High-speed rail fastener pre-configuration method based on virtual track simulation

Abstract

The application discloses a high-speed rail fastener pre-configuration method based on virtual track simulation, which comprises the following steps: S1, collecting full-element point clouds on a track plate of a line to be finely adjusted; S2, removing noise and cleaning data of the full-element point clouds by using a filtering algorithm; S3, target region segmentation; S4, rail support platform point cloud pose representation: rail support platform point cloud structure subdivision; rail support platform point cloud attitude extraction; rail support platform point cloud center position extraction; S5, virtual track environment establishment; S6, track simulation fine adjustment: calculating a TQI value of a target unit section; simulation fine adjustment principle; judging whether the TQI value meets acceptance indexes and outputting a simulation fine adjustment scheme; establishing an LP model; nonlinear optimization of a change amount Z of a total adjustment amount in the simulation fine adjustment scheme; and S7, converting the simulation fine adjustment scheme into a fastener pre-configuration scheme. The method shortens a high-speed rail construction period, reduces laying cost and improves construction quality and safety.

Description

Technical Field

[0001] The present invention belongs to the technical field of high-speed railway ballastless track construction, and in particular relates to a high-speed railway fastener pre-configuration method based on virtual track simulation. Background Art

[0002] To ensure that high-speed railway tracks meet the requirements for smooth and safe operation of high-speed trains, the smoothness of the high-speed railway tracks must be measured during construction and operation. The long rails must be fine-tuned by adjusting the combination of fasteners and structural components to adjust the horizontal and vertical offsets of the long rails and ensure that the track smoothness meets the requirements. The traditional fastener configuration scheme is to install standard fasteners on each sleeper along the entire line before the rails are laid. After the long rails are stressed and locked, static track testing and fastener configuration are performed. The initial track measurement process does not take into account the actual conditions of the rail support platform, and a standard fastener system is uniformly installed. After the track is laid, a "fixed tool" is used to obtain track measurement data. The specifications of the adjustment parts are then calculated, the fastener system is updated, and the track data is remeasured. The adjustment part specifications are repeatedly calculated and the fastener system is updated until the track smoothness meets the requirements. Adjustment parts refer to structural parts that need to be adjusted, and structural parts are divided into different models.

[0003] The above solution has the following problems:

[0004] (1) The time efficiency of the working mode is low, which greatly slows down the construction progress. At the same time, changes in environmental conditions (temperature, light, external force, accidental interference) over a long span of time will also affect the accuracy and precision of the measurement results, thereby affecting the design effectiveness of the fine-tuning plan. Repeated operations result in low fine-tuning efficiency.

[0005] (2) In addition to the reduction in manpower and material costs due to improved measurement efficiency, one of the economic costs that accounts for a relatively high proportion in the current working conditions is the cost of replacing standard fasteners with fine-tuning fasteners during the fine-tuning of long rails. In a certain project, the average replacement rate of insulation blocks during the fine-tuning of long rails was 95%, the average replacement rate of gauge blocks was 12%, and the average replacement rate of rail pads was 205%. The replacement cost was approximately RMB 93,000 per kilometer of track laid. Therefore, it is particularly important to conduct laying environment measurements before laying long rails, especially precise measurements of track plates and rail supports, and to carry out simulation fine-tuning and solution simulation work.

[0006] (3) Currently, the measurement mode of "CPIII + total station + track support frame" is still used from the track plate fine-tuning to the track laying stage. Under this working mode, on the one hand, the measurement accuracy is severely limited by multiple factors such as the accuracy of CPIII point verification, the tolerance of the total station setting and the overlap accuracy between multiple stations, the placement deviation of the track support frame, and manual measurement errors; on the other hand, the large-scale three-dimensional spatial information is replaced by local sparse point information, which directly affects the coverage completeness, information distribution uniformity, and feature description ability of the measurement results.

[0007] (4) The current measurement and fine-tuning perspective is still relatively small, focusing mainly on the local smoothness of the line and the static smoothness of the entire line, and it is impossible to intervene in the dynamic joint debugging and testing process in advance. Summary of the Invention

[0008] In order to solve the problems existing in the prior art, the present invention provides a high-speed rail fastener pre-configuration method based on virtual track with high construction efficiency, low fastener replacement rate and low laying cost. This method uses virtual track simulation technology to achieve track scene reconstruction and precise configuration of fasteners during the laying stage of long steel rails on ballastless track of high-speed lines, thereby greatly reducing construction costs.

[0009] A high-speed rail fastener preconfiguration method based on virtual track simulation, comprising:

[0010] S1, data acquisition: collect and obtain the full-element point cloud of the track slab of the line to be fine-tuned, including the track slab point cloud, track support platform point cloud and CPIII point cloud;

[0011] S2, noise filtering: using a filtering algorithm to remove noise and clean the data of the full-factor point cloud;

[0012] S3, target area segmentation: the point cloud obtained in S2 is preliminarily divided into multiple sub-point clouds to form a sub-point cloud set, where each sub-point cloud contains only one rail support; the coordinates of each point in all sub-point clouds in the sub-point cloud set are converted to the virtual track coordinate system with the help of the CPIII control network; the rail support in each sub-point cloud is numbered and recorded; the rail support point cloud is extracted from each sub-point cloud; a standard rail support point cloud in an ideal posture is established according to the virtual track coordinate system {p k}, where k represents the number of the midpoint of the standard rail support point cloud;

[0013] S4, extract the posture and center position of the point cloud of the track support platform;

[0014] S5, virtual track environment establishment: perform step S4 on the rail support point clouds of all other sub-point clouds in S3 in sequence, obtain the pose representation information of all rail support point clouds, and establish the virtual track based on the design parameters of the long rails in the line to be fine-tuned;

[0015] S6, track simulation fine-tuning, specifically includes the following steps:

[0016] S61, calculating the TQI value of the virtual track of the target unit segment;

[0017] S62, simulation fine-tuning principle: perform simulation fine-tuning on the virtual track and output a simulation fine-tuning plan, the steps are as follows:

[0018] S621, determining the reference track of the curve segment, transition curve segment, and straight line segment;

[0019] S622: Using the peak-cutting and valley-filling principle, simulate and adjust the virtual track in S5 to ensure that the overall smoothness index of the virtual track meets the requirements;

[0020] S623: When adjusting the elevation locally, on the basis that the elevation deviation of the reference rail determined in S61 meets the acceptance requirements, the 30m medium wave height irregularity value and the 300m long wave height irregularity value of the reference rail are adjusted in sequence. When both the medium and long wave height irregularity values ​​of the reference rail meet the acceptance requirements, the non-reference rail is adjusted and controlled using level and distortion.

[0021] S644: When adjusting the plane locally, based on the plane deviation of the reference rail determined in S61 meeting the acceptance requirements, the 30m medium-wave and 300m long-wave track irregularities of the reference rail are adjusted in sequence. When both the medium-wave and long-wave track irregularities of the reference rail meet the acceptance requirements, the non-reference rail is adjusted and controlled using the track gauge and track gauge change rate.

[0022] S63, calculating the TQI value of the target unit section of the virtual track after the simulation fine-tuning in S62, and determining whether the TQI value meets the fine-tuning acceptance index: if so, outputting the simulation fine-tuning plan and executing step S64; if not, repeating S62-S63;

[0023] S64, using the adjustable amount of the fastener as a constraint factor and minimizing the change Z of the total adjustment amount in the simulated fine-tuning scheme as the optimization goal, establish an LP model that is a simplification of the integer programming model ILP:

[0024]

[0025] Wherein, σ is the fine adjustment amount corresponding to each rail support platform number to be optimized; α is the global nonlinear optimization factor constant; β is the nonlinear optimization factor constant under local constraints; B is the optimization cutoff setting threshold; and the values ​​of m and n depend on the acceptable degree of optimization.

[0026] S65, performing nonlinear optimization on the change Z of the total adjustment amount in the simulation fine-tuning scheme based on the LP model established in S64 and the branch-and-bound method, and generating a new simulation fine-tuning scheme;

[0027] S7, fastener configuration: configure fastener components according to the new simulation fine-tuning scheme generated in S65.

[0028] In step S3, the origin of the virtual track coordinate system is set at the geometric center of the acquisition device during the first measurement; the Y-axis direction is the extension direction from the small mileage at the origin to the large mileage and is parallel to the track plate; the X-axis direction is the direction from the reference rail at the origin to the non-reference rail and is parallel to the track plate; the Z-axis direction is the direction of the right-hand coordinate system established by the X-axis and the Y-axis.

[0029] In step S3, the method for extracting the track support platform point cloud from each sub-point cloud is: using the random sampling consistency algorithm to eliminate the track plate surface from each sub-point cloud, and then using the minimum bounding box algorithm for each sub-point cloud after eliminating the track plate surface to accurately extract the track support platform point cloud in each sub-point cloud.

[0030] Step S4 includes the following sub-steps:

[0031] S41, subdividing the point cloud structure of the rail support platform: extracting surface point cloud data of the rail support platform point cloud described in S3 based on point and line features, wherein the surface point cloud data includes point cloud data of the rail support surface, point cloud data of the inner and outer jaw surfaces, and point cloud data of the embedded casing center;

[0032] S42, extracting the point cloud posture of the rail support platform, includes the following steps:

[0033] S421, calculate the attitude vector of the rail support platform point cloud: perform implicit surface parameterization on the rail support surface point cloud data obtained in S41 to obtain an implicit surface; solve the normal vector of each point of the implicit surface; normalize the normal vector of each point based on the error averaging concept and Abbe principle, and fit to obtain the three-axis attitude vector

[0034] S422, calculation of the registration vector of the track support platform point cloud posture: using the ICP algorithm to calculate the track support surface point cloud data obtained in S41 and the standard track support platform point cloud data established in S3. k}Perform iterative registration of the nearest neighbor point and output the posture transformation vector

[0035] S423, track support platform point cloud posture correction: If the three-axis posture vector and the posture transformation vector The following conditions are met:

[0036]

[0037] The posture representation of the current support platform point cloud is the three-axis posture vector If the above conditions are not met, then steps S421 to S423 are repeated;

[0038] S43, extracting the center position of the track support platform point cloud, includes the following steps:

[0039] S431, performing data cleaning on the inner and outer jaw surface point cloud data obtained in S41, removing outliers and abnormal values ​​in the inner and outer jaw surface point cloud data, and obtaining the point set {p l}, where l represents the number of the midpoint of the point cloud data of the inner and outer jaw surfaces;

[0040] S432, fitting the point set of the inner and outer jaw surface point cloud data {p l}: the centroid of the point set {p l The total number of points in the point set {p l The coordinates of the t-th point in the virtual orbit coordinate system are Among them, t=1,2,3,…,s, then the point set {p l The center of mass point p in the virtual orbit coordinate system described in S32 l (x l ,y l ,z l ) is fitted to:

[0041]

[0042] S433, track support platform point cloud center position correction: the center of mass point p l (x l ,y l ,z l ) along the three-axis attitude vector when satisfying equation (1) in S423 Vectors in Project it onto the implicit surface corresponding to equation (1) in S423 to obtain the projection intersection point p r (x r ,y r ,z r );

[0043] S434, determining the projection intersection point p r (x r ,y r ,z r ) and the center of mass point p l (x l ,y l ,z l ) is less than the threshold: If yes, the center position of the support platform point cloud is (x r ,y r ,z r ); if not, repeat steps S431 to S434.

[0044] The three-axis attitude vector of the S421 middle, is the comprehensive average vector in the point cloud along the long side of the rail surface; is the comprehensive average vector in the point cloud along the short side of the rail surface; is the comprehensive mean vector on the implicit surface obtained, and and and Form a right-handed coordinate system.

[0045] Preferably, the threshold value described in S434 is 0.25 mm.

[0046] Preferably, the point cloud acquisition device in S1 is a self-moving RC-SLAM high-precision three-dimensional scanning system.

[0047] Preferably, the filtering algorithm in S2 uses a multi-size region statistical filtering algorithm.

[0048] Preferably, in S3, the rail supports in each sub-point cloud are numbered and recorded according to the principle of "small mileage first, large mileage last; left side of the track first, right side of the track last".

[0049] Preferably, in S64, m=5 and n=8.

[0050] Compared with the prior art, the present invention has the following beneficial effects:

[0051] 1. This invention can rely on virtual tracks to carry out simulated fine-tuning before construction, reducing the time spent on actual fine-tuning, avoiding inefficient fine-tuning and repeated work processes, further accelerating construction progress and improving work efficiency;

[0052] 2. Based on virtual track simulation technology, this invention enables pre-configuration of high-speed rail fasteners before long rails are laid, significantly reducing the replacement rate and cost of replacing standard fasteners with fine-tuning fasteners during the original fine-tuning work;

[0053] 3. This invention replaces the sparse sampling point data of the traditional working mode with full-scene reconstruction data of the virtual track, further improving the coverage completeness, information distribution uniformity and feature description capabilities of the data required for fine-tuning, thereby reducing construction safety and quality issues.

[0054] 4. Full-line simulation work is carried out during the long rail laying stage. The method of the present invention can intervene in the dynamic joint debugging and testing stage and the dynamic unevenness acceptance stage in advance, providing a basis for construction quality and cost control. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 is a flow chart of the present invention;

[0056] Figure 2 It is a schematic diagram of all-element point cloud acquisition of the present invention;

[0057] Figure 3 Schematic diagram of the preliminary segmented sub-point cloud of the present invention;

[0058] Figure 4 This is a schematic diagram of redundant data elimination and precise segmentation of the rail support platform of the present invention;

[0059] Figure 5 It is a schematic diagram of the subdivision of the rail support platform structure based on point and line features of the present invention;

[0060] Figure 6 This is a schematic diagram of the orientation of the rail support platform of the present invention;

[0061] Figure 7 It is a schematic diagram of the virtual track establishment result of the present invention. DETAILED DESCRIPTION

[0062] The technical solution of the present invention is further described in detail below with reference to the accompanying drawings and embodiments.

[0063] A high-speed rail fastener pre-configuration method based on virtual track simulation, the process is as follows Figure 1 As shown, the following steps are included:

[0064] S1, data collection: collect and obtain the full-element point cloud on the track plate of the line to be fine-tuned. The collection process is as follows Figure 2 As shown in the figure, the collected full-element point cloud includes the track plate point cloud, the track support platform point cloud, and the CPIII point cloud. The point cloud collection equipment is a self-propelled RC-SLAM high-precision 3D scanning system.

[0065] S2, Noise Filtering: Use a filtering algorithm to remove noise and clean the data of the full-factor point cloud. Noise filtering can prevent the impact of various types of severe environmental noise during high-speed railway construction on the accuracy and precision of the full-factor point cloud. The filtering algorithm should avoid using threshold segmentation filtering algorithms. A multi-scale regional statistical filtering algorithm is used. This filtering algorithm can maximize the integrity of the internal features of the full-factor point cloud with uneven density.

[0066] S3, target area segmentation, specifically includes the following steps:

[0067] S31, point cloud segmentation: Based on the design plan, prior mileage and track plate size parameters, the point cloud obtained in S2 is initially segmented into multiple sub-point clouds. Multiple sub-point clouds constitute a sub-point cloud set, and each sub-point cloud contains only one track support platform. A sub-point cloud in the sub-point cloud set is as follows: Figure 3 shown.

[0068] S32, coordinate system conversion: convert the coordinates of each point in all sub-point clouds in the sub-point cloud set in S31 into the virtual track coordinate system by means of the CPIII control network. Among them, the origin of the virtual track coordinate system is located at the geometric center of the collection device at the first measurement; the Y-axis direction is the extension direction of the small mileage to the large mileage at the origin and is parallel to the track plate; the X-axis direction is the pointing direction of the reference rail to the non-reference rail at the origin and is parallel to the track plate; the Z-axis direction is the right-hand coordinate system direction established by the X-axis and the Y-axis.

[0069] S33, numbering and recording the only rail support platform in all sub-point clouds in the virtual track coordinate system in S32: according to the spatial distribution of the sub-point cloud set, the rail support platform of each sub-point cloud in the sub-point cloud set is numbered and recorded according to the principle of "small mileage first, large mileage later; rail left first, rail right later", wherein the number of each rail support platform does not change in the process of fastener configuration.

[0070] S34, using the random sampling consistency algorithm to respectively remove the track plate surface of each sub-point cloud in the sub-point cloud set in the virtual track coordinate system in S32, and then using the minimum bounding box algorithm to accurately extract the rail support platform point cloud of each sub-point cloud in the sub-point cloud set, as shown in Figure 4 .

[0071] S35, establishing a standard rail support platform point cloud: according to the virtual track coordinate system in S32, a standard rail support platform point cloud {p k} in an ideal attitude is established, wherein k represents the number of points in the standard rail support platform point cloud.

[0072] S4, rail support platform point cloud pose representation, specifically including the following steps:

[0073] S41, rail support platform point cloud structure subdivision: based on point-line feature extraction, the surface point cloud data of the rail support platform point cloud in a sub-point cloud in S34 is extracted, and the extracted surface point cloud data includes rail surface point cloud data, inner and outer jaw surface point cloud data, and embedded sleeve center point cloud data, as shown in Figure 5 .

[0074] S42, rail support platform point cloud attitude extraction, including the following steps:

[0075] S421, calculate the attitude vector of the track support platform point cloud: perform implicit surface parameterization on the track support surface point cloud data obtained in S41, and the implicit surface obtained includes the inner surface and the outer surface; since the track support surface should be a plane under ideal conditions, solve the normal vectors of each point of the implicit surface; based on the error averaging concept (Xie Zhongshu. On the random variation of sampling average error in interval estimation [J]. Statistical Research, 1985, (03): 77-79+69.) and Abbe's principle (Shi Zhaoyao, Zhang Bin, Fei Yetai. Re-understanding of Abbe's principle [J]. Chinese Journal of Scientific Instrument, 2012, 33 (05): 1128-1133.), normalize the normal vectors of each point of the implicit surface and fit it as follows Figure 6 The three-axis attitude vector shown

[0076] Among them, the vector is the comprehensive average vector in the point cloud along the long side of the rail surface; vector is the comprehensive average vector in the point cloud along the short side of the rail surface; vector is the comprehensive average vector on the implicit surface obtained, and the vector With vector and vector Forming a right-handed coordinate system, that is, the vector vector and vector Two by two are perpendicular.

[0077] S422, calculation of the registration vector of the track support platform point cloud posture: using the ICP algorithm to calculate the track support surface point cloud data obtained in S41 and the standard track support platform point cloud data established in S35. k}Perform iterative registration of the nearest neighbor point and output the posture transformation vector

[0078] S423, Track support platform point cloud posture correction: If the three-axis posture vector in S421 and the posture transformation vector The following conditions are met:

[0079]

[0080] The posture representation of the current support platform point cloud is the three-axis posture vector If the above conditions are not met, then steps S421 to S423 are repeated;

[0081] S43, extracting the center position of the track support platform point cloud, includes the following steps:

[0082] S431, cleaning the inner and outer jaw face point cloud data: cleaning the inner and outer jaw face point cloud data obtained in S41, eliminating outliers and abnormal values in the inner and outer jaw face point cloud data, and obtaining a point set {p l} of the inner and outer jaw face point cloud data, wherein l represents the number of points in the inner and outer jaw face point cloud data.

[0083] S432, fitting the centroid of the point set {p l} of the inner and outer jaw face point cloud data: the total number of points in the point set {p l} is s, and the coordinate of the tth point in the point set {p l} in the virtual track coordinate system described in S32 is wherein t = 1, 2, 3, …, s, and the centroid point p l of the point set {p l} in the virtual track coordinate system described in S32 is fitted as: l l l

[0084]

[0085] S433, correcting the center position of the rail support point cloud: projecting the centroid point p l (x l ,y l ,z l ) along the vector in the three-axis attitude vector satisfying formula (1) in S423 to the corresponding implicit surface when formula (1) in S423 is satisfied, to obtain a projection intersection point p r (x r ,y r ,z r ).

[0086] S434, judging whether the Euclidean distance between the projection intersection point p r (x r ,y r ,z r ) and the centroid point p l (x l ,y l ,z l ) is less than a threshold value: if yes, the center position of the rail support point cloud is (x r ,y r ,z r ); if not, repeating steps S431-S434, wherein as a preferred, the threshold value is 0.25 mm.

[0087] ​​​S5, virtual track environment establishment: perform step S4 on the support platform point clouds of all other sub-point clouds in S34 in sequence, and obtain the position and posture representation information of all support platform point clouds (the three-axis posture vector of the support platform point cloud The center position of the point cloud of the support platform is (x r ,y r ,z r )) and, combined with the design parameters of the long rails in the line to be fine-tuned, create a virtual track, such as Figure 7 shown.

[0088] S6, track simulation fine-tuning, specifically includes the following steps:

[0089] S61, calculate the TQI value of the virtual track of the target unit section: calculate the TQI value of the virtual track in S5 within the length of the unit section. The TQI value is the sum of the standard deviations of the seven geometric deviations within the length of the unit section, including height (left and right tracks), track direction (left and right tracks), gauge, level (superelevation), and triangular pit. The calculation formula is as follows:

[0090]

[0091] Where, σ i is the standard deviation of any one of the seven geometric deviations within the unit segment length:

[0092]

[0093] Where N is the number of centroids of all rail supports within the unit section length, x ij represents the amplitude of the center of mass of any rail support platform, Represents the amplitude x of the center of mass of N rail supports ij The arithmetic mean of .

[0094] The length of the corresponding unit section at different speeds is different, and the standards are as follows: the length of the unit section at a speed of 350km / h is 200m; the length of the unit section in the speed range of 200km / h to 350km / h is 150m; the length of the unit section at a speed less than 200km / h is 100m.

[0095] S62, simulation fine-tuning principle: Perform simulation fine-tuning on the virtual track described in S5 and output a simulation fine-tuning plan, specifically including the following steps:

[0096] S621, determine the reference track:

[0097] Determine the reference rail for curved and transition curve sections: When optimizing the track direction, the high rail (where there is a superelevation design in the curved section) is usually selected as the reference rail for curved and transition curve sections. When optimizing the height, the low rail should be selected as the reference rail for curved sections.

[0098] Determine the reference rail for the straight segment: When optimizing the track direction and height of the straight segment, the selected reference rail and the reference rail of the closest curved segment in the direction of short mileage should be the same rail;

[0099] S622: Overall, waveforms of the seven geometric deviations of the virtual track are generated based on the seven geometric deviations, and the waveforms are analyzed. Using a principle similar to "peak shaving and valley filling", the virtual track in S5 is simulated and adjusted to ensure that the track smoothness indicators of the virtual track are qualified as a whole: the actual direction of the line is as consistent as possible with the designed line shape, achieving a state of straight lines, smooth curves, no sudden changes, and no periodic small oscillations.

[0100] S623: When adjusting the elevation locally, on the basis that the elevation deviation of the reference rail determined in S61 meets the acceptance requirements, the 30m medium wave height irregularity value and the 300m long wave height irregularity value of the reference rail are adjusted in sequence. When the medium and long wave height irregularity values ​​of the reference rail have met the acceptance requirements, the non-reference rail is adjusted and controlled using the level (superelevation) and distortion.

[0101] S644: When adjusting the plane locally, on the basis that the plane deviation of the reference rail determined in S61 meets the acceptance requirements, the 30m medium-wave track irregularity value and the 300m long-wave track irregularity value of the reference rail are adjusted in sequence. When the medium-wave and long-wave track irregularities of the reference rail have met the acceptance requirements, the non-reference rail is adjusted and controlled using the track gauge and track gauge change rate.

[0102] After the virtual track is simulated and fine-tuned, the simulation fine-tuning plan is output to facilitate the subsequent determination of the fastener configuration plan:

[0103] Elevation adjustment: When the actual rail surface is higher than the reference rail surface, the adjustment amount sign is "minus (-)", indicating that the rail needs to be adjusted downward; when the actual rail surface is lower than the reference rail surface, the adjustment amount sign is "positive (+)", indicating that the rail needs to be adjusted upward.

[0104] Plane adjustment: When the actual position of the rail is on the right side of the design position, the adjustment amount sign is "minus (-)", indicating that the rail needs to be adjusted to the left. When the actual position of the rail is on the left side of the design position, the adjustment amount sign is "plus (+)", indicating that the rail needs to be adjusted to the right. The left and right sides are defined as follows: the forward direction of the track is from small mileage to large mileage, and the left and right sides of the track are defined according to the forward direction of the track.

[0105] S63, according to the TQI calculation formula described in S61, calculate the TQI value of the target unit section of the virtual track after the simulation fine-tuning in S62, and determine whether the TQI value meets the fine-tuning acceptance index: if it does, output the simulation fine-tuning plan and execute step S64; if it does not, repeat S62-S63.

[0106] S64, taking the adjustable amount of the fastener as the constraint factor and minimizing the change Z of the total adjustment amount in the simulation fine-tuning scheme as the optimization goal, an integer programming model (ILP) is established (Luo Yucai. A hybrid algorithm for integer linear programming [J]. Journal of Guizhou University (Natural Science Edition), 1987, (04): 193-199.) The simplified LP model is:

[0107]

[0108] Among them, σ is the fine-tuning amount corresponding to each rail platform number to be optimized; α is the global nonlinear optimization factor constant; β is the nonlinear optimization factor constant under local constraints; B is the optimization cutoff setting threshold; Among them, the values ​​of m and n depend on the acceptable degree of optimization. In the embodiment of the present invention, m=5 and n=8.

[0109] The reason for using an integer programming model to optimize the simulated fine-tuning scheme is that the adjustment amount given in the simulated fine-tuning scheme output by the simulated fine-tuning principle does not take into account the specification constraints of the track pads / gauge blocks. Since ballastless track fine-tuning is achieved by increasing or decreasing the number of track pads / gauge blocks, the final track adjustment effect is limited by the experience and level of the on-site fine-tuning personnel. At the same time, since the adjustable amount of the supporting fasteners of ballastless track rails is limited, if the track is fine-tuned according to the horizontal and vertical deviations of the track obtained by the design values ​​of the track geometric parameters, the adjustment amount will exceed the adjustable range of the fasteners. Therefore, on the basis of ensuring the smoothness of the virtual track, the integer programming model is used to further optimize the simulated fine-tuning scheme, giving the minimum number of track pads / gauge blocks that need to be adjusted at each sleeper, so as to avoid the impact of human intervention on the track adjustment effect during on-site fine-tuning, making the simulated fine-tuning scheme more consistent with on-site operation conditions.

[0110] S65, based on the LP model established in S64 and the branch and bound method, nonlinear optimization is performed on the change Z of the total adjustment amount in the simulation fine-tuning scheme: the calculation is started from the relaxation problem without considering the integer constraints of the decision variables. If the optimal solution obtained does not meet the integer constraints, the original problem is decomposed into several problems (branch problems), and the original feasible solution area is compressed by continuously adding new constraints (constraints determined by the integer requirements) on the basis of the original constraints, gradually approaching the integer optimal solution until the TQI value of the target segment and other constraint indicators meet the acceptance requirements, and a new simulation fine-tuning scheme is generated.

[0111] S7, fastener configuration: configure fastener parts according to the new simulation fine-tuning scheme generated by S65: select the specifications and corresponding models of adjustable parts, and purchase, distribute and install fastener parts at the construction site according to the scheme.

[0112] Preferably, the method further comprises:

[0113] S8, Process Review During Fine-Tuning: Use a track irregularity evaluation model based on the vector distance difference method to verify the irregularity of the fine-tuning solution obtained above. If the irregularity value exceeds the target end acceptance standard, re-execute step S6 and modify the fine-tuning solution.

Claims

1. A high-speed rail fastener pre-configuration method based on virtual track simulation, characterized in that include: S1, data acquisition: collect and obtain the full-element point cloud of the track slab of the line to be fine-tuned, including the track slab point cloud, track support platform point cloud and CPIII point cloud; S2, noise filtering: using a filtering algorithm to remove noise and clean the data of the full-factor point cloud; S3, target area segmentation: the point cloud obtained in S2 is preliminarily divided into multiple sub-point clouds to form a sub-point cloud set, where each sub-point cloud contains only one rail support; the coordinates of each point in all sub-point clouds in the sub-point cloud set are converted to the virtual track coordinate system with the help of the CPIII control network; the rail support in each sub-point cloud is numbered and recorded; the rail support point cloud is extracted from each sub-point cloud; a standard rail support point cloud in an ideal posture is established according to the virtual track coordinate system {p k }, where k represents the number of the midpoint of the standard rail support point cloud; S4, extract the posture and center position of the point cloud of the track support platform; S5, virtual track environment establishment: perform step S4 on the rail support point clouds of all other sub-point clouds in S3 in sequence, obtain the pose representation information of all rail support point clouds, and establish the virtual track based on the design parameters of the long rails in the line to be fine-tuned; S6, track simulation fine-tuning, specifically includes the following steps: S61, calculating the TQI value of the virtual track of the target unit segment; S62, simulation fine-tuning principle: perform simulation fine-tuning on the virtual track and output a simulation fine-tuning plan, the steps are as follows: S621, determining the reference track of the curve segment, transition curve segment, and straight line segment; S622: Using the peak-cutting and valley-filling principle, simulate and adjust the virtual track in S5 to ensure that the overall smoothness index of the virtual track meets the requirements; S623: When adjusting the elevation locally, on the basis that the elevation deviation of the reference rail determined in S61 meets the acceptance requirements, the 30m medium wave height irregularity value and the 300m long wave height irregularity value of the reference rail are adjusted in sequence. When both the medium and long wave height irregularity values ​​of the reference rail meet the acceptance requirements, the non-reference rail is adjusted and controlled using level and distortion. S644: When adjusting the plane locally, based on the plane deviation of the reference rail determined in S61 meeting the acceptance requirements, the 30m medium-wave and 300m long-wave track irregularities of the reference rail are adjusted in sequence. When both the medium-wave and long-wave track irregularities of the reference rail meet the acceptance requirements, the non-reference rail is adjusted and controlled using the track gauge and track gauge change rate. S63, calculating the TQI value of the target unit section of the virtual track after the simulation fine-tuning in S62, and determining whether the TQI value meets the fine-tuning acceptance index: if so, outputting the simulation fine-tuning plan and executing step S64; if not, repeating S62-S63; S64, using the adjustable amount of the fastener as a constraint factor and minimizing the change Z of the total adjustment amount in the simulated fine-tuning scheme as the optimization goal, establish an LP model that is a simplification of the integer programming model ILP: Wherein, σ is the fine adjustment amount corresponding to each rail support platform number to be optimized; α is the global nonlinear optimization factor constant; β is the nonlinear optimization factor constant under local constraints; B is the optimization cutoff setting threshold; and the values ​​of m and n depend on the acceptable degree of optimization. S65, performing nonlinear optimization on the change Z of the total adjustment amount in the simulation fine-tuning scheme based on the LP model established in S64 and the branch-and-bound method, and generating a new simulation fine-tuning scheme; S7, fastener configuration: configure fastener components according to the new simulation fine-tuning scheme generated in S65.

2. The high-speed rail fastener pre-configuration method according to claim 1, characterized in that: The origin of the virtual track coordinate system described in S3 is set at the geometric center of the acquisition device during the first measurement; the Y-axis direction is the extension direction from the small mileage at the origin to the large mileage and is parallel to the track plate; the X-axis direction is the direction from the reference rail at the origin to the non-reference rail and is parallel to the track plate; the Z-axis direction is the direction of the right-hand coordinate system established by the X-axis and the Y-axis.

3. The high-speed rail fastener pre-configuration method according to claim 1, characterized in that: The method for extracting the track support platform point cloud from each sub-point cloud in S3 is: use the random sampling consistency algorithm to eliminate the track plate surface from each sub-point cloud, and then use the minimum bounding box algorithm on each sub-point cloud after eliminating the track plate surface to accurately extract the track support platform point cloud in each sub-point cloud.

4. The high-speed rail fastener pre-configuration method according to claim 3, characterized in that: S4 includes the following sub-steps: S41, subdividing the point cloud structure of the rail support platform: extracting surface point cloud data of the rail support platform point cloud described in S3 based on point and line features, wherein the surface point cloud data includes point cloud data of the rail support surface, point cloud data of the inner and outer jaw surfaces, and point cloud data of the embedded casing center; S42, extracting the point cloud posture of the rail support platform, includes the following steps: S421, calculate the attitude vector of the rail support platform point cloud: perform implicit surface parameterization on the rail support surface point cloud data obtained in S41 to obtain an implicit surface; solve the normal vector of each point of the implicit surface; normalize the normal vector of each point based on the error averaging concept and Abbe principle, and fit to obtain the three-axis attitude vector S422, calculation of the registration vector of the track support platform point cloud posture: using the ICP algorithm to calculate the track support surface point cloud data obtained in S41 and the standard track support platform point cloud data established in S3. k }Perform iterative registration of the nearest neighbor point and output the posture transformation vector S423, track support platform point cloud posture correction: If the three-axis posture vector and the posture transformation vector The following conditions are met: The posture representation of the current support platform point cloud is the three-axis posture vector If the above conditions are not met, then steps S421 to S423 are repeated; S43, extracting the center position of the track support platform point cloud, includes the following steps: S431, performing data cleaning on the inner and outer jaw surface point cloud data obtained in S41, removing outliers and abnormal values ​​in the inner and outer jaw surface point cloud data, and obtaining the point set {p l }, where l represents the number of the midpoint of the point cloud data of the inner and outer jaw surfaces; S432, fitting the point set of the inner and outer jaw surface point cloud data {p l }: the centroid of the point set {p l The total number of points in the point set {p l The coordinates of the t-th point in the virtual orbit coordinate system are Among them, t=1,2,3,…,s, then the point set {p l The center of mass point p in the virtual orbit coordinate system described in S32 l (x l ,y l ,z l ) is fitted to: S433, track support platform point cloud center position correction: the center of mass point p l (x l ,y l ,z l ) along the three-axis attitude vector when satisfying equation (1) in S423 Vectors in Project it onto the implicit surface corresponding to equation (1) in S423 to obtain the projection intersection point p r (x r ,y r ,z r ); S434, determining the projection intersection point p r (x r ,y r ,z r ) and the center of mass point p l (x l ,y l ,z l ) is less than the threshold: If yes, the center position of the support platform point cloud is (x r ,y r ,z r ); if not, repeat steps S431 to S434.

5. The high-speed rail fastener pre-configuration method according to claim 4, characterized in that: S421's three-axis attitude vector middle, is the comprehensive average vector in the point cloud along the long side of the rail surface; is the comprehensive average vector in the point cloud along the short side of the rail surface; is the comprehensive mean vector on the implicit surface obtained, and and and Form a right-handed coordinate system.

6. The high-speed rail fastener pre-configuration method according to claim 4, characterized in that: The threshold value described in S434 is 0.25 mm.

7. The high-speed rail fastener pre-configuration method according to claim 1, characterized in that: The point cloud acquisition device in S1 is a self-moving RC-SLAM high-precision three-dimensional scanning system.

8. The high-speed rail fastener pre-configuration method according to claim 1, characterized in that: The filtering algorithm described in S2 uses a multi-size regional statistical filtering algorithm.

9. The high-speed rail fastener pre-configuration method according to claim 1, characterized in that: In S3, the rail supports in each sub-point cloud are numbered and recorded according to the principle of "small mileage first, large mileage last; left track first, right track last".

10. The high-speed rail fastener pre-configuration method according to claim 1, characterized in that: In S64, m=5 and n=8.

Images