An urban subway track maintenance operation evaluation method

Abstract

The present application belongs to the technical field of metro track maintenance evaluation, and specifically discloses a kind of urban metro track maintenance operation evaluation method, by the straight section and section zoning of metro track, after maintenance, different disease identification methods are used for disease elimination identification for different line types, can follow the law of wheel-rail interaction and disease distribution characteristics, effectively improve the pertinence of disease detection, and combined with the zoned disease list before and after maintenance, realize the accurate evaluation of disease elimination rate in different areas, can accurately reflect the actual effectiveness of maintenance operation under different mechanical environment, improve the scientificity and guidance of evaluation, at the same time, on the basis of completing disease elimination rate evaluation in sub-section, the geometric length of straight section and the curvature characteristics of curved section are fused, respectively, the response relationship model of disease elimination rate is constructed, the different section maintenance cycle is different, and the data-driven adjustment is realized, which is beneficial to improve the scientificity and effectiveness of maintenance resource allocation.

Description

Technical Field

[0001] This invention belongs to the field of subway track maintenance evaluation technology, and specifically discloses a method for evaluating urban subway track maintenance operations. Background Technology

[0002] As the arteries of urban public transportation, urban subway tracks endure high-frequency, high-volume train cyclical loads over extended periods, inevitably leading to various defects such as rail wear, corrugated wear, rail surface cracks, and track geometry deviations. These defects not only affect passenger comfort but also seriously threaten operational safety. Therefore, regular track maintenance and scientific evaluation of its effectiveness are crucial for ensuring the safe, stable, and efficient operation of the subway.

[0003] In the subway track maintenance evaluation system, the defect elimination rate is the core quantitative indicator for measuring maintenance effectiveness. It reflects the extent to which maintenance operations cover existing defects and directly reflects the quality of maintenance. The primary step in conducting this evaluation is the accurate monitoring and identification of defects. Currently, relevant technical solutions support automated defect diagnosis. For example, Chinese invention patent CN113255825B proposes a method and device for identifying track bed defects. This method divides the track into several segments according to a preset unit length. Based on multi-period track geometric irregularity detection data, it constructs a time series of the standard deviation of specific irregularity parameters. By identifying significant change points in the time series, it divides the track into multiple segments and fits the slope of each segment to form a time series model of irregularity deterioration rate, thereby determining whether there are potential track bed defect segments. This technology enables trend analysis based on long-term monitoring data, improving the timeliness and objectivity of defect identification.

[0004] However, this scheme employs an equal-length segmentation strategy during the section division process, failing to incorporate the actual geometric characteristics of the track (such as straight and curved sections), and neglecting the significant differences in wheel-rail dynamic mechanisms, stress distribution patterns, and typical defect evolution under different track types. These morphological blind spots make it difficult for the defect identification model to accurately characterize the degradation behavior under specific track types, especially in curved sections where defect boundaries are prone to misjudgment or missed detection, weakening the physical consistency and sensitivity of the diagnosis.

[0005] Furthermore, when evaluating the maintenance effectiveness based on such identification results, the entire line is often treated as a homogeneous system, lacking a classification and evaluation mechanism for different mechanical environment sections such as straight sections and curved sections. If the line type is not distinguished for unified evaluation, the maintenance effect will be masked by averaging, failing to truly reflect the adaptability and effectiveness of the operation strategy under different working conditions. Ultimately, the evaluation results are coarse, lack mechanistic support, and are difficult to guide the fine optimization of subsequent maintenance cycles and process parameters. Summary of the Invention

[0006] Therefore, one objective of this application is to provide an evaluation method for urban subway track maintenance operations. By dividing the subway track into straight and curved sections, and performing post-maintenance defect elimination identification and targeted maintenance evaluation in each section, the problem mentioned in the background art is effectively solved.

[0007] The objective of this invention can be achieved through the following technical solution: a method for evaluating urban subway track maintenance operations, comprising the following steps: Step 1: dividing the subway track into straight and curved sections based on track line design parameters, and generating a track partition model with section attribute labels.

[0008] Step 2: Construct an initial dataset of track defects containing defect locations and initial feature vectors based on historical track detection information, and then associate it with the track partitioning model to generate a list of defects before maintenance by classifying them into straight / curved segments.

[0009] Step 3: After the maintenance work is completed, call the 3D scanning device to obtain track surface data, and perform a comparison of the elimination status of defect features in the straight section and the curved section using differentiated feature analysis areas, and output a list of defect elimination for each section.

[0010] Step 4: Evaluate the disease elimination rate of straight and curved sections based on the disease elimination list and the pre-maintenance disease list.

[0011] Step 5: Combine the geometric length of the straight section and the curvature of the curved section, and construct their response relationship with the disease elimination rate. Based on the response relationship, adjust the maintenance cycle of different sections accordingly.

[0012] Combining all the above technical solutions, the positive effects of this invention are as follows: 1. By dividing the subway track into straight sections and segments, and after maintenance, this invention uses differentiated defect identification methods for different track types to identify and eliminate defects. This can follow the wheel-rail interaction law and defect distribution characteristics, effectively improving the targeting and accuracy of defect detection. Furthermore, by combining the defect lists before and after maintenance, it can achieve accurate evaluation of defect elimination rates by region and type, accurately reflecting the actual efficiency of maintenance operations under different mechanical environments, improving the scientific nature and guidance of the evaluation, and providing data support for the optimization of differentiated maintenance strategies.

[0013] 2. This invention, based on the evaluation of track defect elimination rates across different sections, integrates the geometric length of straight sections and the curvature characteristics of curved sections to construct response relationship models between these models and the defect elimination rates. This enables differentiated, data-driven adjustments to the maintenance cycles of different sections. This method can accurately respond to the deterioration characteristics of tracks under different geometric and mechanical environments, improve the scientific and effective allocation of maintenance resources, promote the transformation of operation and maintenance models from uniform cycles to on-demand maintenance, and enhance the safety and sustainability of the track system. Attached Figure Description

[0014] The present invention will be further described with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present invention. For those skilled in the art, other drawings can be obtained based on the following drawings without creative effort.

[0015] Figure 1 This is a diagram illustrating the implementation steps of the method of the present invention.

[0016] Figure 2 This is a flowchart illustrating the specific process of generating the pre-maintenance disease list in this invention.

[0017] Figure 3 This is a schematic diagram illustrating the operation of adjusting the maintenance cycle of different sections by integrating the geometric length of the straight section and the curvature of the curved section in this invention. Detailed Implementation

[0018] 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.

[0019] See Figure 1 As shown, this invention proposes an evaluation method for urban subway track maintenance operations, including the following steps: Step 1: Based on the track line design parameters, the subway track is divided into straight and curved sections, and a track partitioning model with section attribute labels is generated.

[0020] As a preferred implementation of the above scheme, the specific implementation process of Step 1 is as follows: extract the spatial curvature radius sequence of the track centerline and the absolute geographical coordinates or track mileage codes of the sleepers along the line from the track line design parameters.

[0021] It should be noted that urban rail transit lines are composed of alternating straight and curved segments on the plane, and their geometry can be precisely characterized by a spatial curvature radius sequence of the track centerline. This is because the track centerline, as the reference axis of the line's spatial geometry, is the core geometric entity describing the track's planar position and orientation, reflecting the geometric continuity and morphological changes of the line in three-dimensional space. Specifically, the track centerline of a straight segment approximates a straight line in Euclidean space, and its differential geometric characteristics are characterized by curvature approaching zero and curvature radius approaching infinity; in design parameters, it is usually represented by a maximum constant or an ideal infinite radius. The track centerline of a curved segment has a finite and continuously varying curvature radius along the line direction, commonly exhibiting a combination of transition curves and circular curves; its curvature gradient reflects the gradual transition characteristics of the wheel-rail contact geometry. Therefore, based on the curvature radius sequence, high-precision analysis of the track line's geometry and topologically continuous segment classification can be achieved.

[0022] It should be further pointed out that the spatial distribution of sleepers along the track centerline reflects the physical discretization layout and spatial positioning benchmark of the track structure. As the supporting unit of the track structure, the sleepers provide a high-precision spatial mapping benchmark and physical entity association basis for distinguishing the location of defects.

[0023] The curvature radius sequence of the track centerline is scanned segment by segment with a set step size: if the curvature radius of all sampling points in a certain continuous segment is greater than or equal to the preset straight segment judgment threshold and the spatial projection length of the segment is not less than the minimum straight segment length, then the segment is identified as a straight segment. If a continuous segment meets any of the following conditions, it is determined to be a curved segment: a) the average curvature radius in the segment is less than the preset straight segment judgment threshold.

[0024] b) There is a curvature gradient, that is, the rate of change of curvature along the path direction is greater than the gradient threshold.

[0025] c) It has a curved transition structure.

[0026] As an application of the above operation, when dividing straight segments, satisfying the curvature condition alone is not enough to constitute a valid straight segment. It is also necessary to ensure its spatial continuity. The minimum straight segment length is introduced as a criterion to exclude pseudo-straight segments caused by short-distance fluctuations or measurement noise, and to ensure that the identification results conform to the actual line structure characteristics.

[0027] In addition, the minimum straight section length should not be less than the minimum stable operating length of maintenance equipment such as rail grinding vehicles and inspection trains, so as to ensure that the section has the engineering significance for independent maintenance operations.

[0028] The straight-segment determination threshold reflects the engineering definition boundary of straight lines in track design, that is, the maximum allowable curvature level corresponding to a functional straight segment that can be considered to have no curvature impact in actual construction and operation. The setting of this threshold can refer to the provisions on minimum curve radius in the track design specifications.

[0029] It should also be noted that when classifying curved sections, condition a) reflects that the overall alignment of the section has significant bending characteristics; condition b) indicates that the curvature has significant non-uniform variation characteristics, usually corresponding to transition sections such as transition curves, reflecting the gradual evolution of the alignment curvature; and condition c) possesses a curved transition structure, such as a transitional point, a rounded transitional point, a straight transitional point, or a transitional straight point, conforming to the curve connection patterns in the track geometry design specifications. The above three types of conditions cover the main geometric forms of curved sections, and satisfying any one of these conditions determines it to be a curved section, which conforms to the actual design logic and alignment continuity requirements of the track line.

[0030] The aforementioned curvature gradient threshold reflects the critical level of the rate of change of curvature of the track alignment in the spatial domain. Its core physical significance lies in identifying whether there are transition curve sections with significant geometric transition characteristics. According to the *Metro Design Code* and the *Railway Line Design Code*, the setting of transition curves must meet the requirements of train dynamics. Their length is related to the radius of the connecting circular curve and the design train speed to achieve a gradual transition of curvature. Under the ideal spiral model, the curvature changes linearly with the arc length, and its average curvature gradient can be expressed as… ,in Indicates the radius of the circular curve. This indicates the minimum design length of the transition curve, combined with typical parameters of urban rail transit. , Substituting this value into the equation yields the average curvature gradient, which can serve as an important reference for setting the gradient threshold.

[0031] Based on the above classification rules, the entire line is divided into sections, a track partition model with topological continuity is constructed, the type attributes of each partition and its spatial start and end range are marked, and a mapping index table between partition boundaries and sleeper physical numbers is further established.

[0032] Step 2: Construct an initial dataset of track defects containing defect locations and initial feature vectors based on historical track detection information, and then associate it with the track partitioning model to generate a list of defects before maintenance by classifying them into straight / curved segments.

[0033] See Figure 2 As shown, as an optional implementation of the above scheme, the pre-maintenance defect list generated by associating with the track zoning model and classifying by straight / curved segments includes the following: based on the location information of each defect recorded in the initial track defect dataset, it is orthogonally projected onto the track centerline to obtain the arc length mileage of the defect point along the track direction.

[0034] It is important to understand that urban subway tracks have a long and narrow linear structure. The distance that defects deviate from the centerline laterally is usually small, but the distribution along the track direction is key information. Orthogonally projecting the two-dimensional or three-dimensional coordinates of the defects onto the track centerline to obtain its arc length along the track is a standard spatial normalization method.

[0035] Based on the arc length mileage and the sleeper spatial database, a proximity query is performed to determine the range of sleeper physical numbers adjacent to this location.

[0036] By using a pre-established index table mapping sleeper numbers to sections, the section to which a sleeper number belongs can be queried, thereby deriving the logical section to which the defect point belongs.

[0037] It is important to understand that sleepers are the most basic and stable physical unit in the track structure, and their numbers are unique and identifiable on-site. Using known sleeper spacing and starting reference points, the sleeper number corresponding to any mileage can be quickly located through a database index. The section-sleeper number mapping index table constructed in the early stage is essentially a spatial semantic dictionary, associating the underlying physical unit, the sleeper, with its corresponding functional section.

[0038] By comparing the arc length of the defect point with the spatial start and end range of each segment in the track partitioning model, the segment to which the defect point belongs is determined based on geometric spatial matching.

[0039] The consistency of the two types of attribution segments obtained above is checked: if the logical attribution is consistent with the spatial attribution, then the segment to which the disease belongs is confirmed to be valid.

[0040] If the logical attribution and spatial attribution are inconsistent, a manual review will be initiated to determine the section to which the disease belongs.

[0041] The above-mentioned dual verification mechanism, which combines logical structure-based attribution with geometric space-based attribution, achieves high accuracy and reliability in disease attribution determination, providing a precise spatial benchmark and reliable classification basis for subsequent disease elimination and identification.

[0042] After determining the segment affiliation of all disease points, the disease data is classified and collected according to segment attributes. Specifically, all diseases located in straight segments are classified into the straight segment disease subset, and diseases located in curved segments are classified into the curved segment disease subset.

[0043] For each subset, a list of defects is generated before maintenance. The list includes the location of the defects, the initial feature vector, and the segment identifier.

[0044] The initial feature vectors mentioned above include geometric shape, disease severity, etc., which comprehensively reflect the physical characteristics of the disease and provide a quantifiable state description basis for subsequent disease elimination analysis. The geometric shape includes, but is not limited to, length, width, height, surface area, etc., and the disease severity includes, but is not limited to, the amplitude of lateral protrusions / indentations, surface roughness, etc.

[0045] Step 3: After the maintenance work is completed, call the 3D scanning device to obtain track surface data, and perform a comparison of the elimination status of defect features in the straight section and the curved section using differentiated feature analysis areas, and output a list of defect elimination for each section.

[0046] As one possible approach to the above solution, the following operations are performed to compare the state of disease feature elimination for straight and curved sections using differentiated feature analysis regions: For straight sections, a feature analysis cube is constructed for each recorded disease location in the pre-maintenance disease list based on the spatial scale of the disease obtained from the initial detection. This cube is centered on the disease center and extends outward along the length, width, and height directions of the original disease boundary with a preset safety reserve scale to encompass the disease-affected area and the possible diffusion range of maintenance effects, ensuring that the analysis window completely covers the potential change area.

[0047] Understandably, track defects such as peeling, dents, and corrugations have a defined spatial distribution range, and their impact is mainly concentrated in local areas. The effects of maintenance operations such as grinding and milling also decrease outward from the original location of the defect. Constructing a feature analysis cube centered on the defect location and expanding according to the actual size conforms to the spatial locality law of physical action.

[0048] After the maintenance work is completed, high-precision point cloud data of the track surface is obtained by a three-dimensional laser scanning device. Track geometry and surface condition information are extracted within the feature analysis cube to generate a post-maintenance feature vector.

[0049] The feature vectors of the diseased points after maintenance are compared with the initial feature vectors item by item, and the feature difference of each dimension is calculated. The feature difference is the difference between the feature vectors after maintenance and the initial feature vectors. It is determined whether the feature difference exceeds the preset disease elimination judgment threshold. At the same time, the feature vectors after maintenance are compared with the feature space range of the healthy track state. If all feature differences reach the disease elimination judgment threshold or the feature vectors after maintenance fall within the feature space range of the healthy track state, then the disease is determined to be eliminated.

[0050] As an explanation of the above scheme, the defect elimination judgment threshold reflects the minimum degree of improvement that defect characteristics need to achieve after maintenance, that is, the quantitative standard for judging whether a certain defect has been effectively treated. Its setting should be based on the detection accuracy of track geometry and surface condition, the repair tolerance of typical defects, and operational safety limits, and is usually taken as 2-3 times the equipment resolution or determined through statistical analysis of historical repair data.

[0051] The characteristic space range of a healthy track condition characterizes the multidimensional feature distribution boundary of the track surface geometry and structure under conditions without significant defects, reflecting the normal fluctuation range of various indicators under ideal service conditions. This range is defined by a high-density compact region identified by a clustering algorithm after standardization of the feature vectors of a large number of historical healthy track samples, forming a benchmark reference domain in multidimensional space. If the feature vectors after maintenance fall entirely within this region, it indicates that the track condition has recovered to a normal level, and defects can be determined to have been eliminated.

[0052] It should be noted that the essence of defect elimination lies in the structural transformation of the track surface condition from abnormal deterioration to functional recovery, the core of which is the rebalancing of geometric morphology and load-bearing capacity. Compared with primary criteria that rely solely on characteristic changes, a higher-order evaluation logic should focus on the regressivity of the condition, i.e., whether the condition returns to the normal service range after maintenance. By integrating the dual criteria of functional recovery and normal regression, a multi-dimensional collaborative assessment of the defect elimination effect is achieved, significantly improving the accuracy, physical interpretability, and engineering credibility of the evaluation results, and deepening the evolution of the evaluation system from a process-response type to a state-oriented type.

[0053] For curved sections, based on the location of each defect recorded in the pre-maintenance defect list, an arc-shaped analysis channel is constructed along the track local method plane. The track local method plane is a plane perpendicular to the tangent direction of the line and containing the curvature vector. The channel takes the defect center as the starting section and extends a preset distance along the direction of line advance to form a three-dimensional spatial region covering the defect influence area, ensuring sufficient spatial inclusiveness for the unique asymmetric wear and geometric evolution characteristics of curved sections.

[0054] Understandably, curved tracks exhibit significant curvature effects, leading to asymmetrical stress distribution and an arc-shaped geometric shape across the track cross-section. Using the cubic analysis region commonly employed for straight sections would result in projection distortion, inner rail compression, and outer rail stretching on the curve, distorting the analysis range. Constructing an arc-shaped analysis channel extending along the local normal plane, conforming to the spatial morphology of the curved section, ensures accurate coverage of the disease's influence zone both radially and longitudinally.

[0055] Within the constructed arc-shaped analysis channel, three-dimensional analysis grid points with equal spacing are arranged along the longitudinal and transverse directions. For each grid point, the corresponding local feature vector is extracted from the high-precision point cloud data obtained after maintenance, referring to the feature extraction method of the straight track area.

[0056] As illustrated in the example, the horizontal direction refers to the track direction, and the vertical direction refers to the cross-sectional direction of the track. The grid resolution is usually set to 5 mm × 5 mm × 5 mm to achieve high-precision spatial sampling of track surface morphology changes.

[0057] Furthermore, it is understandable that track defects in curved sections are often distributed continuously in bands or patches along the track direction, rather than as isolated points. Single-point assessments are easily affected by noise and cannot reflect the overall repair quality. High-density sampling is achieved by deploying an evenly spaced three-dimensional grid of points within the channel.

[0058] The local feature vector of each grid point is compared with the initial feature vector of the corresponding position before maintenance, and combined with the feature space range of the pre-constructed healthy trajectory state, a single-point judgment is made according to the same disease elimination judgment logic as the straight section area. If either the feature difference is significant or the state returns to the healthy domain, the grid point is judged as eliminated; otherwise, it is marked as not eliminated.

[0059] Based on the judgment results of all grid points within the channel, a channel-level overall elimination status assessment is performed to obtain the judgment result of the elimination of defects in the curved section.

[0060] Specifically, the overall elimination status assessment at the channel level is implemented as follows: the proportion of the number of grid points determined to have been eliminated to the total number of grid points in the channel is used as the elimination coverage rate.

[0061] For all unresolved grid points, contour extraction and closure connections are performed along their spatial distribution boundaries to form unresolved patches. Unresolved patches represent continuous anomalous regions with clear spatial boundaries and geometric shapes.

[0062] For unresolved patches, the proportion of unresolved grid points within the unresolved patch to the total number of grid points within the unresolved patch is used as the degree of unresolved clustering.

[0063] It should be noted that when constructing unresolved patches from unresolved grid points, if already resolved grid points are interspersed between adjacent unresolved grid points, direct connection may result in the inclusion of non-disease area grid points within the patch. In this case, the unresolved patches formed by morphological closure or boundary tracing algorithms will contain a certain number of internal filling points. The ratio of the number of unresolved grid points within a patch to the total number of grid points within the patch is defined as the unresolved clustering degree. This indicator reflects the spatial clustering and continuity of the disease residue area. The higher the unresolved clustering degree, the more concentrated the distribution of residual disease, and the closer it is to the true state of local untreated or insufficiently treated areas.

[0064] The elimination coverage rate is compared with the coverage rate limit, and the non-eliminated clustering degree is compared with the allowable clustering degree. If the elimination coverage rate reaches the coverage rate limit and the non-eliminated clustering degree is lower than or equal to the allowable clustering degree, the disease of the curve segment is determined to be eliminated; otherwise, the disease of the curve segment is determined to be not eliminated.

[0065] The instructions for the above operations state that the coverage limit reflects the minimum effective elimination rate that maintenance work must achieve on the spatial coverage area. The specific limit can be set according to the engineering requirements for the overall removal of defects. The allowable clustering reflects the tolerance for the spatial distribution pattern of residual defects, i.e., the lower limit of the concentration of unremoved areas, used to identify whether there are large, continuous untreated areas. High clustering indicates that residual defects are distributed in a blocky pattern, possibly due to blind spots or insufficient processes; low clustering shows discrete spots, which is normal fluctuation. The allowable clustering can be determined by analyzing the distribution of residual defects after high-quality maintenance, using clustering statistics combined with quantiles to determine the threshold. Values ​​below this threshold are considered uneven elimination and require rework.

[0066] It should be explained that when conducting a segment-level overall assessment based on the elimination judgment results of each grid point, relying solely on the achievement of the overall elimination coverage rate while ignoring the spatial distribution characteristics of residual diseases may obscure the existence of large-scale continuous untreated areas. Therefore, by introducing the elimination coverage rate to measure the global integrity of the restoration, and combining it with the degree of unresolved clustering to identify potential blind spots or weak areas in the process, a dual test of the spatial uniformity and structural continuity of the elimination effect can be achieved.

[0067] As another possible way to implement the above scheme, the output of the disease elimination list for each section is as follows: count the number of disease points in each section that have been determined to have been eliminated, and extract the location information of each eliminated disease point, thereby forming the disease elimination list for each section.

[0068] Step 4: Evaluate the disease elimination rate of straight and curved sections based on the disease elimination list and the pre-maintenance disease list.

[0069] Specifically, the evaluation of the disease elimination rate for straight and curved sections is carried out as follows: For any section, the total number of diseases in the disease list before maintenance and the number of diseases that have been eliminated in the disease elimination list are counted, and the proportion of the number of eliminated diseases to the total number of diseases is used as the disease elimination rate.

[0070] Step 5: Combine the geometric length of the straight section and the curvature of the curved section, and construct their response relationship with the disease elimination rate. Based on the response relationship, adjust the maintenance cycle of different sections accordingly.

[0071] See Figure 3As shown, in a preferred embodiment, the above steps are specifically implemented as follows: obtain the spatial start and end range of each straight section along the entire line, calculate its geometric length, and then construct a length-disease elimination response curve with the straight section length as the horizontal axis and the disease elimination rate as the vertical axis to characterize the nonlinear trend of maintenance effect with the change of section spatial scale.

[0072] The inflection point of the length-disease elimination response curve is identified to obtain the critical length.

[0073] For example, inflection point identification can be achieved by using mathematical analysis of the response curve, such as the second derivative extreme value method, the curvature maximum method, or piecewise linear regression breakpoint analysis, to identify the key turning point where the elimination rate growth trend changes from significant to gradual. The length value corresponding to this point is defined as the critical length, reflecting the minimum segment length threshold required for maintenance operations to achieve stable performance.

[0074] Based on the critical length, all straight segments are divided into short straight segments with a length lower than or equal to the critical length and standard straight segments with a length higher than the critical length.

[0075] It is important to clarify that while the length of the straight section is not a direct cause of track defects, it is a key spatial parameter reflecting the geometric continuity and structural integrity of the track. Its dimensional characteristics are closely related to the dynamic response characteristics of large maintenance equipment and the requirements of operational processes. When the straight section length is too short, failing to meet the minimum stable operating range requirements of the equipment, the proportion of the start-stop transition section in the total operating length will increase significantly, leading to problems such as lag in grinding power adjustment, pressure control fluctuations, and unstable initialization of detection data, thereby reducing the uniformity and completeness of defect treatment. In contrast, a standard straight section has sufficient spatial extensibility, which is conducive to the maintenance equipment entering and maintaining a steady-state operating mode, ensuring the convergence of process parameters and controllable operation, thus achieving a stable and efficient defect elimination effect.

[0076] Therefore, the length of a straight section indirectly affects the disease elimination rate by influencing the process stability and spatial coverage consistency of maintenance operations. To ensure that the zoning based on critical length has statistical significance and engineering interpretability, it is necessary to verify the correlation between the disease elimination rate and the length of a straight section. For example, the Pearson correlation coefficient can be used to quantify the degree of linear correlation between the two, or piecewise regression analysis can be used to test whether the trend of the elimination rate in different length intervals shows a significant positive correlation, thereby improving the scientific nature and decision reliability of the zoning strategy.

[0077] For standard straight sections, the baseline maintenance cycle is maintained. For short straight sections, the average disease elimination rate of all standard straight sections is calculated. Then, for each short straight section, the ratio of its elimination rate to the average level of standard straight sections is calculated, and the baseline maintenance cycle is shortened accordingly.

[0078] It should be noted that maintaining the baseline maintenance cycle for standard straight sections ensures efficient resource utilization. For short straight sections, the dynamic adjustment expression for their maintenance cycle is as follows: In the formula This indicates the adjusted maintenance cycle. Indicates the baseline maintenance cycle. This indicates the disease elimination rate of the short, straight section. This represents the average level of disease elimination rate in standard straight sections. This indicates the minimum maintenance cycle.

[0079] The explanation applied to the above expression is as follows: This reflects the relative maintenance efficiency of short straight sections compared to standard sections. In practice, it is necessary to consider... Limited to The smaller this value, the worse the elimination effect of defects in short straight sections, the higher the degree of inadequate treatment, and the more necessary it is to increase the frequency of intervention to compensate for insufficient operational efficiency. Multiplying this value by the baseline cycle constitutes the cycle scaling factor, realizing a negative feedback linkage mechanism between maintenance frequency and actual treatment effect: the lower the elimination rate, the smaller the scaling factor, and the shorter the adjusted cycle, which conforms to the predictive maintenance logic that the more severe the condition deterioration, the more frequent the maintenance. Simultaneously, a lower limit constraint is introduced. To prevent excessive compression of maintenance cycles due to abnormally low elimination rates, and to avoid problems such as excessive resource occupation, overloaded equipment and manpower, and non-linear increase in maintenance costs, the adjustment strategy is implemented within the boundaries of engineering feasibility and operational sustainability.

[0080] Obtain the curvature of each section along the entire line, and construct a curvature-disease elimination response curve with curvature on the horizontal axis and disease elimination rate on the vertical axis.

[0081] The inflection point of the curvature-disease elimination response curve is identified to obtain the critical curvature.

[0082] The identification method for the critical curvature mentioned above is similar to the identification method for the critical length.

[0083] Based on the critical inflection point, all curved sections are divided into gentle curved sections with curvature below or equal to the critical curvature and steep curved sections with curvature above the critical curvature.

[0084] It is important to clarify that there is a significant correlation between the defect elimination rate and track curvature in curved sections. The fundamental reason is that curvature, as a core geometric parameter of wheel-rail interaction, directly determines the mechanical environment and contact state when a train passes, thus affecting the occurrence mechanism, development rate, and effectiveness of maintenance operations. Specifically, as curvature increases, the lateral force between the wheel and rail significantly strengthens, leading to a rise in the frequency and severity of defects such as rail side wear, rail head plastic deformation, corrugation, and spalling. Therefore, high-curvature sections typically exhibit lower defect elimination rates, reflecting limited actual effectiveness of maintenance operations. In contrast, gently curved sections experience relatively balanced stress, better wheel-rail matching, higher stability of maintenance equipment operations, and easier maintenance of treatment effects. Therefore, to ensure the statistical significance and engineering interpretability of section division based on critical curvature, it is necessary to verify the statistical correlation between curvature and defect elimination rate. It is recommended to use methods such as Pearson correlation coefficient or piecewise regression analysis to examine the statistical strength and trend consistency of the correlation between the two, thereby enhancing the scientific nature of the zoning strategy.

[0085] For gently curved sections, the baseline maintenance cycle is maintained; for sharply curved sections, the baseline maintenance cycle is shortened by referring to the cycle adjustment logic of short straight sections.

[0086] This invention, based on the evaluation of track defect elimination rates across different sections, integrates the geometric length of straight sections and the curvature characteristics of curved sections to construct response relationship models between these models and the defect elimination rates. This enables differentiated, data-driven adjustments to the maintenance cycles of different sections. This method can accurately respond to the deterioration characteristics of tracks under different geometric and mechanical environments, improve the scientific and effective allocation of maintenance resources, promote the transformation of operation and maintenance models from uniform cycles to on-demand maintenance, and enhance the safety and sustainability of the track system.

[0087] The parameters involved in the above formula are all dimensionless and calculated numerically. The formula is a formula obtained from the most recent real situation by collecting a large amount of data and simulating it with software. The preset parameters in the formula are set by those skilled in the art according to the actual situation.

[0088] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0089] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0090] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0091] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0092] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for evaluating urban subway track maintenance operations, characterized in that, Includes the following steps: Step 1: Based on the track design parameters, divide the subway track into straight and curved sections, and generate a track partition model with section attribute labels; Step 2: Construct an initial dataset of track defects containing defect locations and initial feature vectors based on historical track detection information, and then associate it with the track partitioning model to generate a list of defects before maintenance by classifying them into straight / curved segments; Step 3: After the maintenance work is completed, call the 3D scanning device to obtain track surface data, perform a comparison of the elimination status of defect features in the straight section and the curved section using differentiated feature analysis areas, and output a list of defect elimination for each section. Step 4: Evaluate the disease elimination rate of straight and curved sections based on the disease elimination list and the pre-maintenance disease list; Step 5: Combine the geometric length of the straight section and the curvature of the curved section to construct their response relationship with the disease elimination rate, and adjust the maintenance cycle of different sections accordingly based on the response relationship; The procedure for comparing the elimination status of defects using differentiated feature analysis regions for straight and curved sections is as follows: For straight sections, a feature analysis cube is constructed for each recorded defect location in the pre-maintenance defect list based on the defect scale boundary obtained from the initial detection. This cube has the defect center as its geometric center and extends outward along the length, width, and height directions of the original defect boundary by a preset safety allowance scale. After the maintenance work is completed, high-precision track surface point cloud data is acquired using a 3D laser scanning device. Track geometry and surface state information are extracted within the feature analysis cube to generate a post-maintenance feature vector. The post-maintenance feature vector of the defect points is compared item by item with the initial feature vector to calculate the feature difference in each dimension. It is determined whether the feature difference exceeds the preset defect elimination judgment threshold. At the same time, the post-maintenance feature vector is compared with the feature space range of the healthy track state. If all feature differences reach the defect elimination judgment threshold or the post-maintenance feature vector falls within the feature space range of the healthy track state, the defect is determined to have been eliminated. The method of performing a differential feature analysis region comparison for the elimination of disease features for straight sections and curved sections also includes the following: For curved sections, based on the location of each disease recorded in the pre-maintenance disease list, an arc-shaped analysis channel is constructed along the local plane of the track. The channel takes the center of the disease as the starting section and extends a preset distance along the direction of the track to form a three-dimensional spatial region covering the disease influence area. Within the constructed arc-shaped analysis channel, equidistant three-dimensional analysis grid points are arranged longitudinally and laterally. For each grid point, the corresponding local feature vector is extracted from the high-precision point cloud data obtained after maintenance, referring to the feature extraction method of the straight track area. The local feature vector of each grid point is compared with the initial feature vector of the corresponding position before maintenance, and combined with the feature space range of the pre-constructed healthy track state, a single-point judgment is made according to the same defect elimination judgment logic as the straight section area. Based on the judgment results of all grid points in the channel, a channel-level overall elimination status assessment is performed to obtain the judgment result of defect elimination in the curved section.

2. The method for evaluating urban subway track maintenance operations as described in claim 1, characterized in that: The specific implementation process of Step 1 is as follows: Extract the spatial curvature radius sequence of the track centerline and the absolute geographical coordinates or track mileage codes of the sleepers along the track from the track design parameters. The curvature radius sequence of the track centerline is scanned segment by segment using a set step size: if the curvature radius of all sampling points in a continuous segment is greater than or equal to the preset straight segment determination threshold and the spatial projection length of the segment is not less than the minimum straight segment length, then the segment is identified as a straight segment; if a continuous segment meets any of the following conditions, it is determined to be a curved segment: a) The average radius of curvature within the segment is less than the preset straight segment determination threshold; b) There is a curvature gradient, that is, the rate of change of curvature along the path direction is greater than the gradient threshold; c) It has a curved transition structure; Based on the above classification rules, the entire line is divided into sections, a track partitioning model with topological continuity is constructed, the type attributes of each section and its spatial start and end range are marked, and a mapping index table between sections and sleeper physical numbers is further established.

3. The method for evaluating urban subway track maintenance operations as described in claim 2, characterized in that: The pre-maintenance defect list includes the following generated content: Based on the location information of each defect recorded in the initial dataset of track defects, it is orthogonally projected onto the track centerline to obtain the arc length mileage of the defect point along the track direction. Based on the arc length mileage and the sleeper spatial database, a proximity query is performed to determine the range of sleeper physical numbers adjacent to this location; By using a pre-established index table mapping between sections and sleeper physical numbers, the section to which a sleeper number belongs can be queried, thereby deriving the logical section to which the defect point belongs. The arc length of the defect point is compared with the spatial start and end range of each segment in the track partitioning model to determine the interval inclusion, and the segment to which the defect point belongs is obtained based on geometric spatial matching. Perform a consistency check on the two types of attribution segments obtained above: If the logical attribution and spatial attribution are consistent, then the section to which the disease belongs is confirmed to be valid; If the logical attribution and spatial attribution are inconsistent, a manual review will be initiated to determine the section to which the disease belongs; After determining the segment affiliation of all disease points, the disease data is classified into straight segment disease subsets and curved segment disease subsets according to segment attributes; For each subset, a list of defects is generated before maintenance. The list includes the location of the defects, the initial feature vector, and the segment identifier.

4. The method for evaluating urban subway track maintenance operations as described in claim 3, characterized in that: The overall elimination status assessment at the channel level is as follows: The elimination coverage rate is calculated as the proportion of the number of grid points that have been determined to be eliminated to the total number of grid points in the channel. For all the unremoved grid points, contour extraction and closed connection are performed along their spatial distribution boundaries to form unremoved patches; For unresolved patches, the proportion of unresolved grid points within the unresolved patch to the total number of grid points within the unresolved patch is used as the degree of unresolved clustering. The elimination coverage rate is compared with the coverage rate limit, and the non-eliminated clustering degree is compared with the allowable clustering degree. If the elimination coverage rate reaches the coverage rate limit and the non-eliminated clustering degree is lower than or equal to the allowable clustering degree, the disease of the curve segment is determined to be eliminated; otherwise, the disease of the curve segment is determined to be not eliminated.

5. The method for evaluating urban subway track maintenance operations as described in claim 1, characterized in that: The output section's disease elimination list is as follows: The number of disease points in each section that have been determined to have been eliminated is counted, and the location information of each eliminated disease point is extracted to form a list of disease eliminations for each section.

6. The method for evaluating urban subway track maintenance operations as described in claim 1, characterized in that: Step 4 includes the following: For any given section, the total number of defects in the pre-maintenance defect list and the number of defects eliminated in the defect elimination list are counted, and the proportion of eliminated defects to the total number of defects is used as the defect elimination rate.

7. The method for evaluating urban subway track maintenance operations as described in claim 2, characterized in that: The specific implementation process of Step 5 is as follows: The geometric length is calculated based on the spatial start and end range of each straight section along the entire line. Then, a length-disease elimination response curve is constructed with the straight section length as the horizontal axis and the disease elimination rate as the vertical axis. The inflection point of the length-disease elimination response curve is identified to obtain the critical length. Based on the critical length, all straight segments are divided into short straight segments with a length lower than or equal to the critical length and standard straight segments with a length higher than the critical length. For standard straight sections, the baseline maintenance cycle is maintained. For short straight sections, the average disease elimination rate of all standard straight sections is calculated. Then, for each short straight section, the ratio of its elimination rate to the average level of standard straight sections is calculated, and the baseline maintenance cycle is shortened accordingly.

8. The method for evaluating urban subway track maintenance operations as described in claim 7, characterized in that: Step 5 also includes the following: Obtain the curvature of each segment along the entire line, and construct a curvature-disease elimination response curve with curvature on the horizontal axis and disease elimination rate on the vertical axis; The inflection point of the curvature-disease elimination response curve is identified to obtain the critical curvature. Similarly, all curve segments are divided into gentle curve segments and steep curve segments based on the critical curvature. For gently curved sections, the baseline maintenance cycle is maintained; for sharply curved sections, the baseline maintenance cycle is shortened by referring to the cycle adjustment logic of short straight sections.

Images