Marking point laser positioning monitoring method and system for railway track

Abstract

The application provides a marking point laser positioning monitoring method and system for a railway track, relates to the technical field of laser positioning monitoring, and comprises the following steps: collecting a laser reflection signal of a track marking point, constructing a bimodal feature set, and calculating a marking point credibility index; obtaining real space coordinates and relative space coordinates of the track marking point, adjusting the weight of the track marking point, solving a track multi-attitude parameter vector, obtaining a predicted position of each track marking point, extracting a resonance energy spectrum feature, performing coupling relationship inversion suppression on the resonance energy spectrum feature, and outputting a monitoring anomaly according to the inversion suppression result and the track multi-attitude parameter vector. The technical problems that the existing railway track monitoring technology has low efficiency, limited precision, and leads to the fact that track anomalies cannot be identified and positioned in real time and accurately are solved. The technical effects of identifying and positioning track anomalies in real time and with high precision and improving the railway track monitoring level and train operation safety are achieved.

Description

Technical Field

[0001] This invention relates to the field of laser positioning and monitoring technology, and specifically to a laser positioning and monitoring method and system for marker points on railway tracks. Background Technology

[0002] During long-term use, railway tracks are affected by various factors such as repeated train loads, natural environmental erosion, and changes in geological conditions, resulting in abnormal phenomena such as track geometry deviations, rail wear, and sleeper damage. If these abnormalities are not detected and addressed in a timely manner, they will seriously affect the safety and stability of train operations and may even lead to serious traffic accidents.

[0003] Traditional railway track monitoring methods mainly rely on manual inspections and periodic static testing equipment. Manual inspections are not only labor-intensive and inefficient, but also difficult to cover the entire track area, easily leading to omissions or false detections. While periodic static testing equipment can detect track conditions to some extent, the inspection cycle is long, making it impossible to grasp the dynamic changes of the track in real time, resulting in poor timeliness of monitoring data.

[0004] Existing railway track monitoring technologies suffer from low efficiency and limited accuracy, resulting in the inability to identify and locate track anomalies in real time and accurately. Summary of the Invention

[0005] The purpose of this application is to provide a laser positioning and monitoring method and system for marking points on railway tracks, in order to solve the technical problems of low efficiency and limited accuracy in existing railway track monitoring technologies, which result in track anomalies being unable to be identified and located in real time and accurately.

[0006] In view of the above problems, this application provides a method and system for laser positioning and monitoring of marker points on railway tracks.

[0007] The first aspect of this application provides a laser positioning and monitoring method for track markers on railway tracks. The method includes: acquiring laser reflection signals from track markers, extracting differential reflection spectral characteristics and geometric distribution characteristics of laser spots to construct a dual-modal feature set of the marker's spectrum and geometry; calculating a marker confidence index using the dual-modal feature set; obtaining the true spatial coordinates of the track markers and reading the real-time acquired relative spatial coordinates; adjusting the track marker weights based on the marker confidence index; calculating the lateral displacement, vertical displacement, torsion angle, and gauge changes of the track during the current monitoring period to form a multi-attitude parameter vector of the track; predicting the predicted positions of each track marker at the next monitoring time based on the multi-attitude parameter vector of the track; dynamically updating the marker status based on the comparison results of the actual observed positions and predicted positions at the next monitoring time and the confidence index; acquiring track vibration signals; extracting resonant energy spectrum characteristics; filtering the multi-attitude parameter vector of the track based on the updated marker status; performing inversion suppression of the coupling relationship with the resonant energy spectrum characteristics; and outputting monitoring anomalies based on the inversion suppression results and the multi-attitude parameter vector of the track.

[0008] Optionally, baseline correction and segmented filtering are performed on the laser reflection signal, the target spectrum is divided into multiple sub-bands according to a preset band, and the spectral difference gradient sequence of adjacent sub-bands is calculated; multi-scale Gaussian pyramid decomposition is performed on the spot image of the track marker point, and the ellipse fitting parameters, centroid offset and edge gradient distribution of the spot are extracted at each scale to form a multi-scale set feature vector; the spectral difference gradient sequence and the multi-scale set feature vector are used as a bimodal feature set, the spectral-geometric coupling consistency index is calculated, and the spectral feature components and geometric feature components are jointly normalized and weighted and fused using the coupling consistency index to output the marker point confidence index.

[0009] Optionally, a multi-point cooperative attitude model containing the neighborhood spatial relationships of each track marker point is constructed based on the real spatial coordinates of each track marker point. A model matrix for attitude calculation is generated based on the multi-point cooperative attitude model. After constructing an adaptive weighted matrix using the marker point confidence index of each track marker point, the model matrix is ​​corrected. The real-time collected relative spatial coordinates are input into the corrected model matrix, and the three-dimensional rigid body transformation matrix of the track is obtained by solving it through weighted least squares or iterative optimization. The lateral displacement, vertical displacement, torsion angle and track gauge changes are analyzed using the multi-point cooperative attitude model and the three-dimensional rigid body transformation matrix to form a multi-attitude parameter vector of the track.

[0010] Optionally, a track node topology matrix is ​​constructed based on the actual spatial coordinates of the track markers. The track node topology matrix represents the spatial connectivity, neighborhood order, and structural constraint relationships between the nodes of the track. For each track marker, a historical motion trend vector of the corresponding track is constructed using historical monitoring data. The attitude calculation weights of each track marker are calculated based on the track node topology matrix and the historical motion trend vector. The model matrix of the multi-point cooperative attitude model is optimized according to the calculation results to establish a multi-attitude parameter vector of the track.

[0011] Optionally, the confidence index of each track marker point is mapped to an initial weight, which reflects the contribution of each marker point to the attitude calculation; the coordinate residual between the relative spatial coordinates and the real spatial coordinates is obtained, an auxiliary weight factor is configured based on the coordinate residual, and the initial weight compensation is performed using the auxiliary weight factor to construct a compensation weight; the compensation weight is classified with confidence, a confidence label is configured, and neighborhood collaborative weighting is performed using the confidence label to construct an adaptive weighting matrix.

[0012] Optionally, the orbital multi-attitude parameter vector is parsed, and the parsed feature set is used as the basic data to construct a time-series fitting prediction model for each orbital marker point; the monitoring interval for the next monitoring time is obtained, and the monitoring interval is used as auxiliary data and sent to the time-series fitting prediction model to output the predicted position of each orbital marker point; residual calculation is performed using the actual observed position and the predicted position to generate a residual comparison result, and the marker point status is dynamically updated using the residual comparison result and the confidence index.

[0013] Optionally, a mapping relationship between the screening results and the resonant energy spectrum characteristics is constructed, including: a: mapping the temporal changes of the orbital multi-attitude parameter vectors of each orbital marker point to the resonant energy spectrum amplitude, frequency drift, and energy distribution changes of the corresponding frequency band; b: constructing a high-dimensional coupling matrix to characterize the multi-point, multi-frequency coupling relationship between orbital attitude and vibration characteristics; suppressing the anomalous energy spectrum components of the high-dimensional coupling matrix based on an adaptive attenuation strategy, and establishing the inversion suppression results.

[0014] Optionally, an abnormal early warning signal is configured based on the abnormal indicators of the monitored anomaly; an anomaly is reported using the abnormal early warning signal, and a monitoring focus is generated simultaneously; and continuous monitoring and management of the railway track is performed based on the monitoring focus.

[0015] Optionally, vibration anomalies can be identified using track vibration signals, vibration anomaly triggering early warning signals can be established, and early warning issuance management of vibration anomaly triggering early warning signals can be implemented.

[0016] A second aspect of this application provides a laser positioning and monitoring system for track markers. The system includes: a reliability calculation module, used to collect laser reflection signals from track markers, extract the differential characteristics of the reflection spectrum and the geometric distribution characteristics of the laser spots, construct a dual-modal feature set of the marker's spectrum and geometry, and calculate a marker reliability index using the dual-modal feature set; and an attitude parameter acquisition module, used to acquire the true spatial coordinates of the track markers, read the real-time acquired relative spatial coordinates, adjust the track marker weights based on the marker reliability index, and calculate the lateral position of the track during the current monitoring period. The system generates a multi-attitude parameter vector for the track by analyzing changes in displacement, vertical displacement, torsion angle, and track gauge. A state update module predicts the positions of track markers at the next monitoring time based on this multi-attitude parameter vector. It dynamically updates the marker state using the residuals between the actual observed and predicted positions at the next monitoring time and the confidence index. A monitoring anomaly output module collects track vibration signals. After extracting the resonant energy spectrum features, it filters the multi-attitude parameter vector based on the updated marker state, performs inversion suppression based on the coupling relationship with the resonant energy spectrum features, and outputs monitoring anomalies based on the inversion suppression results and the multi-attitude parameter vector.

[0017] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0018] The method provided in this application collects laser reflection signals from track markers, extracts the differential characteristics of the reflection spectrum and the geometric distribution characteristics of the laser spots, constructs a dual-modal feature set of the marker's spectrum and geometry, and calculates the marker's confidence index using the dual-modal feature set. It obtains the true spatial coordinates of the track markers and reads the real-time acquired relative spatial coordinates. After adjusting the track marker weights based on the confidence index, it calculates the lateral displacement, vertical displacement, torsion angle, and gauge changes of the track during the current monitoring period, forming a multi-attitude parameter vector of the track. Based on the multi-attitude parameter vector, it predicts the predicted positions of each track marker at the next monitoring time, and dynamically updates the marker state using the comparison results of the actual observed positions and predicted positions at the next monitoring time, and the confidence index. It collects track vibration signals, extracts resonant energy spectrum features, filters the multi-attitude parameter vector based on the updated marker state, and performs inversion suppression of the coupling relationship with the resonant energy spectrum features. Based on the inversion suppression results and the multi-attitude parameter vector, it outputs the monitoring anomaly. This achieves the technical effect of real-time, high-precision identification and location of track anomalies, improving the level of railway track monitoring and train operation safety.

[0019] The above description is merely an overview of the technical solution of this application. To enable a clearer understanding of the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in this application 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 merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0021] Figure 1 A flowchart illustrating the laser positioning and monitoring method for marker points on railway tracks provided in this application.

[0022] Figure 2 This is a schematic diagram of the structure of the laser positioning and monitoring system for marking points on railway tracks provided in this application.

[0023] Figure labeling: 11 Reliability calculation module, 12 Attitude parameter acquisition module, 13 State update module, 14 Monitoring anomaly output module. Detailed Implementation

[0024] This application provides a laser positioning and monitoring method and system for marker points on railway tracks, addressing the technical problems of low efficiency and limited accuracy in existing railway track monitoring technologies, which prevent the real-time and accurate identification and location of track anomalies. It achieves real-time, high-precision identification and location of track anomalies, thereby improving the level of railway track monitoring and the safety of train operation.

[0025] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. It should also be noted that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, not all of them.

[0026] Example 1, as Figure 1As shown, this application provides a laser positioning and monitoring method for marker points on railway tracks, the method comprising:

[0027] By collecting laser reflection signals from track markers, extracting the differential characteristics of the reflection spectrum and the geometric distribution characteristics of the laser spots, a dual-modal feature set of the marker's spectrum and geometry is constructed, and the confidence index of the marker is calculated using the dual-modal feature set.

[0028] Specifically, track markers are identifiable positioning references pre-installed at specific locations on railway tracks for monitoring track geometry; they are fixed points. These track markers possess stable spectral reflectance characteristics, regular geometric shapes, and clear spatial repeatability; examples include high-reflectance coated markers and geometrically regular positioning blocks. High-precision laser emitters and receiving sensors mounted on fixed supports beside the track collect the laser reflection signals from the track markers. The laser emitter, such as a semiconductor laser or a pulsed laser, emits a narrow pulse or continuous wave laser beam towards the track marker. When the laser shines on the marker surface, a reflection signal is generated. Receivers, such as photodiodes or high-speed cameras, collect the returned reflected light and convert it into an electrical signal. This signal is then digitized by an analog-to-digital converter, forming the differential reflectance spectral characteristics and the geometric distribution characteristics of the laser spot. The differential reflectance spectral characteristics refer to the trend of reflection intensity variation in different wavelength ranges, reflecting the marker's characteristic response to laser light. The geometric distribution characteristics reflect the consistency of the laser spot morphology and space, such as whether the spots are regular, symmetrical, and concentrated, reflecting the surface condition and positioning stability of the marker.

[0029] A bimodal feature set of spectral and geometric characteristics is constructed by fusing spectral difference properties and geometric distribution properties to comprehensively and accurately reflect the state of the marker points. Then, the consistency index between the spectral and geometric features is calculated using this bimodal feature set to obtain a marker point reliability index. This reliability index is used to evaluate the reliability of marker point identification and the validity of the data.

[0030] By collecting laser reflection signals from marker points and calculating the marker point reliability index, reliable and comprehensive data is provided for marker point positioning and monitoring, thereby improving the accuracy and reliability of marker point positioning and monitoring.

[0031] Furthermore, the confidence index of the marker point is calculated using the dual-modal feature set, including: performing baseline correction and segmented filtering on the laser reflection signal; dividing the target spectrum into multiple sub-bands according to a preset band; and calculating the spectral difference gradient sequence of adjacent sub-bands; performing multi-scale Gaussian pyramid decomposition on the spot image of the track marker point; extracting the ellipse fitting parameters, centroid offset, and edge gradient distribution of the spot at each scale to form a multi-scale set feature vector; using the spectral difference gradient sequence and the multi-scale set feature vector as the dual-modal feature set, calculating the spectral-geometric coupling consistency index; and using the coupling consistency index to jointly normalize and weightedly fuse the spectral feature components and geometric feature components to output the confidence index of the marker point.

[0032] Specifically, the acquired laser reflection signal undergoes baseline correction and segmented filtering. Baseline correction eliminates baseline drift caused by instrument characteristics and environmental factors such as initial deviation of the laser emitter, zero-point drift of the receiver, temperature changes, and electromagnetic interference. The laser reflection signal is fitted using the least squares method to obtain baseline polynomial coefficients. The polynomial order can be started from low order and gradually increased until the fitting effect meets the requirements. The corrected laser reflection signal is obtained by subtracting the fitted baseline polynomial coefficients from the laser reflection signal.

[0033] The corrected laser reflection signal is subjected to a Fourier transform to convert it into a frequency domain signal. Through frequency domain analysis, the corrected laser reflection signal is divided into multiple frequency bands, such as low-frequency, mid-frequency, and high-frequency bands. Bandpass filtering or wavelet segmented filtering is used to suppress high-frequency noise in different frequency bands while preserving key spectral features. The filtered frequency domain signal is then converted back to the time domain using an inverse Fourier transform to obtain the filtered laser reflection signal. According to a preset band division rule, the target spectrum is divided into several continuous sub-bands, each corresponding to a specific wavelength range. By using a point-by-point subtraction method, the difference between every two adjacent sub-bands is calculated and divided by the wavelength interval to obtain the local gradient of each sub-band, forming a spectral difference gradient sequence. This spectral difference gradient sequence reflects the rate and trend of spectral transformation between different bands. The preset band division rule is based on material property analysis, environmental factors, and experimentally determined preset bands that accurately reflect the state of the marked points and have minimal mutual interference.

[0034] Multi-scale Gaussian pyramid decomposition is performed on the laser spot images of track marker points. This multi-scale Gaussian pyramid decomposition is an image processing technique based on an image pyramid structure. By performing multiple Gaussian blurring and downsampling operations on the original image, a pyramid structure composed of images at different scales is constructed. These scales include high-resolution, medium-resolution, and low-resolution scales. At each scale, features of the laser spot image are obtained from different resolutions and levels of detail, including the ellipse fitting parameters, centroid offset, and edge gradient distribution of the spot. At each scale, the ellipse fitting parameters, centroid offset, and edge gradient distribution of the spot are extracted. The ellipse fitting parameters reflect the shape characteristics of the laser spot in the image; the shape of the laser spot formed by the reflection from marker points in different states will vary. For example, the spot from a normal marker point is closer to a circle, while the spot from a deformed marker point appears elliptical or other irregular shapes. The centroid offset reflects the positional changes of the laser spot in the image and is related to factors such as the vibration and displacement of the marker point. The edge gradient distribution reflects the brightness changes at the edges of the laser spot. The extracted ellipse fitting parameters, centroid offset, and edge gradient distribution are integrated to form a multi-scale ensemble feature vector. This multi-scale geometric feature vector comprehensively reflects the distribution characteristics of the marker points from multiple scales.

[0035] The obtained spectral difference gradient sequence and multi-scale ensemble feature vectors are combined to form a bimodal feature set of the marker points' spectral and geometric dimensions. This bimodal feature set contains information from both spectral and geometric dimensions, providing a more comprehensive and accurate reflection of the marker points' state. The spectral difference gradient sequence and multi-scale ensemble feature vectors in the bimodal feature set are then standardized to eliminate the influence of dimensions; this can be achieved using z-score standardization. A similarity or correlation coefficient calculation algorithm is used to calculate the spectral-geometric coupling consistency index. For example, the Pearson correlation coefficient formula is used; the standardized spectral and geometric data are substituted into the formula to obtain the spectral-geometric correlation coefficient, which serves as the coupling consistency index. The closer the correlation coefficient is to 1, the stronger the spectral-geometric coupling consistency. The coupling consistency index is used to perform joint normalization and weighted fusion of spectral and geometric feature components. Joint normalization refers to normalizing the spectral and geometric feature data, which can be done using min-max normalization. Weighted fusion means assigning different weights based on different coupling consistency levels; that is, the higher the coupling consistency of the marker point, the greater the weight of its feature component in the fusion. The normalized spectral and geometric feature components are weighted according to weighting coefficients, and the confidence index T of each orbital marker point is output. i =w i ×(S i +G i ), where T iw is the confidence index of the i-th marker. i For spectral characteristic components, S i For geometric characteristic components G i The weighting coefficients.

[0036] For example, in a simulated railway track scenario, 10 track markers were set up. High-precision laser transmitting and receiving equipment was used to collect the laser reflection signals from each marker, while a high-speed camera captured the light spot images of the corresponding markers. The collected laser reflection signals underwent baseline correction and segmented filtering. Taking one marker as an example, the corrected original laser reflection signal was divided into low, medium, and high frequency bands after Fourier transform, and bandpass filtering was used to suppress high-frequency noise. The filtered signal was converted back to the time domain. Based on material property analysis, environmental factors, and experiments, four sub-bands were determined: 400-500nm, 500-600nm, 600-700nm, and 700-800nm. The target spectrum was divided according to the preset bands, and the calculated spectral difference gradient sequences were 0.2, 0.15, and 0.18. Multi-scale Gaussian pyramid decomposition was performed on the spot image. The ellipse fitting parameters of the spot at point 1 showed that the major axis was 5 pixels, the minor axis was 4.8 pixels, and the centroid offset was (0.2, 0.3) pixels, forming a multi-scale set feature vector.

[0037] After standardizing the spectral differential gradient sequence and multi-scale ensemble feature vector, the Pearson correlation coefficient formula was used to calculate the spectral-geometric coupling consistency index, yielding a correlation coefficient of 0.85 for this marker point. Using min-max normalization and weighted fusion, the reliability index of this marker point was obtained as 0.78. Repeating the above process for 10 marker points, statistical analysis revealed that marker points with a reliability index higher than 0.7 had an average positioning error of 0.5 mm in subsequent track monitoring, while marker points with a reliability index lower than 0.5 had an average positioning error of 2 mm. This demonstrates that the marker point reliability index can effectively quantify the reliability of marker points and improve the effectiveness and accuracy of track monitoring.

[0038] Calculating the reliability coefficient of track markers can quantify their reliability, thereby improving the effectiveness and accuracy of track monitoring and ensuring the long-term stable and safe operation of railway tracks.

[0039] The true spatial coordinates of the track markers are obtained, and the relative spatial coordinates collected in real time are read. After adjusting the track marker weights based on the marker confidence index, the lateral displacement, vertical displacement, torsion angle and track gauge changes of the track during the current monitoring period are calculated to form a multi-attitude parameter vector of the track.

[0040] Furthermore, the lateral displacement, vertical displacement, torsion angle, and gauge changes of the track during the current monitoring period are calculated to form a multi-attitude parameter vector of the track. This includes: constructing a multi-point cooperative attitude model that includes the neighborhood spatial relationships of the track markers, based on the real spatial coordinates of each track marker; generating a model matrix for attitude calculation based on the multi-point cooperative attitude model; constructing an adaptive weighted matrix using the marker confidence index of each track marker; performing model matrix correction; inputting the real-time collected relative spatial coordinates into the corrected model matrix; and obtaining the three-dimensional rigid body transformation matrix of the track during the current monitoring period through weighted least squares or iterative optimization; and using the multi-point cooperative attitude model and the three-dimensional rigid body transformation matrix to analyze the lateral displacement, vertical displacement, torsion angle, and gauge changes to form a multi-attitude parameter vector of the track.

[0041] Specifically, high-precision measuring equipment, such as total stations and GNSS, is used to acquire the true spatial coordinates of track marker points. These true position coordinates refer to the precise coordinate values ​​of the track marker points within a pre-defined and fixed global coordinate system. They accurately reflect the actual geographical location of the track marker points in real three-dimensional space. Without considering changes in actual position caused by abnormal conditions such as track deformation or displacement, these coordinates are objectively existent and unchanging, providing an absolute standard reference for track monitoring. The global coordinate system is constructed using a fixed point on the railway line, such as the starting point or a specific mileage marker, with the positive x-axis representing the direction of line movement, the positive y-axis representing the direction perpendicular to the line, and the positive z-axis representing vertically upward. Simultaneously, the relative spatial coordinates of each marker point are collected in real time. These relative spatial coordinates are based on the true spatial coordinates of the track marker points, and the offset or change in coordinates of the marker points relative to the reference position measured during monitoring reflects the displacement and attitude changes of the track in real time. Using the actual spatial coordinates of each track marker point as a reference, a topological structure of the marker points is established according to the track geometry and the spatial neighborhood relationships between them. This involves establishing connections between each marker point and its neighbors, forming a neighborhood matrix to characterize spatial connectivity and relative positional constraints. Based on the neighborhood matrix, the relative distances, angles, and track node order between marker points are calculated, constructing a multi-point cooperative attitude model that includes the neighborhood spatial relationships of the track marker points. This model reflects the spatial connectivity, relative positional relationships, and structural constraints between the marker points. For example, in railway tracks, there are fixed geometric constraints between adjacent marker points; by constructing a multi-point cooperative attitude model, the accuracy of track attitude calculation can be improved.

[0042] Based on a multi-point cooperative attitude model, the spatial coordinate relationships of marker points are transformed into a system of linear equations to generate a model matrix for attitude calculation. For example, based on the neighborhood spatial relationships of the multi-point cooperative attitude model, the true spatial coordinates and neighborhood spatial constraints of each track marker point are converted into constraint equations, including the relative position, distance, and angle constraints between marker points, forming linear equations. Following the marker point order and coordinate dimensions, the equations of all marker points are synthesized to generate a model matrix for attitude calculation. Each row of the model matrix corresponds to a constraint equation, and each column corresponds to the three-dimensional coordinates and attitude variables to be solved. An adaptive weighted matrix is ​​constructed using the marker point confidence index of each track marker point. This adaptive weighted matrix uses the confidence level of each marker point as a weight coefficient. The adaptive weighted matrix is ​​applied to the model matrix, and through weighted calculation, the model matrix is ​​corrected so that the corrected model matrix can more accurately reflect the actual attitude of the track.

[0043] The real-time acquired relative spatial coordinates are input into the corrected model matrix. The model matrix is ​​then solved using weighted least squares or iterative optimization methods to obtain the three-dimensional rigid body transformation matrix of the track during the current monitoring period. This matrix reflects the overall translational and rotational angle changes of the track. For example, by inputting the relative spatial coordinates into the corrected model matrix and using the weighted least squares formula, the parameters of the three-dimensional rigid body transformation matrix are obtained by minimizing the sum of squared residuals. Alternatively, iterative optimization methods, such as the Gauss-Newton iterative algorithm, can be used to progressively update the estimated values ​​until a convergence threshold or the maximum number of iterations is reached, obtaining the translational and rotational parameters to form the three-dimensional rigid body transformation matrix. Using a multi-point cooperative attitude model and the three-dimensional rigid body transformation matrix, the lateral displacement, vertical displacement, torsion angle, and gauge changes of the track during the current monitoring period are analyzed. The lateral displacement reflects the track's horizontal offset, the vertical displacement reflects its vertical rise and fall, the torsion angle reflects the track's rotation angle, and the gauge change reflects the distance change between the rails on both sides of the track. Based on the changes in lateral displacement, vertical displacement, torsion angle, and track gauge, a multi-attitude parameter vector of the track is obtained, which comprehensively reflects the spatial geometric state of the track at the current monitoring time.

[0044] For example, the multi-attitude parameter vector of the track obtained during the current monitoring period is analyzed. This vector contains key attitude parameters such as lateral displacement, vertical displacement, torsion angle, and gauge change, reflecting the dynamic geometric state changes of the track over time. For instance, during the current monitoring period, the lateral displacement recorded by the multi-attitude parameter vector is 1mm, 1.2mm, and 1.5mm, the vertical displacement is 0.8mm, 0.9mm, and 1mm, the torsion angle is 0.1°, 0.12°, and 0.15°, and the gauge change is 0.2mm, 0.25mm, and 0.3mm. The attitude parameters are sorted chronologically to form an analytical feature set of track motion, which serves as the basis for prediction. For each track marker point, autoregressive and moving average models are used to fit the stationary data to predict future data values. Taking track marker point 1 as an example, analytical feature data from its past five monitoring times are collected and input into the autoregressive and moving average models for training, constructing a time-series fitting prediction model for this track marker point.

[0045] The monitoring interval for the next monitoring time is obtained. The monitoring interval refers to the time interval between the current monitoring time and the next monitoring time, acquired through a preset monitoring plan or a real-time time synchronization mechanism. For example, if the monitoring plan specifies monitoring every 2 hours, then the monitoring interval is 2 hours. This acquired monitoring interval is used as auxiliary data and sent to the pre-built time-series fitting prediction model for each track marker point. Based on the input monitoring interval and the analytical feature set corresponding to the multi-attitude parameter vector of the track at the current monitoring time, the time-series fitting prediction model outputs the predicted position of each track marker point at the next monitoring time. Through analysis, the predicted lateral displacement is 1.8 mm, the vertical displacement is 1.1 mm, the torsion angle is 0.18°, and the track gauge change is 0.35 mm at the next monitoring time, thus determining its predicted position coordinates. Using measuring equipment, such as a total station, the actual observation position at the next monitoring time is collected. The actual observation shows that the lateral displacement of the track marker point is 1.7 mm, the vertical displacement is 1.05 mm, the torsion angle is 0.17°, and the track gauge change is 0.33 mm, obtaining the actual observation position coordinates. The residual between the actual and predicted positions at the next monitoring time is calculated using the Euclidean distance formula. The distance between the actual and predicted coordinates is used as the residual comparison result. The calculated residual between the actual and predicted positions of this track marker is 0.12 mm. The reliability index of this track marker is known to be 0.85. A minimum residual threshold of 0.1 mm and a maximum residual threshold of 0.2 mm are set. Since the residual comparison result of 0.12 mm falls between the minimum and maximum residual thresholds of 0.1 mm and 0.2 mm, the track marker is updated to a suspicious state according to the rules. By dynamically updating the marker status based on the residual comparison results and the reliability index, the working status of the track marker can be grasped in a timely manner, ensuring the accuracy and reliability of track monitoring. By utilizing the real spatial coordinates, real-time relative spatial coordinates, and reliability index of the marker, the accurate and comprehensive acquisition of track attitude parameters is improved, thereby enhancing the accuracy and reliability of railway track monitoring.

[0046] Furthermore, the calculation of the lateral displacement, vertical displacement, torsion angle, and gauge changes of the track during the current monitoring period to form a multi-attitude parameter vector of the track also includes: constructing a track node topology matrix based on the real spatial coordinates of the track markers, wherein the track node topology matrix represents the spatial connectivity, neighborhood order, and structural constraint relationships between the track nodes; constructing a historical motion trend vector for each track marker using historical monitoring data; calculating the attitude calculation weights for each track marker based on the track node topology matrix and the historical motion trend vector; and optimizing the model matrix of the multi-point cooperative attitude model based on the calculation results to establish the multi-attitude parameter vector of the track.

[0047] Specifically, using the actual spatial coordinates of track markers as a reference, the connectivity between markers is analyzed based on the track structure to determine the neighboring markers of each marker. For example, on a straight track, adjacent markers are neighbors. Following the marker numbering order, the spatial connectivity, neighborhood order, and structural constraints between markers are represented in matrix form, constructing a track node topology matrix. The rows and columns of this matrix correspond to track markers, and the matrix elements represent the relationships between markers. Spatial connectivity refers to whether there is a direct connection or association between markers on the track, reflecting the physical continuity of the track structure—that is, whether one marker can be directly reached from another via the track. Neighborhood order refers to the arrangement order of markers in the surrounding area of ​​a track marker, reflecting the relative positional information between markers. Structural constraints refer to the inherent conditions and rules of the track structure that restrict the position and movement of markers on the track, such as the curve radius constraint of the track and the distance constraint between tracks.

[0048] Spatial coordinate data of track markers over multiple monitoring periods are collected using long-term monitoring equipment, such as a track monitoring sensor network. For each marker, the coordinate data at different times is then processed and analyzed. Time series analysis methods, such as the autoregressive moving average (ARMA) model, are used to fit the changes in the coordinates of the track markers over time, yielding their motion trend equation. Based on this equation, characteristic parameters reflecting the historical motion trend of the markers, such as the rate of change of displacement and acceleration, are extracted to form the historical motion trend vector of the track.

[0049] Based on the track node topology matrix, the spatial connectivity, neighborhood order, and structural constraints among marker points are analyzed. Higher topology weights are assigned to adjacent marker points or those significantly impacting the track structure. For example, on curved tracks, marker points adjacent to key control points of the curve have a greater impact on the overall track shape due to their positional changes, thus receiving higher topology weights. In turnout areas, the topology weights of interconnected marker points on different turns are determined based on their connection relationships and their roles in turnout switching. Simultaneously, the motion stability and trend consistency of track marker points are analyzed using historical motion trend vectors. Marker points with clear motion patterns and stable trends are assigned higher motion trend weights. For example, a track marker point exhibiting uniform displacement and stable acceleration in historical monitoring indicates reliable motion and warrants greater weight in attitude calculation. The topology weights and motion trend weights are then weighted and fused using methods such as the Analytic Hierarchy Process (AHP) or entropy weighting to obtain the attitude calculation weight for each marker point. The model matrix of the multi-point collaborative attitude model is optimized based on the attitude calculation weight of each marker point. During the optimization process, weighted least squares or iterative optimization methods are used to solve the problem. The lateral displacement, vertical displacement, torsion angle and track gauge changes of the track during the current monitoring period are obtained analytically, forming a multi-attitude parameter vector of the track. The multi-attitude parameter vector of the track can comprehensively and accurately reflect the attitude changes of the track during the current monitoring period.

[0050] By introducing track topology constraints and historical motion trends, the accuracy of track attitude calculation is further improved, providing reliable data support for track condition prediction and anomaly monitoring, thereby enhancing the effectiveness and accuracy of railway track monitoring.

[0051] Furthermore, an adaptive weighted matrix is ​​constructed using the confidence index of each track marker point, including: mapping the confidence index of each track marker point to an initial weight, wherein the initial weight is used to reflect the contribution of each marker point to the attitude solution; obtaining the coordinate residual between the relative spatial coordinates and the real spatial coordinates, configuring auxiliary weight factors based on the coordinate residuals, performing initial weight compensation using the auxiliary weight factors to construct compensated weights; performing confidence classification on the compensated weights, configuring confidence labels, and performing neighborhood collaborative weighting processing using the confidence labels to construct an adaptive weighted matrix.

[0052] Specifically, a confidence index for each orbital marker is obtained, and this confidence index is mapped to an initial weight. The initial weight reflects the contribution of each marker to the attitude calculation. For example, the confidence index is linearly normalized so that it is directly used as the initial weight, ensuring that high-confidence markers correspond to larger weights and low-confidence markers correspond to smaller weights.

[0053] The coordinate residuals between the relative spatial coordinates and the true spatial coordinates are calculated. The coordinate reference reflects the degree of deviation between the relative spatial coordinates and the true spatial coordinates of the marker point. To optimize the stability of the solution, auxiliary weighting factors are configured for compensation based on the coordinate residuals. For example, the magnitude of the coordinate residuals is calculated and compared with a reference magnitude threshold. The magnitude threshold is obtained empirically, for example, 5 mm. When the magnitude of the coordinate residuals is less than or equal to the magnitude threshold, the auxiliary weighting factor is slightly increased by 0.05-0.1. For track marker points with larger residuals, the weight is appropriately reduced by 0.03-0.1. The auxiliary weighting factor will not exceed 0.1 to avoid affecting the final solution result. The initial weights are compensated using the configured auxiliary weighting factor, and the auxiliary weighting factor and the initial weights are added together to form the compensated weights.

[0054] The compensation weights are classified into trustworthy categories, with different classification thresholds set according to the magnitude of the compensation weights, and trustworthy labels are configured. Trustworthy labels include three categories: high trustworthiness, medium trustworthiness, and low trustworthiness. For example, thresholds T1 and T2 are set, where T1 is less than T2. ​​If the compensation weight is greater than or equal to T2, it is marked as high trustworthiness; if the compensation weight is between T1 and T2, it is marked as medium trustworthiness; and if the compensation weight is less than T1, it is marked as low trustworthiness. Neighborhood collaborative weighting is performed using trustworthy labels. That is, for each marker point, neighborhood collaborative weighting is performed based on the trustworthiness of its neighboring marker points. For example, a distance-based approach is used, setting a fixed distance threshold to determine the neighborhood range of each track marker point, counting the trustworthy labels of each track marker point within the neighborhood, and assigning corresponding weight contribution values ​​based on different trustworthy labels. For example, high trustworthiness labels correspond to larger contribution values, and low trustworthiness labels correspond to smaller contribution values. The average of the weight contribution values ​​of all marker points within the neighborhood is calculated as the neighborhood collaborative factor. The compensation weight of each marker point is multiplied by the corresponding neighborhood co-factor to obtain the final weighted value of the marker point. The final weighted values ​​are arranged in the order of the marker point numbers to construct an adaptive weighted matrix.

[0055] By employing reliable classification and neighborhood collaborative weighting, the constructed adaptive weighting matrix can fully utilize the spatial relationships and reliability information between orbit markers, thereby enabling attitude calculation to more accurately reflect the actual attitude changes of the orbit and effectively improve the accuracy and stability of attitude calculation.

[0056] The predicted positions of each track marker point at the next monitoring time are predicted based on the track multi-attitude parameter vector. The status of the marker points is dynamically updated by comparing the residual between the actual observed position and the predicted position at the next monitoring time and the confidence index.

[0057] Furthermore, based on the orbital multi-attitude parameter vector, the predicted positions of each orbital marker point at the next monitoring time are predicted. The marker point states are dynamically updated using the residual comparison results between the actual observed positions and the predicted positions at the next monitoring time and the confidence index. This includes: parsing the orbital multi-attitude parameter vector, using the parsed feature set as basic data, and constructing a time-series fitting prediction model for each orbital marker point; obtaining the monitoring interval for the next monitoring time, using the monitoring interval as auxiliary data, sending it to the time-series fitting prediction model, and outputting the predicted positions of each orbital marker point; calculating the residuals using the actual observed positions and the predicted positions, generating residual comparison results, and dynamically updating the marker point states using the residual comparison results and the confidence index.

[0058] Specifically, the multi-attitude parameter vector of the track obtained during the current monitoring period is analyzed. This vector includes key attitude parameters such as lateral displacement, vertical displacement, torsion angle, and gauge change, reflecting the dynamic geometric state changes of the track over time. By sorting these attitude parameters in chronological order, an analytical feature set of track motion is formed, serving as the basis for prediction. For each track marker, a corresponding time-series fitting prediction model is constructed based on the basis data. For example, an autoregressive integral sliding model is used to stabilize historical data through differencing. The stabilized data is then fitted using autoregressive and moving average models to predict future data values. For each marker, its historical analytical feature data is collected and input into the autoregressive integral sliding model for training, constructing a time-series fitting prediction model for each track marker.

[0059] The monitoring interval for the next monitoring time is obtained. This monitoring interval refers to the time interval between the current monitoring time and the next monitoring time, and is obtained through a preset monitoring plan or a real-time time synchronization mechanism. For example, if the monitoring plan specifies monitoring every hour, then the monitoring interval is one hour. The obtained monitoring interval is used as auxiliary data and sent to the pre-constructed temporal fitting prediction model for each orbital marker. Based on the input monitoring interval and the analytical feature set corresponding to the orbital multi-attitude parameter vector at the current monitoring time, the temporal fitting prediction model outputs the predicted position of each orbital marker at the next monitoring time.

[0060] The actual observation position at the next monitoring time is collected using measuring equipment, such as a total station. The residual between the actual observation position and the predicted position at the next monitoring time is calculated. The residual calculation can use the Euclidean distance formula to calculate the distance between the actual observation position coordinates and the predicted position coordinates as the residual comparison result. This residual comparison result reflects the degree of deviation between the current prediction and the actual measurement. The status of the track marker is dynamically updated based on the residual comparison result and the confidence index. The track marker status represents the current working status of the track marker. For example, when the residual comparison result is less than a smaller residual threshold and the confidence index is high, the track marker is considered to be in a normal state and its valid status is maintained. When the residual comparison result is between a smaller residual threshold and a larger residual threshold, the track marker is updated to a suspicious state. When the residual comparison result is greater than a larger residual threshold and the confidence index is low, the track marker is updated to an abnormal state.

[0061] By utilizing historical attitude data and real-time monitoring information of the orbit, the future position of the orbit marker can be accurately predicted, and the residual comparison results between the two can be calculated. Combined with the credibility index, the status of the marker can be dynamically updated, which can promptly detect anomalies of the marker and realize real-time monitoring and dynamic evaluation of the orbit status.

[0062] The system collects track vibration signals, extracts resonant energy spectrum features, filters track multi-attitude parameter vectors based on updated marker point states, performs inversion suppression of the coupling relationship with resonant energy spectrum features, and outputs monitoring anomalies based on the inversion suppression results and track multi-attitude parameter vectors.

[0063] Furthermore, the inversion suppression of the coupling relationship with the resonant energy spectrum characteristics is performed, including: constructing a mapping relationship between the screening results and the resonant energy spectrum characteristics, including: a: mapping the temporal changes of the orbital multi-attitude parameter vectors of each orbital marker point to the corresponding frequency band resonant energy spectrum amplitude, frequency drift, and energy distribution changes; b: constructing a high-dimensional coupling matrix to characterize the multi-point, multi-frequency band coupling relationship between orbital attitude and vibration characteristics; suppressing the anomalous energy spectrum components of the high-dimensional coupling matrix based on an adaptive attenuation strategy, and establishing the inversion suppression results.

[0064] Specifically, vibration sensors deployed along the railway track collect track vibration signals in real time. The time-domain track vibration signals are converted into frequency-domain signals using Fast Fourier Transform (FFT) or time-frequency analysis methods, yielding the spectral distribution of the track vibration signals. From this spectral distribution, the resonant energy spectrum characteristics of the track vibration signals are extracted. These characteristics refer to information such as the energy distribution, amplitude, and frequency drift of the track vibration signals in the frequency domain, reflecting the dynamic response characteristics of the track structure at different frequency bands. Specifically, changes in energy distribution reflect the distribution of vibration energy across different frequency bands, the amplitude of the resonant energy spectrum represents the magnitude of the vibration energy, and the frequency drift reflects changes in the vibration frequency.

[0065] Based on the updated marker point status, the orbital multi-attitude parameter vectors are filtered to obtain the orbital multi-attitude parameter vectors corresponding to the current marker point status. Based on the filtering results, a mapping relationship is established between the orbital multi-attitude parameter vectors and the characteristics of the resonant energy spectrum. Specifically, the temporal changes of the orbital multi-attitude parameter vectors at each orbital marker point are mapped to the amplitude, frequency drift, and energy distribution changes of the resonant energy spectrum in the corresponding frequency band. For example, using polynomial fitting in data fitting techniques, the temporal data of the orbital multi-attitude parameter vectors are used as independent variables, and the temporal data of the amplitude, frequency drift, and energy distribution changes of the resonant energy spectrum in the corresponding frequency band are used as dependent variables to construct a mapping relationship model. For example, for the temporal changes of the orbital multi-attitude parameter vector p(t)=[p1(t),p2(t),…,pn(t)]T at each orbital marker point, a multivariate polynomial regression model can be established to map it to the amplitude A(f,t), frequency drift Δf(t), and energy distribution E(f,t) of the resonant energy spectrum in the corresponding frequency band. The specific mathematical model is as follows: ,in, These are the characteristic time-series values ​​of the resonant energy spectrum for the corresponding frequency band. Here, m represents the regression coefficients, and m is the polynomial order, typically between 2 and 4, which can be selected based on the fitting error and cross-validation. This represents the residual term. Using historically acquired orbital multi-attitude time-series data and corresponding frequency band resonant energy spectrum characteristic time-series data, a least-squares fitting method is employed to solve for the polynomial regression coefficients. The goal of the least-squares method is to minimize the sum of squared residuals. By processing the new orbital multi-attitude parameter vector time-series data using a mapping relationship model, the temporal changes of the orbital multi-attitude parameter vectors can be mapped to the corresponding frequency band resonant energy spectrum amplitude, frequency drift, and energy distribution changes, thus achieving temporal correlation analysis between attitude and vibration.

[0066] Based on the mapping relationship, a high-dimensional coupling matrix is ​​constructed to represent the multi-point, multi-frequency coupling relationship between orbital attitude and vibration characteristics. In this matrix, rows represent different orbital marker points, columns represent different frequency bands, and each element represents the coupling strength between the orbital attitude and vibration characteristics at the corresponding marker point within the corresponding frequency band. For example, by calculating the correlation coefficient between the resonant energy spectrum characteristics of different marker points at different frequency bands and the orbital multi-attitude parameter vectors, this correlation coefficient is used as an element in the high-dimensional coupling matrix. The correlation coefficient can be calculated based on the Pearson correlation coefficient, using the following formula: ,in, The timing sequence of the multi-attitude parameter vector for the i-th orbit marker point. Let cov be the resonant energy spectrum characteristic of the corresponding frequency band j, and let cov be the covariance. and These are the standard deviations of the corresponding variables.

[0067] The high-dimensional coupling matrix is ​​analyzed to identify anomalous energy spectrum components. These anomalous components may be caused by factors such as orbital faults or external interference, and their characteristics may include abnormally increased amplitude, frequency drift exceeding the normal range, or abnormally concentrated energy distribution. Statistical methods, such as setting thresholds, are employed. First, the mean μ and standard deviation σ of the resonant energy spectrum characteristics of each frequency band, including historical data of amplitude, frequency drift, or energy distribution under normal conditions, are calculated. Then, a threshold is set, for example, an amplitude threshold A. th =μA+kσA, where k can be 2-3, corresponding to a 95%–99% confidence interval. Frequency drift thresholds and energy distribution thresholds are set in the same way. During judgment, when the amplitude, frequency drift, or energy distribution of an energy spectrum component exceeds the threshold, it is judged as an abnormal energy spectrum component. An adaptive attenuation strategy is used to suppress the identified abnormal energy spectrum components, dynamically reducing the influence weight of abnormal frequency bands or abnormal attitude components in the overall solution based on the degree of abnormality. The adaptive attenuation strategy automatically adjusts the attenuation coefficient based on the characteristics of the abnormal energy spectrum component, such as amplitude and frequency drift, to effectively suppress the abnormal energy spectrum component. Its attenuation formula can be expressed as:

[0068] ,

[0069] ,in, ∈(0,1] is the attenuation coefficient, which varies with the abnormal amplitude A ij Frequency drift or energy distribution E ij The increase automatically decreases. A represents the anomalous energy spectrum component in the high-dimensional coupling matrix. th f th E th These represent the amplitude, frequency, and energy distribution thresholds, respectively. max f max E max These represent the maximum values ​​of amplitude, frequency drift, and energy distribution in the matrix, respectively. , , These are weighting parameters for adjusting the attenuation rate, which can be dynamically set according to the track vibration characteristics, such as 1.5, 1.0, and 1.0 respectively. For example, a larger attenuation coefficient is used for rapid suppression of anomalous energy spectrum components with large amplitudes, while a smaller attenuation coefficient is used for slow suppression of anomalous energy spectrum components with small frequency drifts, to avoid excessive influence on normal signals. By continuously adjusting the attenuation coefficient, the anomalous energy spectrum components are gradually attenuated to the normal range, thereby establishing the inversion suppression result. The inversion suppression result refers to the high-dimensional coupling matrix obtained after suppressing the anomalous energy spectrum components, which can accurately reflect the railway track anomaly information. Based on the inversion suppression result and the track multi-attitude parameter vector, the monitoring anomalies of track structure anomalies and geometric deformations are output.

[0070] By constructing a mapping relationship with the resonant energy spectrum characteristics, the intrinsic connection between track attitude and vibration characteristics can be analyzed in depth, providing a basis for accurately judging track status. Further suppression of anomalous energy spectrum components effectively eliminates interference factors, improving the accuracy and reliability of track monitoring data. Furthermore, based on the inversion suppression results and the output of track multi-attitude parameter vectors to monitor anomalies, problems on the track can be detected in a timely manner, achieving precise identification and accurate location of track anomalies. This improves the real-time performance, accuracy, and reliability of track anomaly monitoring, thereby enhancing the level of railway track monitoring and train operation safety.

[0071] Furthermore, based on the inversion suppression results and the track multi-attitude parameter vector, the monitoring anomalies are output, including: configuring anomaly warning signals according to the anomaly indicators of the monitored anomalies; using the anomaly warning signals to perform anomaly reporting and simultaneously generating monitoring of concern; and performing continuous monitoring and management of railway tracks according to the monitoring of concern.

[0072] Specifically, the inversion suppression results are comprehensively analyzed with the multi-attitude parameter vectors of the track. The lateral displacement, vertical displacement, torsion angle, and gauge changes of each track marker point are compared with vibration anomaly characteristics. When these exceed preset thresholds or anomaly patterns, anomaly indicators are generated. Anomaly indicators are categorized according to their severity, and corresponding types and intensities of anomaly warning signals are configured for different levels. For example, the anomaly indicators of each track marker point are compared with preset thresholds, and levels are categorized based on the degree of exceedance and historical experience, including minor anomalies, moderate anomalies, and severe anomalies. Minor anomalies are below 1 times the threshold, moderate anomalies are 1–2 times the threshold, and severe anomalies are above 2 times the threshold. The thresholds can be dynamically set based on actual needs, historical data, and expert experience. Corresponding types and intensities of anomaly warning signals are assigned to different levels. For example, minor anomalies generate low-intensity alerts displayed on the monitoring interface or in the log; moderate anomalies trigger audible and visual alarms and generate task tracking records; and severe anomalies trigger emergency reporting, real-time notification to maintenance personnel, and lock down areas of concern for focused monitoring. The anomaly warning signals trigger the anomaly reporting function, promptly issuing anomaly alarms to relevant personnel. Simultaneously, it generates monitoring tasks containing information such as the location and type of anomalies, used to mark track sections or points requiring focused tracking. Utilizing this monitoring information, continuous railway track monitoring and management are implemented, including periodic re-inspections, data recording, trend analysis, and maintenance scheduling recommendations. This achieves dynamic tracking and closed-loop management of track anomalies, thereby ensuring track operation safety and the scientific basis of maintenance decisions.

[0073] Furthermore, the method also includes: using track vibration signals to identify vibration anomalies, establishing vibration anomaly triggering early warning signals, and performing early warning and dispatch management of vibration anomaly triggering early warning signals.

[0074] Specifically, after acquiring the track vibration signal, the extracted vibration features, such as the amplitude of the resonant energy spectrum, frequency drift, and changes in energy distribution, are compared with the benchmark thresholds under normal operating conditions. When the amplitude of the resonant energy spectrum, frequency drift, and energy distribution exceed the preset range, it is determined to be a vibration anomaly. Based on the vibration anomaly, a vibration anomaly trigger warning signal is generated, and corresponding reporting methods are configured according to the severity of the anomaly. For example, minor anomalies are logged and the section of concern is marked; moderate anomalies trigger monitoring interface prompts and maintenance notifications; and severe anomalies trigger real-time alarms and initiate emergency maintenance procedures, realizing early warning and reporting management.

[0075] By analyzing track vibration signals, we can quickly identify and define track vibration anomalies, provide graded early warnings, and continuously monitor and manage them, thereby further improving track structure safety and operational stability.

[0076] Example 2, based on the same inventive concept as the laser positioning and monitoring method for railway track markers in the foregoing examples, such as... Figure 2As shown, this application provides a laser positioning and monitoring system for marker points on railway tracks, wherein the laser positioning and monitoring system for marker points on railway tracks includes:

[0077] The reliability calculation module 11 is used to collect laser reflection signals from track markers, extract the differential characteristics of the reflection spectrum and the geometric distribution characteristics of the laser spots, construct a dual-modal feature set of the marker's spectrum and geometry, and calculate the marker's reliability index using the dual-modal feature set. The attitude parameter acquisition module 12 is used to acquire the real spatial coordinates of the track markers and read the real-time acquired relative spatial coordinates. After adjusting the track marker weights based on the marker reliability index, it calculates the lateral displacement, vertical displacement, torsion angle, and gauge changes of the track during the current monitoring period, forming a multi-attitude parameter vector of the track. The state update module 13 is used to predict the predicted position of each track marker at the next monitoring time based on the multi-attitude parameter vector of the track. It dynamically updates the marker state based on the comparison result of the actual observed position and the predicted position residual at the next monitoring time and the reliability index. The monitoring anomaly output module 14 is used to collect track vibration signals, extract the resonant energy spectrum features, filter the multi-attitude parameter vector of the track based on the updated marker state, perform inversion suppression of the coupling relationship with the resonant energy spectrum features, and output the monitoring anomaly based on the inversion suppression result and the multi-attitude parameter vector of the track.

[0078] Furthermore, the credibility calculation module 11 is also used to: perform baseline correction and segmented filtering on the laser reflection signal, divide the target spectrum into multiple sub-bands according to a preset band, and calculate the spectral difference gradient sequence of adjacent sub-bands; perform multi-scale Gaussian pyramid decomposition on the spot image of the track marker point, extract the ellipse fitting parameters, centroid offset and edge gradient distribution of the spot at each scale, and form a multi-scale set feature vector; use the spectral difference gradient sequence and the multi-scale set feature vector as a bimodal feature set, calculate the spectral-geometric coupling consistency index, and use the coupling consistency index to perform joint normalization and weighted fusion of spectral feature components and geometric feature components, and output the marker point credibility index.

[0079] Furthermore, the attitude parameter acquisition module 12 is also used to: construct a multi-point cooperative attitude model containing the neighborhood spatial relationship of each track marker point based on the real spatial coordinates of each track marker point; generate a model matrix for attitude calculation based on the multi-point cooperative attitude model; construct an adaptive weighted matrix using the marker point confidence index of each track marker point; and perform model matrix correction; input the real-time collected relative spatial coordinates into the corrected model matrix; and obtain the three-dimensional rigid body transformation matrix of the track during the current monitoring period by solving the problem through weighted least squares or iterative optimization; and analyze the lateral displacement, vertical displacement, torsion angle, and track gauge changes using the multi-point cooperative attitude model and the three-dimensional rigid body transformation matrix to form a multi-attitude parameter vector of the track.

[0080] Furthermore, the attitude parameter acquisition module 12 is also used to: construct a track node topology matrix based on the real spatial coordinates of the track marker points, wherein the track node topology matrix represents the spatial connectivity, neighborhood order, and structural constraint relationships between each track node; construct a historical motion trend vector for each track marker point using historical monitoring data; calculate the attitude calculation weights for each track marker point based on the track node topology matrix and the historical motion trend vector; and optimize the model matrix of the multi-point cooperative attitude model according to the calculation results to establish a track multi-attitude parameter vector.

[0081] Furthermore, the attitude parameter acquisition module 12 is also used to: map the confidence index of each track marker point to an initial weight, the initial weight being used to reflect the contribution of each marker point to the attitude solution; obtain the coordinate residual between the relative spatial coordinates and the real spatial coordinates, configure an auxiliary weight factor based on the coordinate residual, perform initial weight compensation using the auxiliary weight factor, and construct a compensation weight; perform a confidence classification on the compensation weight, configure a confidence label, and perform neighborhood collaborative weighting processing using the confidence label to construct an adaptive weighting matrix.

[0082] Furthermore, the state update module 13 is also used to: parse the orbital multi-attitude parameter vector, use the parsed feature set as basic data, and construct a time-series fitting prediction model for each orbital marker point; obtain the monitoring interval for the next monitoring time, use the monitoring interval as auxiliary data, send it to the time-series fitting prediction model, and output the predicted position of each orbital marker point; calculate the residual using the actual observed position and the predicted position, generate a residual comparison result, and dynamically update the marker point state using the residual comparison result and the confidence index.

[0083] Furthermore, the monitoring anomaly output module 14 is also used to: construct a mapping relationship between the screening results and the resonant energy spectrum characteristics, including: a: mapping the temporal changes of the orbital multi-attitude parameter vectors of each orbital marker point to the resonant energy spectrum amplitude, frequency drift, and energy distribution changes of the corresponding frequency band; b: constructing a high-dimensional coupling matrix to characterize the multi-point, multi-frequency coupling relationship between orbital attitude and vibration characteristics; suppressing the abnormal energy spectrum components of the high-dimensional coupling matrix based on an adaptive attenuation strategy, and establishing the inversion suppression result.

[0084] Furthermore, the monitoring anomaly output module 14 is also used to: configure an anomaly warning signal according to the anomaly index of the monitored anomaly; use the anomaly warning signal to perform anomaly reporting and simultaneously generate attention monitoring; and perform continuous monitoring management of railway tracks according to the attention monitoring.

[0085] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The laser positioning and monitoring method for railway track markers in the foregoing embodiment one is also applicable to the laser positioning and monitoring system for railway track markers in this embodiment. Through the foregoing detailed description of the laser positioning and monitoring method for railway track markers, those skilled in the art can clearly understand the laser positioning and monitoring system for railway track markers in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0086] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0087] Obviously, those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A laser positioning and monitoring method for marker points on railway tracks, characterized in that, The method includes: By collecting the laser reflection signals of the track markers, extracting the differential characteristics of the reflection spectrum and the geometric distribution characteristics of the laser spots, constructing a dual-modal feature set of the markers' spectrum and geometry, and using the dual-modal feature set to calculate the marker confidence index; The true spatial coordinates of the track markers are obtained, and the relative spatial coordinates collected in real time are read. After adjusting the weight of the track markers based on the confidence index of the markers, the lateral displacement, vertical displacement, torsion angle and track gauge changes of the track during the current monitoring period are calculated to form a multi-attitude parameter vector of the track. Based on the orbital multi-attitude parameter vector, predict the predicted position of each orbital marker point at the next monitoring time, and dynamically update the marker point status by comparing the residual between the actual observed position and the predicted position at the next monitoring time and the confidence index. The system collects track vibration signals, extracts resonant energy spectrum features, filters track multi-attitude parameter vectors based on updated marker point states, performs coupling relationship inversion suppression with resonant energy spectrum features, and outputs monitoring anomalies based on inversion suppression results and track multi-attitude parameter vectors. The confidence index of the marker points is calculated using the bimodal feature set, including: The laser reflection signal is baseline corrected and segmented filtered, the target spectrum is divided into multiple sub-bands according to a preset band, and the spectral difference gradient sequence of adjacent sub-bands is calculated. Multi-scale Gaussian pyramid decomposition is performed on the spot image of the track marker point. Ellipse fitting parameters, centroid offset and edge gradient distribution of the spot are extracted at each scale to form a multi-scale set feature vector. The spectral differential gradient sequence and the multi-scale set feature vector are used as a bimodal feature set. The spectral-geometric coupling consistency index is calculated. The spectral feature components and geometric feature components are jointly normalized and weighted using the coupling consistency index to output the marker point confidence index. Perform inversion suppression based on the coupling relationship with the resonant energy spectrum characteristics, including: Based on the screening results, a mapping relationship between the results and the characteristics of the resonant energy spectrum is constructed, including: a: Map the temporal changes of the orbital multi-attitude parameter vectors at each orbital marker point to the corresponding frequency band's resonant energy spectrum amplitude, frequency drift, and energy distribution changes; b: Construct a high-dimensional coupling matrix to characterize the multi-point, multi-frequency coupling relationship between orbital attitude and vibration characteristics; The anomalous energy spectrum components of the high-dimensional coupling matrix are suppressed based on an adaptive attenuation strategy, and the inversion suppression results are established.

2. The laser positioning and monitoring method for marker points on railway tracks as described in claim 1, characterized in that, The system calculates the lateral displacement, vertical displacement, torsion angle, and gauge changes of the track during the current monitoring period, forming a multi-attitude parameter vector of the track, including: Based on the real spatial coordinates of each track marker point, a multi-point cooperative attitude model including the neighborhood spatial relationship of the track marker points is constructed. Based on the multi-point cooperative attitude model, a model matrix for attitude calculation is generated. After constructing an adaptive weighted matrix using the confidence index of each track marker point, the model matrix is ​​corrected and modified. The real-time collected relative spatial coordinates are input into the corrected model matrix, and the three-dimensional rigid body transformation matrix of the track during the current monitoring period is obtained by weighted least squares or iterative optimization. The multi-point cooperative attitude model and three-dimensional rigid body transformation matrix are used to analyze the changes in lateral displacement, vertical displacement, torsion angle and track gauge, forming a multi-attitude parameter vector of the track.

3. The laser positioning and monitoring method for marker points on railway tracks as described in claim 2, characterized in that, The calculation of the track's lateral displacement, vertical displacement, torsion angle, and gauge changes during the current monitoring period forms a multi-attitude parameter vector of the track, and also includes: Based on the actual spatial coordinates of the track markers, a track node topology matrix is ​​constructed. The track node topology matrix represents the spatial connectivity, neighborhood order, and structural constraint relationships between the track nodes. For each track marker, a historical motion trend vector of the corresponding track is constructed using historical monitoring data; The attitude calculation weights for each track marker point are performed based on the track node topology matrix and historical motion trend vector. The model matrix of the multi-point cooperative attitude model is optimized based on the calculation results to establish the orbital multi-attitude parameter vector.

4. The laser positioning and monitoring method for marker points on railway tracks as described in claim 2, characterized in that, An adaptive weighted matrix is ​​constructed using the confidence index of each track marker point, including: The confidence index of each track marker point is mapped to an initial weight, which is used to reflect the contribution of each marker point to the attitude solution. Obtain the coordinate residual between the relative spatial coordinates and the real spatial coordinates, configure an auxiliary weight factor based on the coordinate residual, perform initial weight compensation using the auxiliary weight factor, and construct compensation weights; The compensation weights are classified in a reliable manner, and reliable labels are configured. The reliable labels are then used to perform neighborhood collaborative weighting to construct an adaptive weighting matrix.

5. The laser positioning and monitoring method for marker points on railway tracks as described in claim 1, characterized in that, Based on the predicted positions of each orbit marker point at the next monitoring time using the orbital multi-attitude parameter vector, the marker point status is dynamically updated using the comparison results of the actual observed positions and predicted positions at the next monitoring time, and the confidence index. This includes: The orbital multi-attitude parameter vector is analyzed, and the analyzed feature set is used as the basic data to construct a temporal fitting prediction model for each orbital marker point. Obtain the monitoring interval for the next monitoring time, use the monitoring interval as auxiliary data, send it to the time series fitting prediction model, and output the predicted position of each track marker point; The residuals are calculated using the actual observed locations and the predicted locations to generate residual comparison results. The status of the marked points is then dynamically updated using the residual comparison results and the confidence index.

6. The laser positioning and monitoring method for marker points on railway tracks as described in claim 1, characterized in that, Anomalies are monitored based on the inversion suppression results and the output of orbital multi-attitude parameter vectors, including: Configure anomaly warning signals based on the anomaly indicators monitored; The aforementioned abnormal warning signal is used to report an anomaly and simultaneously generate a monitoring item of interest. Continuous monitoring and management of railway tracks will be implemented based on the monitoring data.

7. The laser positioning and monitoring method for marker points on railway tracks as described in claim 1, characterized in that, Vibration anomalies are identified using track vibration signals, vibration anomaly triggering early warning signals are established, and early warning issuance management of vibration anomaly triggering early warning signals is implemented.

8. A laser positioning and monitoring system for marking points on railway tracks, characterized in that, The steps for implementing the laser positioning and monitoring method for marker points on railway tracks according to any one of claims 1 to 7 include: The credibility calculation module is used to collect the laser reflection signals of the track markers, extract the reflection spectrum difference characteristics and the geometric distribution characteristics of the laser spots, construct a dual-modal feature set of the markers' spectrum and geometry, and use the dual-modal feature set to calculate the marker credibility index. The attitude parameter acquisition module is used to acquire the real spatial coordinates of the track marker points and read the relative spatial coordinates collected in real time. After adjusting the track marker point weights based on the marker point confidence index, it calculates the lateral displacement, vertical displacement, torsion angle and track gauge changes of the track during the current monitoring period, forming a multi-attitude parameter vector of the track. The state update module is used to predict the predicted position of each track marker point at the next monitoring time based on the track multi-attitude parameter vector, and dynamically update the state of the marker points by comparing the residual between the actual observed position and the predicted position at the next monitoring time and the confidence index. The monitoring anomaly output module is used to collect track vibration signals. After extracting the resonant energy spectrum features, it filters the track multi-attitude parameter vectors based on the updated marker point status, performs inversion suppression of the coupling relationship with the resonant energy spectrum features, and outputs the monitoring anomaly based on the inversion suppression results and the track multi-attitude parameter vectors.

Images