre-assessment mechanism, the precision acquisition process is activated only when track smoothness indicators exceed a set threshold, significantly reducing data redundancy and energy consumption caused by indiscriminate scanning across the entire track section, thus improving detection efficiency. Placing the assessment and acquisition modules at the front and rear of the train effectively avoids interference from severe front-end vibrations on the optical imaging system, improving image quality and measurement repeatability. The combined positioning and attitude perception capabilities of GNSS and IMU ensure a stable three-dimensional spatial reference even in tunnels, curves, or obstructed environments, guaranteeing the continuity and accuracy of the calculation results. The entire device requires no structural modifications to the locomotive, features rapid assembly and disassembly, is suitable for various vehicle platforms, and supports routine operation on non-dedicated inspection trains, providing railway maintenance departments with a low-cost, all-weather, and sustainable track condition monitoring method.

[0063] Furthermore, the vehicle dynamic response acquisition module 1a of the track condition adaptive online evaluation component 1 uses a triaxial accelerometer 1a1 fixed to an aluminum alloy mounting plate 1a2 at the center of the foremost vehicle underframe. The mounting plate 1a2 is connected to the underframe via countersunk bolts and is located on the transverse symmetrical centerline of the vehicle body. The condition analysis and evaluation module 1b uses a computing chip 1b1 and has a built-in data storage to store a four-level standard threshold library. By setting up an acquisition unit consisting of the mounting plate 1a2 and the triaxial accelerometer 1a1 and arranging it in the middle of the vehicle underframe, high-fidelity acquisition of acceleration signals in key directions during vehicle operation is achieved, thus providing a reliable data foundation for subsequent track condition evaluation.

[0064] The core function of the vehicle dynamic response acquisition module 1a is to acquire acceleration information reflecting the dynamic behavior of the entire vehicle in real time. This module mainly consists of two components: a triaxial accelerometer 1a1 and a mounting plate 1a2. The triaxial accelerometer 1a1 is an inertial sensing device capable of simultaneously measuring acceleration components in three orthogonal directions (typically the X-axis longitudinal, Y-axis lateral, and Z-axis vertical). It features wide frequency response, high precision, and resistance to electromagnetic interference, making it suitable for capturing dynamic signals in complex vibration environments. In this embodiment, the focus is on utilizing its Y-axis and Z-axis outputs, namely lateral and vertical acceleration. These two parameters are directly related to the track smoothness—for example, changes in track gauge, horizontal irregularities, and vertical irregularities can all cause significant vibration responses in the vehicle body in the lateral or vertical directions.

[0065] The triaxial accelerometer 1a1 is fixed to the mounting plate 1a2, which is then detachably attached to the center of the vehicle underframe. "Detachable connection" refers to the use of non-permanent fastening methods such as bolts, quick-connect fittings, or clips, allowing the entire module to be installed and removed without welding or structural modifications. This design greatly enhances the deployment flexibility of the device, making it particularly suitable for temporary installation on non-dedicated inspection trains and meeting the need for rapid switching between different lines and vehicle models.

[0066] Furthermore, the mounting plate 1a2 is positioned in the middle of the vehicle underframe, and optionally on the lateral centerline of the car body. This location has significant engineering implications: firstly, the center of the underframe is far from local excitation sources of the bogie suspension system (such as wheel-rail impact points), effectively avoiding interference from local vibrations on the overall acceleration measurement. Secondly, its position on the centerline of symmetry minimizes torsional coupling effects caused by shifts in the car's center of gravity or uneven load distribution, ensuring that the measured lateral acceleration accurately reflects the lateral excitation caused by track geometric deviations, rather than spurious signals from structural asymmetry. Additionally, this position provides a relatively balanced distance from the front and rear bogies, facilitating the capture of rigid body motion characteristics at the entire vehicle scale and improving data representativeness.

[0067] The above technical solution enables efficient and accurate acquisition of lateral and vertical acceleration during vehicle operation. By placing the triaxial accelerometer 1a1 in the center of the vehicle chassis and along the lateral symmetry centerline, the acquired acceleration signal better represents the overall vehicle's dynamic response characteristics, reducing interference from local structural vibrations and off-center loads, and improving the representativeness and reliability of the data. Simultaneously, the detachable mounting plate 1a2 design allows for rapid installation and removal of the entire acquisition module, significantly improving the adaptability and operational efficiency of the detection device in multi-vehicle and multi-route applications, and solving the problems of long deployment cycles and poor versatility caused by traditional fixed installations.

[0068] Furthermore, this embodiment introduces a computing chip 1b1 with built-in storage capabilities to achieve localized and intelligent processing of lateral and vertical acceleration data collected during vehicle operation. The computing chip 1b1 serves as the core processing unit of the state analysis and evaluation module 1b, undertaking functions such as data reception, logical judgment, and strategy generation. This chip can be a high-performance embedded second processing unit or an industrial-grade microcontroller (such as an ARM Cortex-M series or a DSP), possessing real-time computing capabilities and low power consumption, suitable for long-term stable operation in complex automotive electromagnetic environments. The computing chip 1b1 is connected to the triaxial accelerometer 1a1 in the vehicle dynamic response acquisition module 1a via a communication interface (such as a CAN bus 3a3, RS485, or Ethernet), periodically acquiring acceleration sample values ​​and performing preprocessing operations such as filtering, noise reduction, and unit conversion.

[0069] The 1b1 computing chip's built-in data storage is used to persistently store pre-set four-level standard thresholds. This storage can be on-chip Flash, EEPROM, or an external non-volatile memory unit, ensuring that the threshold data is not lost in the event of a power outage. The stored four-level standard thresholds correspond to the four levels in the national or industry standards for "Dynamic Quality Management Values ​​of Track Lines": Level I is the daily maintenance standard, indicating that the track condition is good and no intervention is required. Level II is the planned maintenance standard, indicating that maintenance should be arranged in the near future. Level III is the temporary repair standard, indicating that the track has obvious defects and needs to be dealt with in a timely manner. Level IV is the speed limit or closure standard, reflecting that there are serious safety hazards on the track. These thresholds can be configured differently according to the actual operating line type (such as high-speed railway, conventional railway, heavy-haul line, etc.) and can be updated through external devices or remotely written wirelessly, improving the system's adaptability and maintainability.

[0070] The four standard threshold levels each set independent judgment thresholds for lateral and vertical acceleration. For example, under typical track conditions, the Level I lateral acceleration threshold could be set to 0.5g, and the vertical acceleration to 0.6g. Level IV, on the other hand, could reach 1.2g and 1.5g respectively (g being the acceleration due to gravity). The computing chip 1b1 compares the real-time collected acceleration values ​​with each threshold level to determine the current exceedance level. If it only slightly exceeds Level I, the event is recorded and the section of concern is marked. If it reaches Level III or above, a high-priority alarm is immediately triggered, and a start command is sent to the track geometry information precision acquisition module to achieve on-demand detection.

[0071] Furthermore, the computing chip 1b1 can also integrate simple decision-making algorithm modules, such as peak statistics, duration discrimination, or trend prediction mechanisms based on sliding windows, to avoid misjudgments caused by instantaneous impacts. For example, if the acceleration value briefly exceeds the Level III threshold but the duration is less than 100ms, the system can determine it as accidental interference and not initiate full detection. Simultaneously, the evaluation results output by the chip can be transmitted to subsequent functional modules via digital signals or communication messages, serving as a basis for adjusting the detection frequency, image acquisition resolution, or data upload strategy.

[0072] The above technical solution enables graded and quantitative assessment of track conditions. By employing a 1b1 computing chip with built-in four-level standard thresholds, the acceleration data analysis process no longer relies on manual experience or back-end processing, enabling rapid response and intelligent decision-making at the front end. Because it can automatically match the corresponding management response level based on the acceleration exceedance level, it solves the problems of "data redundancy due to continuous collection across the entire track section" or "missing sudden defects during fixed-cycle detection" in traditional detection systems, thereby achieving the technical effects of optimizing resource allocation and improving detection efficiency and scientific rigor.

[0073] Furthermore, the track geometry information precision acquisition component 2 includes an acquisition equipment mounting platform 2a made of aviation aluminum. The crossbeam 2a1 is connected to the bottom of the tail locomotive via a quick-release locking structure 2a2. A positioning pin 2a3 is inserted to a certain depth to ensure the crossbeam 2a1 is securely installed. The wedge-shaped locking tongue provides sufficient locking force for quick loading and unloading. The track shape acquisition module 2b uses four high-speed industrial cameras 2b1, each mounted on a pre-reserved mounting port on both sides of the crossbeam 2a1. A laser positioning device 2b2 is used to calibrate the camera lenses, ensuring the lens optical axis coincides with the center line of the rail section. The laser ranging module 2c uses a line laser calibrator to ensure the laser line is perpendicular to the rail section, covering the entire section from the top to the bottom of the rail. It is mounted on a pre-reserved mounting port on the crossbeam 2a1.

[0074] The "high-speed industrial camera 2b1" in the track shape acquisition module 2b refers to an industrial-grade imaging device with a high frame rate (usually no less than 1000 frames / second), high resolution (e.g., 2048×2048 pixels or higher), and strong resistance to ambient light interference. It can capture images of light stripes formed by line lasers illuminating the rail surface in real time during train operation, thereby reconstructing the three-dimensional shape of the track. At least four of these cameras are arranged on the left and right sides of the underside crossbeam 2a1 of the rear vehicle, with at least two cameras on each side, forming a symmetrical distribution. This layout can simultaneously cover the complete cross-sectional area of ​​both rails, improving field-of-view overlap and data redundancy, and enhancing system fault tolerance. For example, if one camera fails due to dirt or temporary obstruction, the remaining cameras can still ensure the effective extraction of key feature points. Furthermore, the camera installation positions are optimized so that the optical axis of its lens is perpendicular to the longitudinal centerline of the track and points to a reference point 16mm inside the working edge of the rail, ensuring that the captured images can be used to accurately calculate key parameters such as track gauge.

[0075] At least two "laser positioning devices 2b2" are installed at both ends of the crossbeam 2a1, near the ends. Their main function is to emit a reference laser beam in the visible or near-infrared band for on-site calibration of the lens of the high-speed industrial camera 2b1. During calibration, the beam emitted by the laser positioning device 2b2 simulates the ideal imaging optical axis direction. The operator or automatic calibration system adjusts the pitch, yaw, and focal length parameters of each camera accordingly, ensuring the camera's optical axis is precisely aligned with the preset centerline of the rail section. This in-situ calibration mechanism effectively compensates for optical system offsets caused by mechanical assembly errors, transportation vibrations, or temperature changes, significantly improving the geometric accuracy of image coordinates. Alternatively, the laser positioning device 2b2 can employ an adjustable gimbal structure, supporting multi-angle adjustment and incorporating a built-in attitude sensor to provide calibration status feedback. Alternatively, a dual-wavelength laser design can be used, corresponding to the calibration needs of different cameras, achieving time-division multiplexing. There is a clear functional synergy between the high-speed industrial camera 2b1 and the laser positioning device 2b2: the laser positioning device 2b2 provides a static spatial reference for establishing the mapping relationship between the camera imaging coordinate system and the track physical coordinate system. The high-speed industrial camera 2b1, based on this reference, continuously acquires track images with structured light encoding during vehicle movement. This combination ensures that the system maintains high image recognition accuracy and measurement stability even under complex conditions (such as sudden changes in lighting or rain / snow). The crossbeam 2a1, serving as the load-bearing foundation of the overall structure, further strengthens this synergistic effect, making the entire acquisition module a compact, robust, and easily maintained integrated unit.

[0076] Through the above technical solution, the track shape acquisition module 2b achieves high-precision and high-reliability image acquisition capabilities. Because a rigid crossbeam 2a1 supports multiple high-speed industrial cameras 2b1 symmetrically arranged on both sides of the vehicle, the limitations of traditional single-point acquisition, such as limited field of view and susceptibility to local occlusion, are solved, improving the integrity and robustness of the detection data. Furthermore, the use of laser positioning devices 2b2 at both ends of the crossbeam 2a1 for in-situ calibration of the camera lenses overcomes imaging inaccuracies caused by mechanical installation deviations and thermal deformation, ensuring the long-term stability of the image coordinate system. Therefore, this embodiment provides a high-quality raw image data foundation for the accurate calculation of subsequent track geometric parameters (such as gauge, level, elevation, and triangular pits), improving the overall accuracy and adaptability of the detection system.

[0077] Preferably, the laser ranging module 2c is installed on the bottom of the crossbeam 2a1 facing the track. This installation position ensures that the laser ranging module 2c can emit a ranging beam towards the track surface at a near-vertical angle, thereby minimizing distance measurement deviations caused by the tilt of the incident angle. Placing the laser ranging module 2c in the central area of ​​the bottom surface of the crossbeam 2a1, so that its ranging optical axis is basically perpendicular to the top surface of the track, facilitates the acquisition of high-precision absolute track height information and provides a stable and reliable vertical reference for subsequent calculations of track level, elevation, and other geometric parameters. This installation position also effectively avoids direct obstruction of the laser window by splashes or dust during vehicle operation, improving the reliability of the equipment in complex environments. Optionally, in some implementation scenarios, the laser ranging module 2c can also be fixed to the bottom of the crossbeam 2a1 using an adjustable mounting bracket, allowing for fine-tuning according to the actual vehicle type or track type to adapt to different track types or installation height requirements. In addition, the laser ranging module 2c can use a pulsed line laser, whose emission frequency is synchronized with the acquisition frame rate of the high-speed industrial camera 2b1, to achieve frame-by-frame matching active illumination imaging, enhance image contrast and suppress ambient light interference.

[0078] The high-speed industrial cameras 2b1 on the crossbeam 2a1 are symmetrically positioned about the laser ranging module 2c. This means that the left and right high-speed industrial cameras 2b1 are located on opposite sides of the laser ranging module 2c, maintaining an equidistant and coaxial geometric relationship with the centerline of the laser ranging module 2c. This symmetrical layout ensures a clear and consistent spatial geometric relationship between the left and right cameras and the laser ranging point, facilitating the establishment of a unified coordinate transformation model. Especially in the three-level mapping process from the pixel coordinate system to the vehicle coordinate system and then to the track coordinate system, it significantly reduces system deviations caused by asymmetric errors. Simultaneously, the symmetrical arrangement helps balance the mass distribution of the entire acquisition system, reducing structural deformation or deflection caused by vehicle vibration or acceleration changes, and improving the overall mechanical stability of the device. Each high-speed industrial camera 2b1 is equipped with a fixed-focal-length lens, whose optical axis is calibrated by the laser positioning instrument 2b2 to coincide with the centerline of the rail section, ensuring that the captured images accurately reflect the true geometry of the track. As an alternative, the number of cameras can be expanded to six, with four used for regular orbit shape capture and two dedicated to edge enhancement recognition, improving the robustness of feature point extraction in low light or rainy / snowy weather conditions.

[0079] The aforementioned laser ranging module 2c and the high-speed industrial camera 2b1, when spatially configured, constitute an integrated, high-precision track geometry information sensing unit. The laser ranging module 2c provides a precise vertical distance reference, while the symmetrically distributed high-speed industrial camera 2b1 is responsible for capturing the lateral contour features of the track. The two complement each other spatially: laser data can be used to correct perspective distortion caused by the camera's viewing angle, while camera images can be used to accurately locate the actual physical positions of each feature point on the laser scanning line. This synergistic effect not only improves the quality of single-frame data but also provides favorable initial conditions for multi-source data fusion algorithms, especially suitable for real-time calculation of parameters such as track gauge and level under dynamic conditions.

[0080] Furthermore, the vehicle positioning module 3a of the dynamic spatial reference stabilization component 3 consists of a GNSS receiver 3a1, a communication module 3a2, and a CAN bus 3a3. The GNSS receiver 3a1 provides three-dimensional coordinates, velocity, and time information. The communication module 3a2 receives all the acquired data and transmits it to the calculation module via the CAN bus 3a3. The inertial reference module 3b consists of a high-precision IMU unit 3b1, which can output X and Y axis attitude data to achieve precise matching of mileage and attitude. By integrating the GNSS receiver 3a1 and the communication module 3a2, a subsystem capable of acquiring and stably transmitting vehicle spatiotemporal position information in real time is constructed, providing crucial spatial reference support for the subsequent accurate calculation of track geometry parameters. Among them, the GNSS receiver 3a1, as the core component of the positioning function, is responsible for acquiring the vehicle's three-dimensional spatial position (longitude, latitude, and elevation), running speed, and precise timestamp in the Earth coordinate system. The communication module 3a2 is responsible for establishing data links and relaying information, ensuring that GNSS data can be delivered to the orbit geometry parameter calculation module 4 in a timely and reliable manner, and participate in multi-source data fusion processing.

[0081] The GNSS receiver 3a1 refers to a Global Navigation Satellite System (GNSS) receiving device. It receives signals from satellite systems such as GPS, BeiDou, GLONASS, or Galileo, and uses the multi-satellite positioning principle to calculate the three-dimensional coordinates (X, Y, Z), three-dimensional velocity (Vx, Vy, Vz), and UTC standard time of the mounted vehicle—i.e., the locomotive—at any given time. This receiver has at least four satellite signal acquisition capabilities and can employ dual-frequency or multi-frequency antenna designs to suppress ionospheric delay errors, improving positioning accuracy to centimeter-level RTK (Real-Time Dynamic Differential) levels. In practical applications, the GNSS receiver 3a1 can be fixedly installed on the top of the leading vehicle or the upper part of the trailing vehicle in an unobstructed area, ensuring good sky visibility and avoiding signal interruptions or drift caused by tunnels, bridges, or dense building complexes. Furthermore, when the receiver incorporates an inertial-assisted algorithm, it can perform dead reckoning based on IMU data in the event of a short-term loss of lock, maintaining positioning continuity.

[0082] The communication module 3a2 is used to realize bidirectional or unidirectional data interaction between the GNSS receiver 3a1 and the track geometry parameter calculation module 4. This module can adopt various forms such as CAN bus 3a3, RS485 serial interface, industrial Ethernet, or wireless communication protocols (such as Wi-Fi 6, 5G NR), and can be flexibly selected according to the vehicle's electrical environment and data throughput requirements. When using CAN bus 3a3, the communication module 3a2 must comply with the ISO 11898 standard, possess electromagnetic interference resistance and multi-node mounting characteristics, and be suitable for vehicle network deployment under complex working conditions. During communication, GNSS data is packaged into a standardized frame format (such as NMEA-0183 or a custom binary protocol), time-stamped, and periodically sent to the calculation unit to achieve data alignment and synchronization with other sensors (such as accelerometers, laser rangefinders, and cameras). To ensure communication stability, the module can also integrate heartbeat detection, retransmission mechanisms, and CRC check functions to prevent data loss or out-of-order delivery.

[0083] The GNSS receiver 3a1 and communication module 3a2 are directly connected via an electrical interface, such as SPI or UART, to output raw observations or positioning results. Communication module 3a2 and track geometry parameter calculation module 4 are interconnected via an onboard communication network, forming a complete data path from sensing to processing. This architecture design ensures that the vehicle's position is no longer isolated information, but serves as a global reference benchmark in the "pixel-vehicle-track" three-level coordinate transformation model, supporting the geographic mapping and trend analysis of parameters such as track gauge, level, and elevation. For example, in track elevation calculation, the absolute elevation provided by GNSS can be used to correct vertical drift caused by IMU integration. In triangulation identification, accurate mileage information combined with level data can accurately extract changes within a specified base length.

[0084] Through the above methods, high-precision and continuous perception of the vehicle's spatial position under dynamic operating conditions is achieved, and this information is seamlessly integrated into the overall data processing flow via a dedicated communication link. The introduction of the GNSS receiver 3a1 enables the system to achieve absolute positioning with geographic coordinates, solving the problem of accurate defect location that traditional relative measurement methods struggle with. Simultaneously, the communication module 3a2 ensures low-latency and high-integrity transmission of positioning data, avoiding calculation deviations caused by data asynchrony. Therefore, this technology effectively improves the spatial consistency of track geometry detection and its value for subsequent operation and maintenance guidance, making it particularly suitable for routine monitoring scenarios on long lines and cross-regional railway networks.

[0085] The inertial reference module 3b, as a core component of the dynamic spatial reference stabilization component 3, provides high-precision, real-time information on the vehicle's motion state to the entire detection system. This module, through the integration of an IMU inertial sensor, continuously senses the vehicle's three-dimensional attitude changes during operation, focusing particularly on the attitude angle output along the X-axis (roll axis) and Y-axis (pitch axis). The IMU inertial sensor itself consists of a three-axis gyroscope, a three-axis accelerometer, and signal processing circuitry. Its working principle is based on inertial measurement principles: the gyroscope measures the angular velocity of the vehicle around each coordinate axis, while the accelerometer measures the specific force (i.e., the non-gravitational acceleration per unit mass) along each axis. By integrating the angular velocity signal and combining it with the static gravity reference information provided by the accelerometer, and using data fusion algorithms such as Kalman filtering or complementary filtering, the real-time roll and pitch angles of the vehicle during operation—that is, the attitude data along the X and Y axes—can be calculated.

[0086] The X-axis attitude data reflects the vehicle's tilt or roll around the longitudinal axis, directly affecting the accuracy of track level parameter measurements. The Y-axis attitude data reflects the vehicle's pitch or tilt around the transverse axis, playing a crucial compensating role in calculating track elevation and triangular craters. This attitude data is synchronously transmitted to the track geometry and position parameter calculation module 4 in the second processing unit to correct measurement deviations caused by changes in the vehicle's own attitude. For example, during high-speed cornering, the vehicle experiences significant roll; without compensation, the rail images captured by the camera will show tilt distortion, leading to errors in gauge and level calculations. Similarly, during ramp operation, vehicle pitch changes the relative relationship between laser ranging and the camera's field of view, affecting the determination of track top elevation. Therefore, by introducing attitude information acquired by the IMU, the measurement results of the acquisition equipment can be dynamically corrected during spatial coordinate transformation, thereby separating the true track geometry deformation.

[0087] As an optional implementation, the IMU inertial sensor can be a highly stable device manufactured using Micro-Electro-Mechanical Systems (MEMS) technology, possessing a wide frequency response range (e.g., 0–200 Hz), low noise density (≤0.01° / √Hz), and high temperature drift suppression capability (temperature drift coefficient <0.005° / ℃), making it suitable for complex operating environments with severe locomotive vibration and large temperature variations. Furthermore, the sensor can communicate with the main control system via RS422 or CAN bus 3a3 interface, ensuring real-time data transmission and interference resistance. In terms of installation layout, the IMU inertial sensor is fixed in the central area of ​​the underframe of the front or rear vehicle, away from high-power motors and strong magnetic field sources, to reduce the impact of external electromagnetic interference on its measurement accuracy.

[0088] By utilizing the real-time attitude angle information output by the IMU inertial sensor, the difference between the vehicle's own motion and the actual deformation of the track is effectively distinguished. Because vehicles inevitably experience roll and pitch motion under dynamic operating conditions, without a precise attitude compensation mechanism, the measured track geometry parameters will deviate significantly from the true values. However, by acquiring and applying attitude data along the X and Y axes, the visual and laser measurement results can be dynamically calibrated during the calculation process. This solves the measurement drift problem caused by the lack of a stable spatial reference in traditional methods, thereby significantly improving the accuracy and reliability of track geometry detection.

[0089] Through the above technical solution, this invention achieves high-precision perception of the vehicle's X-axis and Y-axis attitude. Utilizing the real-time attitude angle information output by the IMU inertial sensor, it effectively distinguishes the difference between the vehicle's own motion and the actual deformation of the track. Because vehicles inevitably experience rolling and pitching motions under dynamic operating conditions, without a precise attitude compensation mechanism, the measured results of track geometry parameters will deviate significantly from the true values. By acquiring and applying the X-axis and Y-axis attitude data, the visual and laser measurement results can be dynamically calibrated during the calculation process, solving the measurement drift problem caused by the lack of a stable spatial reference in traditional methods, thereby significantly improving the accuracy and reliability of track geometry detection.

[0090] Preferably, the wireless transmission module 5a and data storage module 5b consist of two sets of output and receiving devices. One set is installed at the front of the vehicle to enable the response link between the track state adaptive online evaluation component 1, the track geometry information accurate acquisition component 2, and the dynamic spatial reference stabilization component 3, thereby driving whether to activate the corresponding device for data acquisition. The other set is installed inside the mounting bracket on the upper left side of the crossbeam 2a1 to upload the dynamic geometric position information calculated by the track geometry parameter calculation module to the data storage module 5b for later retrieval. The data storage module 5b is connected to the locomotive's onboard power supply through an optocoupler isolation unit, achieving 2500V electrostatic protection to prevent equipment failure from affecting the locomotive's circuitry. Together with the wireless transmission module, it is integrated in the reserved mounting bracket on the upper left side of the crossbeam 2a1 to receive dynamic track geometry data. The maximum power supply is 50W, and it also has a battery backup function, allowing it to operate continuously for 24 hours after a power outage.

[0091] Preferably, the master-slave chip module 4a of the track geometry parameter calculation module integrates a "master processing chip 4a1 + coprocessing chip 4a2", and the algorithm processing module 4b integrates a dual feature point verification method, an inertial vision algorithm, and a 3D visualization algorithm to process the collected data and calculate the dynamic geometry data of the track. The aforementioned master-slave chip module 4a and algorithm processing module 4b are jointly installed in the reserved receiver on the lower right side of the crossbeam 2a1 and connected to the communication module 3a2 of the dynamic spatial reference stabilization component 3 via a CAN bus 3a3. After receiving all the collected data via the bus, they perform geometry calculation. The dynamic geometry data calculation of the track includes track gauge calculation, track level calculation, track elevation calculation, and triangular pit calculation.

[0092] The locomotive-mounted track geometry and position detection device refers to the external detection system described in the preceding claims, which integrates functional modules such as a track state adaptive online evaluation component 1, a track geometry information precision acquisition component 2, a dynamic spatial reference stabilization component 3, and a second processing unit. This device can be installed on the underframe of the front and rear vehicles of ordinary operating locomotives, enabling the dynamic acquisition of high-precision track geometry parameters without the need for a dedicated inspection train. Its core lies in using an accelerometer on the front vehicle to perceive the vehicle's vibration characteristics in real time, and combining this with a rear-end vision and inertial measurement system to calculate key geometric parameters such as track gauge, level, elevation, and triangular crater. The entire device supports wireless control and data transmission, maintains low-power standby in non-operating conditions, and automatically initiates the precision acquisition process under trigger conditions. The specific steps are as follows:

[0093] Step 1: When the locomotive is running, the lateral acceleration and vertical acceleration of the vehicle are collected by the vehicle dynamic response acquisition module 1a.

[0094] The vehicle dynamic response acquisition module 1a is fixed to the front vehicle underframe and is mainly used to detect changes in lateral (i.e., sideways) and vertical acceleration caused by track irregularities during vehicle operation. This module typically includes a triaxial accelerometer 1a1 and a mounting plate 1a2. The mounting plate 1a2 is detachably connected to the middle of the vehicle underframe and located on the lateral symmetry centerline of the vehicle body to ensure that the measurement results are not affected by off-center loading or structural asymmetry. The triaxial accelerometer 1a1 is a microelectromechanical system (MEMS) sensor with high sensitivity, a wide frequency response range (e.g., 0.1 Hz to 200 Hz), and good temperature stability, enabling it to accurately capture the dynamic response signals of the vehicle body under complex track conditions. The acquisition frequency can be set between 100 Hz and 1000 Hz according to the track grade to ensure sufficient data resolution. Lateral acceleration reflects track direction deviation and the rationality of superelevation settings on curve sections, while vertical acceleration mainly reflects issues such as track height differences and unsupported ramps. Together, they constitute the basic input for the preliminary assessment of track smoothness.

[0095] Step 2: Use the state analysis and evaluation module 1b to compare the lateral acceleration and vertical acceleration with preset thresholds to evaluate track smoothness.

[0096] The status analysis and evaluation module 1b receives raw acceleration data from the vehicle dynamic response acquisition module 1a and performs filtering, noise reduction, and peak extraction processing on it. This module has a built-in computing chip 1b1 and is equipped with a data storage device containing four levels of standard thresholds, corresponding to Level I (daily maintenance), Level II (planned maintenance), Level III (temporary repair), and Level IV (speed limit) management limits in the "Railway Line Track Dynamic Quality Management Value Standard". The system automatically matches the corresponding threshold standard based on the current permissible speed of the line, compares the measured acceleration value with it, and determines whether there is an exceedance and its severity. If the acceleration does not exceed the Level I standard, the track condition is considered good, and a low-power mode is maintained. If it reaches or exceeds the Level II standard or above, it is determined that there is a potential defect in the track, triggering subsequent high-precision detection procedures. This evaluation process can be completed within milliseconds, ensuring timely response to track anomalies.

[0097] Step 3: Based on the track smoothness assessment results, the track shape acquisition module 2b is used to acquire the track's geometric appearance, and the laser ranging module 2c is used to acquire the track's distance information relative to the vehicle. The vehicle positioning module 3a is used to acquire the vehicle's position information, and the inertial reference module 3b is used to acquire the vehicle's motion state.

[0098] The track shape acquisition module 2b is located on the underside of the rear vehicle, and a typical configuration includes at least four high-speed industrial cameras 2b1 and laser positioning devices 2b2. The high-speed industrial cameras 2b1 are mounted on both sides of the vehicle horizontally via a crossbeam 2a1. The optical axis of their lenses is aligned with the centerline of the rail section after calibration by the laser positioning devices 2b2, ensuring consistent imaging perspective and measurement accuracy. At least two laser positioning devices 2b2 are located at either end of the crossbeam 2a1, used to periodically calibrate the camera lens attitude and prevent displacement caused by mechanical vibration. When the track shape acquisition module 2b is working, it works in conjunction with the laser ranging module 2c to synchronously acquire images of the track surface contour, forming complete two-dimensional / three-dimensional track shape data. The laser ranging module 2c is installed on the bottom side of the crossbeam 2a1 facing the track, using a pulsed line laser. Its emission frequency is strictly synchronized with the camera frame rate, with each frame corresponding to one laser irradiation, effectively suppressing ambient light interference and reducing the risk of heat accumulation. Its output data includes the vertical distance from the top of the left and right rails to the laser emission point, used for subsequent absolute elevation calculations.

[0099] The vehicle positioning module 3a includes a Global Navigation Satellite System (GNSS) receiver and a communication module 3a2. The GNSS receiver 3a1 acquires the vehicle's three-dimensional coordinates, velocity, and time information (PVT data), achieving centimeter-level positioning accuracy in RTK (Real-Time Differential) mode. The communication module 3a2 transmits the PVT data to the second processing unit in real time via a CAN bus 3a3 or a wireless link, providing a precise spatial reference for orbital geometry parameters. The inertial reference module 3b employs an Inertial Measurement Unit (IMU), containing a three-axis gyroscope and a three-axis accelerometer, to measure the vehicle's angular velocity and specific force along the X-axis (longitudinal), Y-axis (lateral), and Z-axis (vertical), thereby calculating the vehicle's attitude angles (such as pitch angle θ) and mileage increment. After fusing the IMU with GNSS data, continuous and stable attitude estimation can be achieved in highly dynamic environments, overcoming positioning blind spots in areas with GNSS signal obstruction, such as tunnels and bridges.

[0100] Step 4: Based on vehicle acceleration, track geometry, track distance relative to vehicle, vehicle position, and vehicle motion state, acquire and store dynamic geometric shape and position data parameters of the track.

[0101] The acquisition of dynamic track geometry parameters involves the track geometry parameter calculation module 4 in the second processing unit integrating all front-end acquired data and performing fusion calculations of multiple indicators such as track gauge, level, elevation, and triangular pits. The calculation process is based on a three-level transformation model of "pixel coordinates - vehicle coordinates - track coordinates," combined with a stable spatial reference provided by the IMU, to eliminate the influence of vehicle motion on the measurement results. For example, in track gauge calculation, the feature points of the left and right rail interaction edges are first identified from the camera images, and their pixel coordinates are converted into actual spatial distances. In level calculation, the relative elevation difference between the two rails is calculated by combining the pitch angle θ provided by the IMU and the vertical distance between the left and right rail tops obtained from laser ranging. In elevation calculation, the least squares method is used to fit the rail top elevation curve, eliminating vehicle vibration noise. In triangular pit calculation, the difference in horizontal values ​​between the two cross sections is selected based on a fixed base length. All calculation results are labeled with precise mileage and timestamps, and are finally sent to the data storage module 5b for storage via a wireless transmission module. The storage module features optical isolation and electrostatic discharge protection, supports power-off resume function, and ensures data integrity.

[0102] The track gauge calculation involves acquiring the positions of feature points within a 16mm range below the top surface of the inner sides of the left and right rails using image acquisition equipment. Combining vehicle attitude information and camera imaging models, the actual horizontal distance between two feature points is calculated to obtain the current track gauge value. This process is based on visual measurement principles, using a high-speed industrial camera (2b1) to capture light stripe images projected onto the rail cross-section by a line laser. The pixel coordinates of the left and right rail edges are extracted, and calibration parameters are used to convert these pixel coordinates into spatial coordinates in the vehicle coordinate system. Furthermore, considering the influence of vehicle roll, pitch, and yaw angles on the measurement reference, a rotation matrix is ​​used for attitude compensation to ensure that the track gauge calculation results are not affected by changes in the locomotive's dynamic attitude. Optionally, to improve recognition reliability, the system can employ a dual feature point verification mechanism—simultaneously extracting multiple candidate feature points and judging their connection consistency. If the deviation exceeds a preset threshold (e.g., 1mm), a re-recognition process is triggered until a stable matching point pair is obtained. In addition, under low light or strong reflective conditions, the signal-to-noise ratio can be optimized by adjusting the laser pulse frequency and the camera exposure time synchronously. Alternatively, deep learning-assisted edge detection algorithms can be introduced as an alternative to enhance adaptability to complex environments.

[0103] Track leveling calculation refers to the process of obtaining the elevation difference between the top surfaces of the left and right rails on the same cross-section. This calculation is based on the absolute vertical reference provided by the Inertial Measurement Unit (IMU), integrates the vertical distance data from the left and right rail tops to the sensor mounting reference plane output by the laser ranging module 2c, and combines the elevation reference from GNSS positioning information to derive the absolute elevation values ​​of the left and right rail tops, and calculates the difference as the track level value. Specifically, the system first corrects the laser ranging value for tilt based on the pitch angle θ output by the IMU, eliminating measurement errors caused by the tilt of the vehicle body. Then, it deduces the true elevation of the rail tops relative to the geodetic coordinate system using the known installation height relationship. This method avoids the cumulative error problem caused by traditional methods that rely solely on relative displacement measurements, significantly improving the stability and accuracy of horizontal parameters in long-distance detection. As a variant implementation, multi-point laser scanning can also be used instead of single-point ranging. By continuously sampling the rail head area, the top surface curve is fitted, thereby improving the measurement accuracy of sections with large curvature.

[0104] Track elevation calculation refers to the technical means of assessing the vertical undulation of the track surface along the longitudinal direction. Essentially, it involves trend analysis of the rail top elevation at multiple continuous cross-sections within a track section, extracting the degree of deviation from an ideal straight track. During the calculation, the system comprehensively uses displacement information obtained from the vertical acceleration integral of the IMU output and the instantaneous relative rail top height obtained from camera-laser fusion measurements to construct a "vibration filtering" model, effectively separating the locomotive's own vibration components from the actual track elevation changes. Mathematically, the least squares method is typically used to fit a straight line. A linear regression model can accurately reflect the trend of local track depressions or bulges. To further improve robustness, sliding window averaging or Kalman filtering techniques can be introduced during data preprocessing to suppress the influence of random noise. Additionally, for small-radius curves or turnout sections, the sampling density and fitting baseline can be dynamically adjusted to achieve refined characterization.

[0105] Triangular crater calculation refers to measuring the twisting and unevenness of the track within a certain base length range, manifested as the difference in track level values ​​between two cross sections. It is usually set to a fixed value according to railway inspection standards (e.g., an inspection base length of 18 meters or shorter). This parameter is used to identify the presence of track defects, especially significantly affecting the smoothness of train passage. When performing triangular crater calculation, the system relies on continuous mileage information provided by high-precision GNSS / IMU combined positioning to accurately match level data at different locations and automatically selects a base length interval that meets the specifications for difference calculation. To prevent false positives, a dynamic threshold alarm mechanism can be set: when multiple adjacent base length segments show an abnormal triangular crater trend, it is judged as structural deformation rather than accidental fluctuation. Furthermore, it can combine gauge and elevation data for joint diagnosis to identify complex track defects. Specifically, the gauge calculation is determined according to the following formula:

[0106] ,

[0107] In the formula: The final track gauge value after correction by the vehicle roll angle, used for calibration. The original track gauge is represented by the absolute difference in the Y-axis spatial coordinates of the feature points on the left and right rails, derived from the image coordinates. The vehicle roll angle, measured by the IMU inertial sensor, reflects the vehicle's tilt angle around the direction of travel (X-axis). The auxiliary conversion formula involves image coordinates and camera model.

[0108] Among them, the original track gauge Determined according to the following formula:

[0109] ,

[0110] ,

[0111] In the formula: u and v are the image coordinates of the feature points at the end face of the track. The rotation matrix represents the vehicle's attitude, derived from the vehicle's roll angle. Vehicle pitch angle Vehicle yaw angle Composition (all from IMU). , , X, Y, and Z represent the world coordinates of the train's geometric appearance acquisition device, obtained through laser ranging or camera calibration. X, Y, and Z are the world coordinates of feature points in the train's geometric appearance image. K is the camera intrinsic parameter matrix. , This refers to the camera's focal length. , K represents the coordinates of the camera's principal point. 33 is a constant, representing the element in the third row and third column of the camera intrinsic parameter matrix. and Let the principal point coordinates of the camera be, and... , Correspondingly. In the formula This represents the relative position of the feature point and the camera's optical center in the Y direction. In the formula... This represents the relative position of the feature point and the camera's optical center in the Z direction, and is combined with a rotation matrix to eliminate the influence of vehicle attitude. In the formula... It is the relative position of the feature point and the camera optical center in the X direction.

[0112] The orbital level is determined according to the following formula:

[0113] ,

[0114] in, ,

[0115] , ,

[0116] In the formula, S is the original horizontal value of the track. The vertical distance from the top of the left rail to the laser module is obtained through the laser ranging module. The vertical distance from the top of the right rail to the laser module is obtained through the laser ranging module. θ represents the absolute elevation of the laser ranging module. θ is the vehicle pitch angle. This is the absolute elevation of the top of the left rail. This is the absolute elevation of the top of the right rail.

[0117] The orbital elevation is determined according to the following formula:

[0118] ,

[0119] in, ,

[0120] , , ,

[0121] In the formula, Orbital elevation and elevation values; X represents the rail top elevation of the data collection point; X represents the mileage coordinates; L represents the length of the selected track segment; X i H represents the mileage coordinates of the i-th data collection point; i Let be the top elevation of the track at the i-th acquisition point, which is the absolute vertical height of the top of the track at the i-th acquisition point. This is the sampling interval distance; The reference elevation line is obtained by fitting the track segment data using the least squares method; A is the slope of the reference elevation line, which reflects the average trend of the elevation change of the track segment; B is the intercept parameter.

[0122] The solution for the triangular pit is determined by the following formula:

[0123] ,

[0124] in, ,

[0125] In the formula, T is the value of the triangular pit. X i and X j Both S represent the distances of the vehicle at corresponding positions i and j on the track. i For X i The track level value corresponding to the mileage. j For X j The track level value corresponding to the mileage.

[0126] Through the above-described steps, this invention achieves on-demand activation and optimized resource allocation of the detection process. Since the high-power acquisition system is only activated when anomalies are detected in the track smoothness assessment, energy waste and data redundancy issues associated with continuous acquisition across the entire track are avoided. Simultaneously, through the collaborative operation of multiple sensor sources, high-precision, all-weather, and dynamic monitoring of track geometry is achieved. Therefore, this method effectively solves the technical challenges of traditional detection methods, such as high blind detection rates, high costs, and difficulty in routine deployment, achieving the technical effects of improving detection efficiency, reducing maintenance costs, and ensuring train operation safety.

[0127] Based on the above method, the present invention also has the following technical advantages:

[0128] 1. Solve the problems of "high cost, poor deployment flexibility, and coexistence of over- and under-detection in traditional testing"

[0129] The front-end triaxial accelerometer collects the lateral and vertical acceleration of the vehicle body in real time. The track status is intelligently determined by a preset threshold (lateral acceleration ≤ 0.6 m / s² and vertical acceleration ≤ 1.0 m / s² is considered a smooth state). The back-end acquisition and reference unit is activated only when the acceleration exceeds the threshold, avoiding continuous sampling across the entire track.

[0130] 2. Solve the problem of "blurred visual acquisition and large feature point recognition error under complex lighting / severe weather conditions".

[0131] Histogram equalization algorithms specifically improve image quality, particularly in environments with alternating strong light and shadow. By stretching the dynamic range of image grayscale, it suppresses "white block" interference in overexposed areas and enhances details in the darker parts of the rail (rail web and rail bottom edges). Four high-speed industrial cameras are symmetrically arranged, working in conjunction with a laser positioning device to calibrate the lenses in real time, ensuring that the optical axis coincides with the centerline of the rail cross-section. A line laser ranging module covers the entire cross-section from the top to the bottom of the rail, complementing the visual data.

[0132] 3. Solve the problem of "mechanical vibration causing image blurring, IMU data drift, and significant calculation errors".

[0133] Vibration anomalies are precisely eliminated by setting vibration amplitude thresholds of 0.5 m / s² laterally and 0.8 m / s² vertically. Abnormal acceleration data caused by impact vibrations such as track gaps and turnout switching are replaced by a weighted average of five consecutive frames of data, ensuring the accuracy of track condition assessment. A GNSS receiver provides positioning data, and an IMU inertial sensor outputs X / Y axis attitude data. These two data points are fused using Kalman filtering to construct a dynamic spatial reference, effectively offsetting position and attitude drift caused by vehicle vibrations and ensuring a high degree of alignment between mileage and attitude.

[0134] The calculation algorithm ensures parameter accuracy. The second processing unit adopts a "master-slave dual-chip" architecture, combining dual feature point verification method and inertial vision algorithm, integrating multi-source data such as acceleration, vision, laser, and spatial reference, and finally realizes the calculation of four major parameters: track gauge, level, elevation, and triangular pit, effectively solving the pain point of insufficient accuracy of existing external devices.

[0135] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for detecting the geometric shape and position of an externally mounted track on a locomotive, characterized in that, Includes the following steps: Real-time acquisition of the vehicle's lateral and vertical acceleration; The lateral acceleration and vertical acceleration are compared with their respective preset thresholds. When either acceleration exceeds the preset threshold, the vibration interference is determined to be excessive. When vibration interference exceeds the standard, the geometric appearance of the track, the distance information of the track relative to the vehicle, the position information of the vehicle, and the motion state of the vehicle are obtained respectively; wherein, the motion state of the vehicle includes the roll angle, pitch angle, and yaw angle of the vehicle. Based on the track's geometric appearance, the track's distance from the vehicle, the vehicle's position information, and the vehicle's roll, pitch, and yaw angles, the system performs real-time track gauge calculation, track level calculation, track elevation calculation, and triangular pit calculation to obtain the track's dynamic geometric position parameters.

2. The detection method according to claim 1, characterized in that, The track gauge is determined according to the following formula: ， In the formula: For calibration of track gauge; This is the original track gauge; The vehicle's roll angle; Among them, the original track gauge Determined according to the following formula: ， In the formula: u and v are the image coordinates of feature points at the end face of the track; The rotation matrix represents the vehicle's attitude. The vehicle's pitch angle; The vehicle's yaw angle; , , X, Y, Z are the world coordinates of the camera; X, Y, Z are the world coordinates of feature points in the geometric appearance image of the train; K is the camera intrinsic parameter matrix; K 33 is a constant, representing the element in the third row and third column of the camera intrinsic parameter matrix; and These are the coordinates of the camera's principal point.

3. The detection method according to claim 1, characterized in that, The horizontal trajectory calculation is determined according to the following formula: ， in, , , In the formula, S is the original horizontal value of the track; This is the vertical distance from the top of the left rail; This is the vertical distance from the top of the right rail; This is the absolute elevation of the distance measuring point; The vehicle's pitch angle; This is the absolute elevation of the top of the left rail; This is the absolute elevation of the top of the right rail.

4. The detection method according to claim 1, characterized in that, The orbital elevation calculation is determined according to the following formula: ， in, , ， ， ， In the formula, Orbital elevation and elevation values; X represents the rail top elevation of the data collection point; X represents the mileage coordinates; L represents the length of the selected track segment; X i H represents the mileage coordinates of the i-th data collection point; i Let be the top elevation of the track at the i-th acquisition point, which is the absolute vertical height of the top of the track at the i-th acquisition point. This is the sampling interval distance; The reference elevation line is obtained by fitting the track segment data using the least squares method; A is the slope of the reference elevation line, which reflects the average trend of the elevation change of the track segment; B is the intercept parameter.

5. The detection method according to claim 1, characterized in that, The triangular pit solution is determined according to the following formula: ， in, , In the formula, T is the triangular pit value; X i and X j Both S represent the mileage of the vehicle at corresponding positions i and j on the track; i For X i The track level value corresponding to the mileage; S j For X j The track level value corresponding to the mileage.

6. A locomotive-mounted track geometry and position detection device, used to implement the detection method according to any one of claims 1-5, characterized in that, include: The first acquisition unit is used to acquire the vehicle's lateral acceleration and vertical acceleration in real time; The first processing unit is used to compare the lateral acceleration and vertical acceleration with preset thresholds respectively; The second acquisition unit is used to acquire the geometric appearance of the track, the distance information of the track relative to the vehicle, the position information of the vehicle, and the motion state of the vehicle when the lateral acceleration or vertical acceleration is greater than a preset threshold. The second processing unit is used to perform parameter calculations based on vehicle acceleration, track geometry, track distance relative to vehicle, vehicle position, and vehicle motion state to obtain dynamic geometric position parameters of the track.

7. The locomotive external track geometry and position detection device according to claim 6, characterized in that, The first acquisition unit includes a triaxial accelerometer and a mounting plate. The mounting plate is detachably connected to the vehicle's chassis and is located in the middle of the vehicle's chassis. The triaxial accelerometer is fixed to the mounting plate and is used to simultaneously detect the vehicle's lateral acceleration and vertical acceleration.

8. The locomotive external track geometry detection device according to claim 7, characterized in that, The second acquisition unit includes a track shape acquisition module, a laser ranging module, a vehicle positioning module, and an inertial reference module. The track shape acquisition module is used to acquire the geometric appearance of the track, the laser ranging module is used to acquire the distance information of the track relative to the vehicle, the vehicle positioning module is used to acquire the position information of the vehicle, and the inertial reference module is used to acquire the motion state of the vehicle.

9. The locomotive external track geometry detection device according to claim 8, characterized in that, The track shape acquisition module includes a laser positioning device and at least four high-speed industrial cameras. The high-speed industrial cameras are respectively mounted on both sides of the rear vehicle in the horizontal direction via a crossbeam. There are at least two laser positioning devices, which are respectively installed at both ends of the crossbeam for calibrating the lenses of the high-speed industrial cameras.

Images