A dynamic detection method for track irregularities based on inertial navigation and non-filtering

Abstract

The application discloses a kind of based on inertial navigation and non-filtered track irregularity dynamic detection method, belong to railway track detection field, comprising: based on inertial measurement unit, with zero speed information, speed sensor, GNSS positioning result is carried out global combined navigation solution, obtain global reference navigation state sequence;To the two-dimensional point cloud obtained by scanning the surface of track to be measured is carried out point cloud registration and key point extraction, obtain key point coordinates;With target chord length as window length, it is carried out sliding window inertial navigation system mechanical arrangement, obtain bogie motion trajectory;Based on global reference navigation state sequence, key point coordinates and bogie motion trajectory, it is carried out three-dimensional dynamic rod arm position compensation, obtain track shape trajectory: calculate track irregularity parameter, complete track irregularity dynamic detection.The application realizes track irregularity dynamic detection, need not carry out band-pass filtering processing, information loss is less, can avoid waveform distortion.

Description

Technical Field

[0001] This invention belongs to the field of railway track inspection, specifically relating to a dynamic detection method for track irregularities based on inertial navigation and non-filtering. Background Technology

[0002] Track smoothness is fundamental to the stable operation of railway wheel-rail systems, and its accuracy and reliability are crucial factors affecting train safety and passenger comfort. Track irregularity detection is essentially a relative measurement, determining the geometric characteristics of the track within a local section, and can be categorized into static and dynamic detection scenarios. Static detection refers to using relatively lightweight portable track inspection trolleys or stationary observation methods to measure track irregularity parameters under no dynamic loads. This method provides relatively rich constraint information, thus typically achieving high detection accuracy, but with lower efficiency. Dynamic detection involves mounting measuring equipment (usually inertial sensors) on a high-speed train to measure track irregularity parameters under dynamic loads. This method offers higher detection efficiency and is often used for periodic track inspections and overall condition assessments, but the lack of motion constraints presents challenges in terms of measurement accuracy.

[0003] Current methods for dynamic detection of track irregularities primarily rely on the inertial reference method. This method involves mounting inertial sensors on the steering angle or the train car to measure data. First, a unidirectional displacement sequence is obtained through double integration of unidirectional acceleration or single integration of unidirectional angular velocity and velocity. Then, compensation for the relative motion between the train and track is performed using laser or other displacement sensors. Finally, frequency domain bandpass filtering extracts irregularity information within the target waveband. A key objective of frequency domain bandpass filtering is to eliminate the accumulation of errors caused by long-term single-axis integration, thus restoring detailed waveforms.

[0004] In practice, the application of the aforementioned inertial reference method has the following problems: Firstly, due to the simplification of the single-axis integration algorithm, some motion accelerations and angular velocities can interfere with the measurement results, leading to a decrease in the accuracy of long-wavelength irregularity measurements. Secondly, due to information loss caused by the bandpass filter, the results are not only prone to waveform distortion but also lack clear geometric meaning, making it impossible to directly output track deviations for track adjustment. In summary, achieving high-precision, lossless, and geometrically clear dynamic measurement of track irregularities is clearly of great value for track inspection and maintenance in the railway industry. Summary of the Invention

[0005] To address the issues of information loss caused by bandpass filtering and interference with measurement results from motion acceleration and angular velocity in existing technologies, this invention provides a dynamic track irregularity detection method based on inertial navigation and without filtering. This method effectively avoids the theoretical defects and information loss of traditional filtering methods, and the measurement results can be directly used to guide track fine-tuning. Through global integrated navigation solution, key point extraction, mechanical arrangement of the sliding window inertial navigation system, three-dimensional dynamic lever position compensation, and calculation of track irregularity parameters, dynamic track irregularity detection is achieved. This method eliminates the need for bandpass filtering, resulting in less information loss. It avoids the theoretical defects, information loss, and waveform distortion of traditional filtering methods, and can output geometrically meaningful track deviations to guide track fine-tuning.

[0006] According to one aspect of the present invention, a method for dynamic detection of track irregularities based on inertial navigation and without filtering is provided, comprising:

[0007] Based on the data from the inertial measurement unit, combined with zero-velocity information, velocity sensor detection results, and GNSS positioning results, a global integrated navigation solution is performed to obtain a global reference navigation state sequence.

[0008] Based on the two-dimensional point cloud obtained by scanning the surface of the track to be tested, point cloud registration and key point extraction are performed to obtain the coordinates of the key points in the coordinate system of the inertial measurement unit.

[0009] Based on the global reference navigation state sequence, the window length is set according to the target chord length, and the mechanical arrangement of the sliding window inertial navigation system is performed to obtain the bogie motion trajectory in each window;

[0010] Based on the attitude sequence in the global reference navigation state sequence, the coordinates of key points in the inertial measurement unit coordinate system, and the bogie motion trajectory in each window, three-dimensional dynamic lever position compensation is performed to obtain the track shape trajectory in each window.

[0011] Based on the attitude sequence in the global reference navigation state sequence, the coordinates of key points in the inertial measurement unit coordinate system, the bogie motion trajectory in each window, and the track shape trajectory, the track irregularity parameters are calculated.

[0012] Furthermore, a global integrated navigation solution is performed, including:

[0013] Position initialization is performed based on GNSS positioning results or known approximate location; velocity initialization is performed based on velocity sensor detection results, and zero velocity correction is performed based on zero velocity information; attitude initialization is performed using anti-sway methods.

[0014] The extended Kalman filter method is used to estimate the data of the inertial measurement unit, the initial position, velocity and attitude, and obtain the global reference navigation state sequence.

[0015] Furthermore, based on the two-dimensional point cloud obtained by scanning the surface of the track under test, point cloud registration and key point extraction are performed, including:

[0016] Based on the standard design profile of the track to be tested, point cloud registration is performed on the two-dimensional point cloud obtained by scanning the surface of the track to be tested.

[0017] Based on the point cloud registration results, the location of the key points is found, and the coordinates of the key points in the inertial measurement unit coordinate system are obtained.

[0018] Furthermore, based on the global reference navigation state sequence, the window length is set according to the target chord length, and the mechanical arrangement of the sliding window inertial navigation system is performed, including:

[0019] The navigation state and inertial measurement unit error state at any specified starting point are extracted from the global reference navigation state sequence by time interpolation.

[0020] Based on the navigation state and inertial measurement unit error state of any specified starting point, the inertial navigation system is mechanically arranged within a sliding window to obtain the updated state of each epoch as the starting state for the next epoch recursion.

[0021] Using the target chord length as the window length, the endpoint time of the window is determined based on the trajectory in the updated state obtained from the mechanical arrangement of the inertial navigation system.

[0022] Furthermore, the expression for three-dimensional dynamic lever arm position compensation is as follows:

[0023] ,

[0024] in, The trajectory of the bogie's movement; The trajectory is the shape of the track; is the projection of the three-dimensional vector from the center of the inertial measurement unit (IMU) to the target point on the top or side of the track in the IMU coordinate system; b represents the IMU coordinate system; n represents the navigation coordinate system; This is the coordinate transformation matrix between the inertial measurement unit coordinate system and the navigation coordinate system.

[0025] Furthermore, based on the global reference navigation state sequence, the coordinates of key points in the inertial measurement unit coordinate system, the bogie motion trajectory within each window, and the track shape trajectory, track irregularity parameters are calculated, including:

[0026] The track gauge of the track to be measured is calculated based on the coordinates of the key points in the inertial measurement unit coordinate system.

[0027] The superelevation of the track under test is calculated based on the coordinates of the key points in the inertial measurement unit coordinate system and the roll angle sequence of the attitude sequence in the global reference navigation state sequence.

[0028] Based on the trajectory shape, calculate the lateral and vertical versines of the track to be measured.

[0029] Furthermore, after calculating the track irregularity parameters, the following is included:

[0030] By subtracting the track irregularity parameters from their standard values, we can obtain track gauge irregularity, superelevation irregularity, track alignment irregularity, and elevation irregularity.

[0031] According to one aspect of the present invention, a dynamic detection device for track irregularities based on inertial navigation and unfiltered method is provided, comprising a detection beam fixedly connected to a railway train, an inertial measurement unit and a two-dimensional laser line scanning sensor mounted on the detection beam, and a data processing unit, wherein the data processing unit receives measurement information transmitted in real time by the inertial measurement unit and the two-dimensional laser line scanning sensor, and executes the aforementioned dynamic detection method for track irregularities based on inertial navigation and unfiltered method.

[0032] Furthermore, the detection beam is used to mount an inertial measurement unit and a two-dimensional laser line scanning sensor; the inertial measurement unit includes a three-axis accelerometer and a three-axis gyroscope, used to measure three-dimensional specific force and angular velocity; the two-dimensional laser line scanning sensor is used to acquire a two-dimensional point cloud of the track surface to be measured.

[0033] Compared with the prior art, the beneficial effects of the present invention are:

[0034] 1. This invention achieves dynamic detection of track irregularities by global integrated navigation solution, key point extraction, mechanical arrangement of sliding window inertial navigation system, three-dimensional dynamic lever position compensation, and calculation of track irregularity parameters. It does not require bandpass filtering, has less information loss, avoids waveform distortion, and can output track deviations with clear geometric meaning to guide track fine-tuning.

[0035] 2. This invention, through the proposed point cloud registration and mechanical arrangement based on the inertial navigation system, more accurately deducts various harmful accelerations and improves the measurement accuracy of long-wave irregularities. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 This is a flowchart of the dynamic detection method for track irregularities based on inertial navigation and non-filtering, according to an embodiment of the present invention.

[0038] Figure 2 This is a schematic diagram of the structural design of a dynamic detection device for track irregularities based on inertial navigation and non-filtering, according to an embodiment of the present invention.

[0039] Figure 3 This is a schematic diagram of the point cloud registration effect in an embodiment of the present invention.

[0040] Figure 4 This is a schematic diagram of three-dimensional lever arm calculation according to an embodiment of the present invention. Detailed Implementation

[0041] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0042] like Figure 1 As shown, this embodiment of the invention proposes a dynamic detection method for track irregularities based on inertial navigation and non-filtering, including: S1: Based on the data from the inertial measurement unit (IMU), combined with the zero-velocity information obtained by zero-velocity correction (ZUPT), the detection results of the velocity sensor, and the GNSS positioning results, a global integrated navigation solution is performed to obtain a global reference navigation state sequence; S2: The two-dimensional point cloud obtained by the two-dimensional laser line scan sensor (line laser) is preprocessed, including point cloud registration and key point coordinate extraction, to obtain the coordinates of key points (track apex and track waist points) in the inertial measurement unit coordinate system; S3: The window length is set according to the target chord length, and sliding is performed. The window-based inertial navigation system (INS) mechanical choreography calculation yields the bogie trajectory (coordinates of the inertial measurement unit center in the navigation system) within each window. The initial state of each window is provided in step S1. S4: Combining the attitude sequence from the global reference navigation state sequence in step S1, the key point coordinates from step S2, and the bogie trajectory from step S3, three-dimensional dynamic lever position compensation is performed to obtain the track shape trajectory (coordinates of key points in the navigation coordinate system) within the window. S5: Based on the results of the above steps, track irregularity parameters are calculated. These parameters include gauge irregularity, superelevation irregularity, alignment irregularity, and elevation irregularity. The global reference navigation state sequence includes the position sequence, velocity sequence, attitude sequence, and inertial navigation error state.

[0043] like Figure 2As shown, this embodiment of the invention proposes a dynamic detection device for track irregularities based on inertial navigation and non-filtering. An inertial measurement unit (IMU) and two line lasers are mounted on the detection beam. The two line lasers are adjusted during installation so that their scanning angles cover the top and sides of the rail head. The IMU and the two line lasers are fixedly connected to the detection beam. Sensor data is transmitted in real-time to a data processing unit for combined navigation and sliding window calculation. The IMU includes a three-axis accelerometer and a three-axis gyroscope for measuring three-dimensional specific force and three-dimensional angular velocity. The raw data measured by the line lasers is a planar point cloud in its own two-dimensional coordinate system. Specifically, the line lasers measure the two-dimensional point cloud formed by the distance between the bogie (or detection beam) and the track under test. The bogie is a core component of all railway trains. Functionally, the bogie is a buffer platform used to isolate the train's wheels and carriages. It can mitigate carriage vibration and improve driving safety through upper and lower secondary suspension systems.

[0044] Specifically, the method for performing global integrated navigation calculation in step S1 of this embodiment of the invention is as follows:

[0045] S11: Initial alignment. Initialize position using a known approximate location or GNSS positioning results. Initialize velocity using a velocity sensor and perform zero-velocity correction based on zero-velocity information. Initialize attitude using anti-sway methods.

[0046] S12: The extended Kalman filter method is used to estimate the state of the carrier. The state to be estimated includes the position, velocity, and attitude of the carrier, as well as the zero bias and scaling factor errors of the three-axis accelerometer and three-axis gyroscope. After the measurement is completed, reverse smoothing methods such as RTS smoothing can be used to optimize the global accuracy.

[0047] Specifically, in step S1, the global integrated navigation uses an IMU + velocity sensor + zero-velocity correction method to construct an extended Kalman filter (EKF). The state vector in the EKF can be represented as:

[0048] (1)

[0049] in, , and Let these represent the position error, velocity error, and attitude error in the n-system, respectively. and These represent the zero-bias errors of the triaxial accelerometer and the triaxial gyroscope, respectively. and These represent the proportional factor errors of the triaxial accelerometer and the triaxial gyroscope, respectively. The superscript T indicates transpose. This represents the global reference navigation state sequence error.

[0050] When using a speed sensor or zero-speed correction for observation updates in the extended Kalman filter, the difference between the INS-calculated speed and the externally observed speed in the vehicle coordinate system is used as the observation vector, and its observation equation is:

[0051] (2)

[0052] in, and These represent the projections of the INS-calculated velocity and the externally observed velocity into the v-frame, respectively. v represents the vehicle coordinate system, i.e., the v-frame; b represents the inertial measurement unit coordinate system, i.e., the b-frame; and n represents the local horizontal coordinate system, also known as the navigation coordinate system, i.e., the n-frame. and Let represent the coordinate transformation matrices between the b-system and the n-system, and between the v-system and the b-system, respectively; This represents the error in the projection of the vehicle's three-dimensional velocity into the n-frame. Indicates attitude error. and Let each represent an antisymmetric matrix of two vectors; The projection of the three-dimensional vector pointing from the IMU center to the velocity sensor position onto the b-frame. This represents the error in the projection of the Earth's rotational angular velocity into the b-frame. This indicates observation noise.

[0053] Global integrated navigation fuses velocity information as an external aid with inertial navigation calculations. Data fusion methods include, but are not limited to, Kalman filtering, extended Kalman filtering, unscented Kalman filtering, particle filtering, sequential least squares, or artificial neural network methods.

[0054] Specifically, the preprocessing method for the linear laser in step S2 of this embodiment of the invention is as follows:

[0055] S21: Based on the standard design profile of the track to be tested, perform point cloud registration on the two-dimensional point cloud measured by the line laser to obtain the rotation angle of the line laser coordinate system relative to the inertial measurement unit coordinate system.

[0056] S22: In the point cloud registration results, find the positions of the track vertex and track waist points according to the correspondence of the points, and record their two-dimensional coordinates in the inertial measurement unit coordinate system.

[0057] Specifically, in step S2, the Iterative Closest Point (ICP) algorithm is used for point cloud registration. Its basic principle is to solve for the rotation parameter R and translation parameter t of the Euclidean transform, and to perform iterative calculations based on the least squares method to minimize the sum of squared errors.

[0058] (3)

[0059] in, and Let represent the coordinates of each point in the two sets of points, and m represent the number of points in each set.

[0060] (4)

[0061] Here, P and Q represent the two point sets to be registered. When solving for the rotation matrix, the two point sets are first centered:

[0062] (5)

[0063] in, and Let represent the coordinates of the points after decentralization, and , where i represents the index of the summation symbol. Then, calculate the covariance matrix H:

[0064] (6)

[0065] Then perform SVD decomposition on H:

[0066] (7)

[0067] Here, U and V represent matrices formed by a set of orthonormal bases in the input and output spaces, respectively, and are also the results obtained from SVD decomposition. Finally, the rotation and translation parameters are obtained:

[0068] (8)

[0069] Point cloud registration is the process of transforming the two-dimensional point cloud of the orbit measured by line laser into a coordinate system associated with the IMU. Registration methods that can be used include, but are not limited to, ICP, normal distribution transformation, consistent point drift, global optimization, or deep learning-based global registration methods.

[0070] Specifically, the sliding window solution method in step S3 of this embodiment of the invention is as follows:

[0071] S31: For any specified starting point, extract the navigation state and IMU error state of the starting point from the global reference navigation state sequence in step S1 by time interpolation.

[0072] S32: Perform INS mechanical orchestration within the sliding window, i.e., iteratively update position, velocity and attitude, with the update state of each epoch serving as the starting state for the next epoch.

[0073] S33: Using the target chord length as the window length, the endpoint time of the window is determined based on the trajectory obtained from speed sensor data or INS mechanical arrangement, and precise interception is ensured through time interpolation. Among the track irregularity parameters, track orientation and elevation irregularities are classified and statistically analyzed according to different chord lengths, such as commonly used 20m chords and 70m chords, and the target chord length is determined according to specific requirements.

[0074] In step S3, local INS mechanical arrangement is performed in each window according to the specified initialization conditions. The specific arrangement steps are conventional methods in this field, so they will not be described in detail here.

[0075] Specifically, the principle of three-dimensional dynamic lever arm compensation in step S4 of this embodiment of the invention is as follows:

[0076] (9)

[0077] in, The projection of the three-dimensional vector from the center of the inertial measurement unit to the target point on the top or side of the track in the b-frame; The bogie motion trajectory obtained in step S3 based on INS mechanical arrangement; The coordinates of the key points in the navigation coordinate system; rail represents the track (e.g., ...). Figure 2 (The left and right rails in the middle).

[0078] In step S4, the required lever arm value for three-dimensional dynamic lever arm compensation. The results are obtained by combining the line laser registration results with the fixed installation position of the equipment. The change in the lever arm value represents the high-frequency relative motion between the vehicle and the rail, such as... Figure 3 As shown. Therefore, it is necessary to accurately calculate the corresponding lever value for each IMU epoch based on the time difference.

[0079] When selecting specific data, it is necessary to perform three-dimensional dynamic lever arm compensation on the key points extracted from the top and side surfaces of the rail head, such as... Figure 4 As shown. This is because the top point is needed to calculate track superelevation and elevation irregularities, while the side point is needed to calculate gauge and directional irregularities. The calculation method for the lever arms in both locations is as follows:

[0080] (10)

[0081] in, and These represent the final total rod arm vectors corresponding to the track vertices and track waists, respectively. This represents the fixed three-dimensional vector from the IMU measurement center to the line laser measurement center during equipment installation. and These represent the time-varying three-dimensional vectors from the center of the line laser measurement to the top and side surfaces of the rail head, respectively.

[0082] Specifically, point cloud registration results in the coordinates of key points in the carrier coordinate system (b system), while lever compensation results in the coordinates of key points in the navigation coordinate system (n system). The former is used for calculating track gauge and superelevation, while the latter is used for calculating lateral and vertical versines.

[0083] Specifically, the method for calculating the track irregularity parameter in step S5 of this embodiment of the invention is as follows:

[0084] S51: Calculate the track gauge based on the point cloud registration results (coordinates of key points in the inertial measurement unit coordinate system):

[0085] (11)

[0086] in, and Let represent the lateral distances from the two line lasers to the two side tracks obtained in step S2, respectively. This indicates the fixed installation distance between two line lasers. This indicates the measured track gauge.

[0087] S52: Calculate the superelevation based on the point cloud registration results and the globally calculated roll angle sequence (which belongs to the attitude sequence):

[0088] (12)

[0089] Where L represents the slant distance of 1500mm corresponding to the standard gauge. This represents the roll angle sequence in the attitude sequence obtained in step S1. and The vertical distances from the two side laser beams to the two side tracks obtained in step S2 are respectively. This indicates that an extremely high altitude was measured.

[0090] S53: Based on the results of the 3D dynamic lever arm compensation (coordinates of key points in the navigation coordinate system), calculate the lateral versine to obtain the trajectory irregularities:

[0091] (13)

[0092] in, , and These represent the planar coordinates of the window start point, midpoint, and end point obtained in step S4 after three-dimensional dynamic lever arm compensation; This represents the measured lateral versine.

[0093] S54: Based on the results of the 3D dynamic lever arm compensation (coordinates of key points in the navigation coordinate system), calculate the vertical versine to obtain the elevation unevenness:

[0094] (14)

[0095] in, , and These represent the elevations of the window start point, midpoint, and end point obtained in step S4 after three-dimensional dynamic lever arm compensation, respectively. This represents the measured vertical versine.

[0096] S55: Subtract the design values ​​from the parameters measured in steps S51, S52, S53, and S54 to obtain the gauge irregularity, superelevation irregularity, track alignment irregularity, and elevation irregularity. The design values ​​refer to the standard values ​​used in railway track design and should be considered existing data during engineering construction and track inspection and maintenance.

[0097] In step S5, track gauge and superelevation parameters are absolute measurements and can be directly calculated using the navigation state sequence obtained from global integrated navigation and the point cloud registration results; track orientation and elevation irregularities are relative measurements and the trajectory length must be accurately truncated within each sliding window.

[0098] It should be noted that, based on relevant theoretical derivations and experimental results, in the measurement scheme proposed in this embodiment of the invention, the position and velocity errors during the initialization of the sliding window have a minimal impact on the measurement error of track irregularities. Therefore, the global integrated navigation calculation in step S1 of this embodiment does not need to achieve high-precision position and velocity calculations; only approximately accurate values ​​are required (typically better than 200m and 0.2m / s). Meanwhile, the initial attitude error of the window has a significant impact on the measurement error of irregularities, especially the initial roll angle error, which has a very large impact on the orbital orientation. Therefore, it is necessary to select an IMU with high gyro accuracy or provide sufficient external observations in the integrated navigation filter to optimize the attitude accuracy of the global calculation as much as possible (typically better than 0.01deg) to support millimeter-level accuracy in track irregularity measurement. These requirements are not difficult to meet in civilian navigation-grade inertial navigation systems; therefore, the track irregularity dynamic detection system proposed in this embodiment of the invention is feasible and practical.

[0099] This invention also provides a dynamic detection device for track irregularities based on inertial navigation and unfiltered operation, comprising a detection beam fixedly connected to a railway train, an inertial measurement unit and a two-dimensional laser line scan sensor mounted on the detection beam, and a data processing unit. The data processing unit receives measurement information transmitted in real time from the inertial measurement unit and the two-dimensional laser line scan sensor, and executes the dynamic detection method for track irregularities based on inertial navigation and unfiltered operation described in the foregoing embodiments. The detection beam carries the inertial measurement unit and the two-dimensional laser line scan sensor; the inertial measurement unit includes a three-axis accelerometer and a three-axis gyroscope for measuring three-dimensional specific force and angular velocity; the two-dimensional laser line scan sensor acquires a two-dimensional point cloud of the track surface to be measured.

[0100] Finally, it should be noted that the above specific embodiments are merely representative examples of the present invention. Obviously, the present invention is not limited to the above specific embodiments and many variations are possible. Any simple modifications, equivalent changes, and alterations made to the above specific embodiments based on the technical essence of the present invention should be considered within the protection scope of the present invention.

Claims

1. A dynamic detection method for track irregularities based on inertial navigation and unfiltered methods, characterized in that, include: Based on the data from the inertial measurement unit, combined with zero-velocity information, velocity sensor detection results, and GNSS positioning results, a global integrated navigation solution is performed to obtain a global reference navigation state sequence. Based on the two-dimensional point cloud obtained by scanning the surface of the track to be tested, point cloud registration and key point extraction are performed to obtain the coordinates of the key points in the coordinate system of the inertial measurement unit. Based on the global reference navigation state sequence, the window length is set according to the target chord length, and the mechanical arrangement of the sliding window inertial navigation system is performed to obtain the bogie motion trajectory in each window; Based on the attitude sequence in the global reference navigation state sequence, the coordinates of key points in the inertial measurement unit coordinate system, and the bogie motion trajectory in each window, three-dimensional dynamic lever position compensation is performed to obtain the track shape trajectory in each window. Based on the attitude sequence in the global reference navigation state sequence, the coordinates of key points in the inertial measurement unit coordinate system, the bogie motion trajectory in each window, and the track shape trajectory, the track irregularity parameters are calculated.

2. The method for dynamic detection of track irregularities based on inertial navigation and non-filtering according to claim 1, characterized in that, Perform global integrated navigation solution, including: Position initialization is performed based on GNSS positioning results or known approximate location; velocity initialization is performed based on velocity sensor detection results, and zero velocity correction is performed based on zero velocity information; attitude initialization is performed using anti-sway methods; The extended Kalman filter method is used to estimate the data of the inertial measurement unit, the initial position, velocity and attitude, and obtain the global reference navigation state sequence.

3. The method for dynamic detection of track irregularities based on inertial navigation and non-filtering according to claim 1, characterized in that, Based on the two-dimensional point cloud obtained by scanning the surface of the track under test, point cloud registration and key point extraction are performed, including: Based on the standard design profile of the track to be tested, point cloud registration is performed on the two-dimensional point cloud obtained by scanning the surface of the track to be tested. Based on the point cloud registration results, the location of the key points is found, and the coordinates of the key points in the inertial measurement unit coordinate system are obtained.

4. The method for dynamic detection of track irregularities based on inertial navigation and non-filtering according to claim 1, characterized in that, Based on the global reference navigation state sequence, the sliding window inertial navigation system is mechanically orchestrated according to the target chord length, including: The navigation state and inertial measurement unit error state at any specified starting point are extracted from the global reference navigation state sequence by time interpolation. Based on the navigation state and inertial measurement unit error state at any specified starting point, the inertial navigation system is mechanically arranged within a sliding window to obtain the updated state of each epoch as the starting state for the next epoch recursion. Using the target chord length as the window length, the endpoint time of the window is determined based on the trajectory in the updated state obtained from the mechanical arrangement of the inertial navigation system.

5. The method for dynamic detection of track irregularities based on inertial navigation and non-filtering according to claim 1, characterized in that, The expression for three-dimensional dynamic lever arm position compensation is: , in, The trajectory of the bogie's movement; The trajectory is the shape of the track; is the projection of the three-dimensional vector from the center of the inertial measurement unit (IMU) to the target point on the top or side of the track in the IMU coordinate system; b represents the IMU coordinate system; n represents the navigation coordinate system; This is the coordinate transformation matrix between the inertial measurement unit coordinate system and the navigation coordinate system.

6. The method for dynamic detection of track irregularities based on inertial navigation and non-filtering according to claim 1, characterized in that, Based on the global reference navigation state sequence, the coordinates of key points in the inertial measurement unit coordinate system, the bogie motion trajectory within each window, and the track shape trajectory, track irregularity parameters are calculated, including: The track gauge of the track to be measured is calculated based on the coordinates of the key points in the inertial measurement unit coordinate system. The superelevation of the track under test is calculated based on the coordinates of the key points in the inertial measurement unit coordinate system and the roll angle sequence of the attitude sequence in the global reference navigation state sequence. Based on the trajectory shape, calculate the lateral and vertical versines of the track to be measured.

7. The method for dynamic detection of track irregularities based on inertial navigation and non-filtering according to claim 6, characterized in that, After calculating the track irregularity parameters, the following is included: By subtracting the track irregularity parameters from their standard values, we can obtain track gauge irregularity, superelevation irregularity, track alignment irregularity, and elevation irregularity.

8. A dynamic detection device for track irregularities based on inertial navigation and non-filtering, characterized in that, The system includes a detection beam fixedly connected to a railway train, an inertial measurement unit and a two-dimensional laser line scanning sensor mounted on the detection beam, and a data processing unit. The data processing unit receives measurement information transmitted in real time from the inertial measurement unit and the two-dimensional laser line scanning sensor, and executes a dynamic detection method for track irregularities based on inertial navigation and non-filtering as described in any one of claims 1-7.

9. The dynamic detection device for track irregularities based on inertial navigation and non-filtering according to claim 8, characterized in that, The detection beam is used to mount an inertial measurement unit and a two-dimensional laser line scanning sensor; the inertial measurement unit includes a three-axis accelerometer and a three-axis gyroscope, used to measure three-dimensional specific force and angular velocity; the two-dimensional laser line scanning sensor is used to acquire a two-dimensional point cloud of the track surface to be measured.

Images