Track point cloud pose correction method and application and measurement equipment and system

Abstract

The invention relates to the technical field of track detection, and provides a track point cloud pose correction method and application and a measuring device and system.The track point cloud pose correction method comprises the following steps that a local coordinate system of three-dimensional point cloud 3DPC of a track is aligned to a coordinate system of a standard track point cloud model; the aligned three-dimensional point cloud (3DPC I) is obtained; and according to at least one of the X value, the Y value and the Z value of the 3DPC I, judging whether the 3DPC I meets a pose correction requirement, and if the pose correction requirement is met, obtaining a three-dimensional point cloud 3DPC II after pose correction. The problem that the track geometric state detection error is large can be solved.

Description

Technical Field

[0001] The present invention relates to the field of track detection technology, and in particular to a track point cloud posture correction method and application, and a measuring device and system. Background Art

[0002] Under the background of high-speed railway operation, the safety of railway transportation has become increasingly the focus. Regular inspection of the main performance parameters of the track and formulation of corresponding maintenance and overhaul plans based on the inspection results are the key means to ensure the safety of railway transportation and an important guarantee for the safe operation of trains.

[0003] However, current track inspection methods mainly rely on traditional measuring tools such as steel rulers, track gauges, offset gauges, wear gauges, chords, etc. to collect data and compare them with standard data. On the one hand, these traditional measuring tools are easily interfered by external environmental factors such as obstacles, resulting in low measurement efficiency or even inability to measure, and measurement accuracy is difficult to guarantee; on the other hand, traditional measuring tools are highly dependent on manual labor, which consumes a lot of manpower, and the measurement results are inaccurate and unreliable. Summary of the invention

[0004] The inventor of the present invention has found that the surface structure optical three-dimensional measurement technology is an emerging incoherent light three-dimensional sensing technology for three-dimensional shape measurement of diffuse reflection objects, which has the advantages of fast speed, non-contact, high precision, etc.; the existing surface structure optical three-dimensional measurement technology needs to project grating stripes on the surface of the object first, and obtain the object image with grating stripes through the visual sensor, and then decode the image and calculate the three-dimensional coordinates of all pixels on the image, so as to obtain the three-dimensional point cloud of the object surface within the visual sensor's field of view. When applied to track detection, the track three-dimensional point cloud data decoded from the image will be affected by the relative position relationship such as the distance and angle of the visual sensor relative to the track; and the relative position relationship between the visual sensor and the track will be affected by various factors, such as the performance of the visual sensor, the operating habits of the operator, and the image shooting environment, and it is difficult to standardize, resulting in unclear position and posture of the three-dimensional point cloud data decoded from different images in space, and it is difficult to compare with the standard track point cloud data, which leads to large errors when the geometric state of the track is detected using the existing surface structure optical three-dimensional measurement technology.

[0005] The purpose of the present invention is to solve the problem of large errors in track geometry state detection and to provide a track point cloud posture correction method and application and a measuring device and system.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is:

[0007] A method for correcting the position and posture of a track point cloud comprises the following steps:

[0008] S1, obtain the three-dimensional point cloud 3DPC of the track; fit the 3DPC into a plane point cloud PPC;

[0009] S2. Calculate the RN vector that measures the average statistical value of the normal vector of each point in PPC, and establish the characteristic plane P of PPC based on the RN vector ⊥ and P || ; Calculate P ⊥ The rotation matrix T of the XOZ plane relative to the standard orbit point cloud model ⊥ ; Calculate P || The rotation matrix T of the YOZ plane relative to the standard orbit point cloud model || , use T ⊥ and T || Perform rotation transformation on 3DPC; calculate the X value of the centroid of 3DPC. If the X value of the centroid is less than the preset threshold, rotate 3DPC 180° around the Y axis to obtain the processed three-dimensional point cloud 3DPCⅠ;

[0010] S3. Determine whether 3DPCⅠ meets the posture correction requirement according to at least one of the X value, Y value, and Z value of the processed three-dimensional point cloud 3DPCⅠ. If the posture correction requirement is met, obtain the three-dimensional point cloud 3DPCⅡ after posture correction.

[0011] As a preferred embodiment of the present invention, in step S1, the step of fitting 3DPC into a plane point cloud PPC includes dividing 3DPC into t pieces of point clouds 3DPC[i], 0≤i<t, and fitting 3DPC[i] into t pieces of plane point clouds PPC[i]; in step S2, 3DPC is subjected to a rotation transformation based on PPC[i] to obtain a processed three-dimensional point cloud 3DPCⅠ[i]; in step S3, whether 3DPCⅠ[i] meets the posture correction requirements is determined. If the posture correction requirements are met, the posture-corrected three-dimensional point cloud 3DPCⅡ is obtained.

[0012] As a preferred embodiment of the present invention, step S3 comprises the following steps:

[0013] S31, comparing the similarity of each 3DPCⅠ[i] with the corresponding part of the standard track point cloud model;

[0014] S32. According to the similarity between each 3DPCⅠ[i] and the corresponding part of the standard track point cloud model, the pose correction requirement is judged in descending order; if a 3DPCⅠ[i] that meets the pose correction requirement appears, then 3DPCⅠ meets the pose correction requirement and 3DPCⅡ is obtained.

[0015] As a preferred solution of the present invention, in step S1, PPC[i] is obtained by performing plane model segmentation on 3DPC[i].

[0016] As a preferred embodiment of the present invention, in step S3, the condition for 3DPCⅠ to meet the posture correction requirement is that the maximum Y value of 3DPCⅠ is less than or equal to the preset threshold; the condition for 3DPCⅠ[i] to meet the posture correction requirement is that the maximum Y value of at least one 3DPCⅠ[i] is less than or equal to the preset threshold.

[0017] As a preferred solution of the present invention, the 3DPC in step S1 is the data of the rail head and the rail waist.

[0018] As a preferred solution of the present invention, in step S1, the data of the rail head and rail waist are selected by automatic area selection and / or manual area selection.

[0019] As a preferred solution of the present invention, the 3DPC is downsampled before step S1.

[0020] As a preferred embodiment of the present invention, step S2 comprises the following steps:

[0021] Calculate the normal vector N[i] of each point P[i] in PPC, 0≤i<n, n is the number of points in PPC; calculate the RN vector that measures the average statistical value of N[i]; calculate the positional relationship d[i] of P[i] to the RN vector respectively, let the maximum value in d[i] be d[s], and the point corresponding to d[s] be P[s]; use RN vector and P[s] to establish plane P ⊥ and plane P || , where P ⊥ Perpendicular to RN vector, P || Parallel to the RN vector;

[0022] Calculate P ⊥ The angle θ between the standard orbit point cloud model coordinate system plane XOZ and the corresponding rotation axis vector V ⊥ , calculate around V ⊥ The rotation matrix T of the rotation θ ⊥ ; Calculate P || The angle between the YOZ plane and the standard track point cloud model coordinate system and the corresponding rotation axis vector V || , calculate around V || Rotation The rotation matrix T || ;

[0023] Using T ⊥ and T || Perform a rotation transformation on the 3DPC; calculate the X value of the centroid of the 3DPC. If the X value of the centroid is less than a preset threshold, rotate the 3DPC 180° around the Y axis to obtain the processed three-dimensional point cloud 3DPCⅠ.

[0024] An application of a method for correcting the position and posture of a track point cloud includes the following steps:

[0025] A. Obtaining 3DPCⅡ by a posture correction method of a track point cloud of the present invention;

[0026] B. According to the type of track identified by 3DPCⅡ, select the corresponding standard track point cloud model to perform deviation analysis on 3DPCⅡ and obtain the deviation analysis result of 3DPCⅡ.

[0027] As a preferred embodiment of the present invention, step B comprises the following steps:

[0028] B1. Roughly align 3DPCⅡ with the corresponding standard track point cloud model;

[0029] B2. Accurately align 3DPCⅡ with the corresponding standard track point cloud model;

[0030] B3. Calculate the deviation between 3DPCⅡ and the corresponding standard track point cloud model.

[0031] As a preferred solution of the present invention, the rail head and rail waist in 3DPCⅡ are segmented before step B1; in step B1, the centroid of the rail head of 3DPCⅡ is aligned with the centroid of the rail head of the standard track point cloud model, or the centroid of the rail waist of 3DPCⅡ is aligned with the centroid of the rail waist of the standard track point cloud model.

[0032] As a preferred embodiment of the present invention, in step B, m sections 3DPCS[i] are cut out from 3DPCⅡ along the Z axis, 0≤i<m, and multiple standard sections corresponding to 3DPCS[i] are cut out from the standard track point cloud model; the similarity between 3DPCS[i] and the standard sections is compared one by one to identify the control section; and the corresponding standard track point cloud model is selected according to the control section.

[0033] As a preferred solution of the present invention, it also includes establishing a track database; the track database at least stores the deviation analysis results of 3DPCⅡ. Further, the deviation analysis results of 3DPCⅡ include track geometric parameter curves obtained by traversing the control section along the track longitudinal direction.

[0034] As a preferred solution of the present invention, a track geometry parameter trend graph is established based on the deviation analysis results of 3DPCⅡ in the track database, and a maintenance plan is formulated based on the track geometry parameter trend graph.

[0035] As a preferred embodiment of the present invention, the following steps are included when establishing a track geometry parameter trend graph:

[0036] Establish a track geometry parameter line graph based on historical data in the track database; use the prediction model to predict track geometry parameters; and establish a track geometry parameter trend graph.

[0037] A three-dimensional track geometric state measuring device comprises a processor and a memory; the memory is communicatively connected to the processor; the memory stores instructions that can be executed by the processor, and the processor can execute a track point cloud posture correction method of the present invention through the instructions.

[0038] A three-dimensional track geometric state measurement system comprises a three-dimensional track geometric state measurement device of the present invention, a surface structured light emitter and a visual sensor; the surface structured light emitter is used to project surface structured light onto the track; the visual sensor is used to obtain a track image with grating stripes; the visual sensor is communicatively connected to the processor.

[0039] As a preferred embodiment of the present invention, it also includes a crossbeam and a wheel assembly; the wheel assemblies are arranged on both sides of the crossbeam along the horizontal direction of the track; at least one of the wheel assemblies is connected to the crossbeam through an adaptive device so that it can swing around the vertical axis relative to the crossbeam; the spacing between the wheel assemblies corresponds to the track gauge of the track; the wheel assembly can move along the track; the surface structured light emitter and the visual sensor are connected to the crossbeam.

[0040] As a preferred embodiment of the present invention, the adaptive device includes a first mounting seat and a second mounting seat, the first mounting seat being capable of rotating relative to the second mounting seat along a vertical axis; one of the first mounting seat and the second mounting seat is connected to the crossbeam, and the other of the first mounting seat and the second mounting seat is connected to the wheel assembly.

[0041] As a preferred solution of the present invention, the other end of the adaptive device is slidably connected to the wheel assembly or the crossbeam, and the sliding connection includes a slide rail whose axis is arranged transversely along the track.

[0042] As a preferred solution of the present invention, the surface structured light emitter and the visual sensor are connected to the crossbeam via a cantilever connecting rod; the cantilever connecting rod is used to adjust the position and angle of the surface structured light emitter and the visual sensor relative to the track.

[0043] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0044] 1. The track point cloud posture correction method of the present invention includes the step of aligning the characteristic plane of the track three-dimensional point cloud 3DPC to the standard track point cloud model coordinate system before identifying and measuring the track three-dimensional point cloud 3DPC, so as to obtain the three-dimensional point cloud 3DPCⅡ after posture correction. Compared with the existing scheme of directly using 3DPC, the present scheme can eliminate the measurement deviation caused by the offset and / or tilt of the 3DPC local coordinate system relative to the standard track point cloud model coordinate system in subsequent use, thereby improving the recognition accuracy and measurement precision of 3DPCⅡ in subsequent processing, and thus enabling the surface structure optical three-dimensional measurement technology to be successfully applied to track detection;

[0045] At the same time, since this solution can eliminate the offset and / or tilt between the 3DPC local coordinate system and the standard track point cloud model coordinate system, there is no need to fix the relative position relationship between the visual sensor and the track when using the visual sensor to capture the track image, so that the staff can freely arrange the shooting distance and angle according to factors such as the visual sensor performance and their own operating habits, thereby improving the efficiency of the shooting link in track detection;

[0046] Furthermore, this scheme can use RN vectors to establish plane P ⊥ and plane P || , thus comprehensively considering all points in PPC, avoiding the influence of a single bad point on the normal calculation, and also avoiding the situation where some points cannot be taken into account when directly calculating the normal of the fitting plane, making this scheme more robust; at the same time, the RN vector is used to establish the plane P ⊥ and plane P || Measuring the characteristic orientation of 3DPC can also make the orientation description more accurate and improve the accuracy of this scheme;

[0047] When rotating the 3DPC, this solution can use the rotation matrix T ⊥ and T || The method of rotating the 3DPC to ensure that the plane P ⊥ and plane P || They are respectively coincident with the XOZ plane and YOZ plane of the standard track point cloud model to eliminate the measurement deviation caused by the offset and / or tilt of the 3DPC coordinate system relative to the standard track point cloud model, thus ensuring the accuracy of this solution;

[0048] At the same time, this scheme also verifies the X value of the centroid of the 3DPC after the rotation transformation, so as to ensure that the extension direction of the 3DPC after the rotation transformation is consistent with the extension direction of the Z axis, and avoid the situation where the local coordinate system of the 3DPC after the rotation transformation is parallel to the coordinate system of the standard track point cloud model, but the direction is inconsistent;

[0049] Finally, this solution also includes the step of judging whether 3DPCⅠ meets the requirements of posture correction, further ensuring the validity of 3DPCⅠ and avoiding errors such as calculation errors or the original data 3DPC not containing the required rail head or rail waist.

[0050] Ⅰ Situations where posture correction requirements are not met.

[0051] 2. Application of the position and posture correction method of the track point cloud of the present invention. Due to the use of the position and posture correction method of the track point cloud of the present invention, the local coordinate system of 3DPCⅡ has no offset and tilt with the standard track point cloud model coordinate system, and has higher accuracy when identifying and analyzing deviations of 3DPCⅡ.

[0052] 3. The track three-dimensional geometric state measurement device of the present invention includes a memory storing a method for executing the track point cloud posture correction method of the present invention and a corresponding processor. When measuring the track using the surface structure optical three-dimensional measurement technology, it can perform track type identification and track geometric parameter measurement based on 3DPCⅡ to improve the identification accuracy and reduce the measurement deviation.

[0053] 4. The track three-dimensional geometric state measurement system of the present invention can obtain the track three-dimensional point cloud 3DPC and obtain 3DPCⅡ through processing by the track three-dimensional geometric state measurement equipment, and then perform track type identification and track geometric parameter measurement based on 3DPCⅡ to improve identification accuracy and reduce measurement deviation. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is a schematic flow chart of a method for correcting the position and posture of a track point cloud of the present invention;

[0055] Figure 2 This is a schematic diagram of the application process of a trajectory point cloud posture correction method of the present invention. Figure 1 ;

[0056] Figure 3 This is a schematic diagram of the application process of a trajectory point cloud posture correction method of the present invention. Figure 2 ;

[0057] Figure 4 The three-dimensional structure diagram of a track three-dimensional geometric state measurement system of the present invention is shown in FIG. Figure 1 ;

[0058] Figure 5 The three-dimensional structure diagram of a track three-dimensional geometric state measurement system of the present invention is shown in FIG. Figure 2 ;

[0059] Figure 6 It is a schematic diagram of a partial three-dimensional structure of a track three-dimensional geometric state measurement system of the present invention installed on a track state and viewed from the inner side of the track to the outer side of the track;

[0060] Figure 7 It is a schematic diagram of the exploded three-dimensional structure of a track three-dimensional geometric state measurement system of the present invention;

[0061] Figure 8 yes Figure 5 The enlarged schematic diagram of the local three-dimensional structure at position Ⅰ in the middle;

[0062] Fig. 9 It is a schematic diagram of the explosion of the local three-dimensional structure at the adaptive device;

[0063] Fig.10 is a three-dimensional structural schematic diagram of a wheel assembly;

[0064] Fig.11 It is a schematic diagram of the partial three-dimensional structure of the cantilever connecting rod;

[0065] Icons: 1-track; 2-crossbeam; 3-wheel assembly; 4-gauge adjustment mechanism; 5-adaptive device; 6-cantilever connecting rod; 21-first mounting seat; 211-recessed portion; 31-traveling wheel; 32-step shaft; 33-slide rail; 34-guide wheel; 35-limiting block; 51-second mounting seat; 511-protruding portion; 52-slider; 53-connecting piece; 61-connecting rod; 62-equipment platform. DETAILED DESCRIPTION

[0066] The present invention will be described in detail below in conjunction with the accompanying drawings.

[0067] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0068] Example 1

[0069] like Figure 1 As shown, a method for correcting the position and posture of a track point cloud adopted in this embodiment comprises the following steps:

[0070] S1, obtain the three-dimensional point cloud 3DPC of track 1; fit the 3DPC into a plane point cloud PPC;

[0071] S2. Calculate the RN vector that measures the average statistical value of the normal vector of each point in PPC, and establish the characteristic plane P of PPC based on the RN vector ⊥ and P || ; Calculate P ⊥ The rotation matrix T of the XOZ plane relative to the standard orbit point cloud model ⊥ ; Calculate P || The rotation matrix T of the YOZ plane relative to the standard orbit point cloud model|| , use T ⊥ and T || Perform rotation transformation on 3DPC; calculate the X value of the centroid of 3DPC. If the X value of the centroid is less than the preset threshold, rotate 3DPC 180° around the Y axis to obtain the processed three-dimensional point cloud 3DPCⅠ;

[0072] S3. Determine whether 3DPCⅠ meets the posture correction requirement according to at least one of the X value, Y value, and Z value of the processed three-dimensional point cloud 3DPCⅠ. If the posture correction requirement is met, obtain the three-dimensional point cloud 3DPCⅡ after posture correction.

[0073] The three-dimensional point cloud 3DPC in step S1 can be obtained by measurement or directly input from existing data; various methods can be used in step S1 to fit the 3DPC into a plane point cloud PPC, such as principal component analysis and least squares plane fitting;

[0074] The specific type of track 1 depends on the detection requirements. Taking the detection of turnout as an example, track 1 may include structures such as point rails, frogs, guard rails, joint rails, and rails. According to different 3DPC acquisition methods, 3DPC may contain some other interference information in addition to the point cloud information of track 1, such as point cloud information of other structures such as the ground, plywood, and bolts. The shape of this part of the point cloud is changeable, which is not conducive to posture correction. Therefore, the 3DPC can be screened before step S1 to filter out other point clouds except the required point cloud, so that the 3DPC in step S1 is the rail head. and rail waist data; specific methods such as automatic selection and / or manual selection can be used for screening, and the rules of automatic selection can be determined based on experience; manual selection is visually determined by the operator; specific selection methods can be frame selection, lasso selection, etc.; automatic selection and manual selection can be performed either at the same time, for example, manual selection can be used as a backup when automatic selection fails; when selecting, a relatively flat rail head tread should be selected; if the 3DPC does not contain interference information, or the interference information has been excluded, automatic selection or manual selection may not be performed.

[0075] At the same time, the information density of 3DPC is generally large. In order to reduce the amount of calculation in subsequent steps, this embodiment also downsamples the 3DPC before step S1, thereby improving the point cloud data processing speed in subsequent steps; when step S1 also includes a 3DPC screening step, the selected 3DPC can be downsampled after screening the 3DPC, or the 3DPC can be screened after all 3DPCs are downsampled; if the information density of 3DPC is small, such as using a visual sensor with lower pixels to shoot track 1, the 3DPC does not need to be downsampled; specifically, the downsampling of this embodiment selects voxel filtering sampling or uniform sampling.

[0076] Step S2 is used to align the local coordinate system of 3DPC with the coordinate system of the standard track point cloud model; in this embodiment, the standard track point cloud model adopts a Cartesian coordinate system including X-axis, Y-axis and Z-axis, and the X-axis corresponds to the track horizontal direction of the standard track point cloud model; the Y-axis corresponds to the track height direction of the standard track point cloud model; and the Z-axis corresponds to the track longitudinal direction of the standard track point cloud model.

[0077] In step S2, P is calculated ⊥ The rotation matrix T of the XOZ plane relative to the standard orbit point cloud model ⊥ , we can first calculate P ⊥ The angle θ relative to the XOZ plane of the standard orbit point cloud model and the corresponding rotation axis vector V ⊥ , and then through θ and V ⊥ Calculate T ⊥ correspondingly, calculate P || The rotation matrix T of the YOZ plane relative to the standard orbit point cloud model || , we can first calculate P || Angle relative to the YOZ plane of the standard track point cloud model and the corresponding rotation axis vector V || , and then through and V || Calculate T || Ways to use T ⊥ and T || After rotating the 3DPC, P ⊥ and P || Align with the XOZ plane and YOZ plane of the standard track point cloud model respectively, so that the local coordinate system of the 3DPC can be aligned with the coordinate system of the standard track point cloud model; the preset threshold value needs to be determined according to the placement of the point cloud in the standard track point cloud model relative to the origin of the coordinate system, and it is sufficient to judge whether the 3DPC is placed in the opposite direction along the Z axis according to the threshold value; specifically, in this embodiment, if the X value of the center of mass of the 3DPC is less than 0, the 3DPC is rotated 180° around the Y axis to avoid the extension direction of the 3DPC being inconsistent with the extension direction of the Z axis; if the X value of the center of mass is greater than or equal to the preset threshold value, the processed three-dimensional point cloud 3DPCⅠ is directly obtained.

[0078] In step S3, due to the physical structure of track 1, the point with the maximum Y value in 3DPCⅠ will only appear in the track 1 part, and will not appear in the non-track 1 part, such as the ground or the plywood. Therefore, the maximum Y value of 3DPCⅠ is determined by the geometric parameters of track 1. When the corresponding threshold is set according to the geometric parameters of the highest point of track 1, the Y value of all points in 3DPCⅠ that meet the posture correction requirements should be less than the threshold: Therefore, the specific posture correction judgment basis adopted in this embodiment is: when the maximum Y value of 3DPCⅠ is less than or equal to the preset threshold, 3DPCⅠ meets the posture correction requirements, and 3DPCⅠ can be used as the three-dimensional point cloud 3DPCⅡ that meets the posture correction requirements.

[0079] Example 2

[0080] On the basis of Example 1, in the step of fitting 3DPC to a plane point cloud PPC in step S1, the step includes dividing 3DPC into t pieces of point clouds 3DPC[i], 0≤i<t, and fitting 3DPC[i] to t pieces of plane point clouds PPC[i]; correspondingly, in step S2, the 3DPC is subjected to posture transformation based on PPC[i], and a processed three-dimensional point cloud 3DPCⅠ[i] is obtained; in step S3, it is determined whether 3DPCⅠ[i] meets the posture correction requirements. If so, a three-dimensional point cloud 3DPCⅡ after posture correction is obtained.

[0081] Specifically, this embodiment uses region growing segmentation to segment 3DPC into t point clouds 3DPC[i], so as to distinguish point clouds of non-same smooth surfaces in 3DPC. Region growing segmentation can make all points on each element in 3DPC[i] located on the same smooth surface, making the point cloud data more holistic and continuous, making it easier to extract local features in the point cloud, and facilitating the recognition and classification of the point cloud in subsequent steps.

[0082] 3DPC[i] is fitted to PPC[i] and plane model segmentation is adopted. Specifically, a plane is fitted to 3DPC using random sampling consistency. Points whose distances to the plane are less than a distance threshold are regarded as points within the plane, and points whose distances are greater than the distance threshold are regarded as points outside the plane. The points within the plane are taken as the segmentation results. Taking the detection of turnouts in this embodiment as an example, plane model segmentation is performed on 3DPC, and the derailment head tread plane can be effectively extracted from 3DPC. The specific method of plane model segmentation can refer to the prior art. In this embodiment, the RANSAC algorithm (Random Sampling Consensus) is preferred.

[0083] Using region growing segmentation and plane model segmentation to process 3DPC can further filter out the required point cloud data from 3DPC, remove unnecessary point cloud data in 3DPC, such as point cloud information of other structures such as the ground, plywood, bolts, etc., and improve the robustness of this embodiment.

[0084] In step S3, the posture correction requirement of 3DPCⅠ[i] is determined as follows:

[0085] S31, comparing the similarity of each 3DPCⅠ[i] with the corresponding part of the standard track point cloud model;

[0086] S32. According to the similarity between each 3DPCⅠ[i] and the corresponding part of the standard track point cloud model, the pose correction requirement is judged in descending order; if a 3DPCⅠ[i] that meets the pose correction requirement appears, then 3DPCⅠ meets the pose correction requirement and 3DPCⅡ is obtained.

[0087] In step S31, the similarity comparison refers to the prior art and can be evaluated by indicators such as mean square error, maximum common point set, land movement distance, chamfer distance, etc. In this embodiment, the Hausdorf distance Hausdorf[i] between 3DPCⅠ[i] and the cross section of the standard straight rail point cloud model is calculated for similarity comparison; correspondingly, in step S32, 3DPCⅠ[i] is reordered from small to large according to the size of Hausdorf[i] of 3DPCⅠ[i] to obtain the reordered 3DPCⅠ[i], 3DPCⅠ[0] is the point cloud with the highest similarity to the standard point cloud model, and 3DPCⅠ[t-1] is the point cloud with the lowest similarity to the standard point cloud model; then, starting from 3DPCⅠ[0] to 3DPCⅠ[t-1], each 3DPCⅠ[i] is judged in turn. Whether DPCⅠ[i] meets the pose correction requirements, that is, whether the maximum Y value is less than or equal to the preset threshold; if there is a 3DPCⅠ[i] that meets the pose correction requirements, the subsequent 3DPCⅠ[i] has a lower similarity with the standard point cloud model and is less likely to meet the pose correction requirements, so there is no need to judge the subsequent 3DPCⅠ[i] one by one; that is, as long as there is one 3DPCⅠ[i] that meets the pose correction requirements, the judgment can be ended in advance and the entire 3DPCⅠ can be used as 3DPCⅡ, thereby ending the judgment in advance to reduce the amount of calculation and improve the calculation speed; if there is still no 3DPCⅠ[i] that meets the pose correction requirements until 3DPCⅠ[t-1], it means that 3DPCⅠ does not meet the pose correction requirements or there is a problem with the initially selected 3DPC data, and 3DPC needs to be reselected.

[0088] Example 3

[0089] Based on Example 1 or 2, step S2 specifically comprises the following steps:

[0090] Calculate the normal vector N[i] of each point P[i] in PPC, 0≤i<n, n is the number of points in PPC; calculate the RN vector that measures the average statistical value of N[i]; calculate the positional relationship d[i] of P[i] to the RN vector respectively, let the maximum value in d[i] be d[s], and the point corresponding to d[s] be P[s]; use RN vector and P[s] to establish plane P ⊥ and plane P || , where P ⊥ Perpendicular to RN vector, P || Parallel to the RN vector;

[0091] Calculate P ⊥ The angle θ between the standard orbit point cloud model coordinate system plane XOZ and the corresponding rotation axis vector V ⊥ , calculate around V ⊥ The rotation matrix T of the rotation θ ⊥ ; Calculate P || The angle between the YOZ plane and the standard track point cloud model coordinate system and the corresponding rotation axis vector V || , calculate around V || Rotation The rotation matrix T || ; Use T ⊥ and T || Perform rotation transformation on 3DPC;

[0092] Calculate the X value of the centroid of the 3DPC. If the X value of the centroid is less than the preset threshold, rotate the 3DPC 180° around the Y axis to obtain the processed three-dimensional point cloud 3DPCⅠ.

[0093] In step S2, the RN vector that measures the average statistical value of N[i] is calculated using formula (1):

[0094]

[0095] In the formula is the RN vector, are the components of the RN vector in the X, Y, and Z directions, respectively, N[i]x, N[i] y 、N[i] z are the components of the N[i] vector in the X, Y, and Z directions, respectively, and n is the number of points in the point cloud PPC;

[0096] In this step, the RN vector is used to take all the points in the PPC into account, which can avoid the influence of a single bad point on the normal calculation; it can also avoid the situation where some points cannot be taken into account due to the direct calculation of the normal of the fitting plane.

[0097] In step S2, the positional relationship d[i] between each point P[i] in PPC and the RN vector is calculated, and when the maximum value in d[i] is set to d[s], formula (2) and formula (3) are used respectively:

[0098]

[0099] d[s]=max(d[i]) (0≤i <n) (3)

[0100] Where d[i] is the positional relationship between point P[i] (0≤i<n) and RN vector, x[i], y[i], z[i] are the coordinates of point P[i] in the X, Y, and Z directions respectively. are the components of the RN vector in the X, Y, and Z directions respectively;

[0101] This step searches for the point in PPC that has the largest positional relationship with the RN vector, which is used as the basis for the subsequent establishment of P ⊥ and plane P || It is more accurate to use the established plane to measure the feature orientation of the point cloud itself.

[0102] Use RN vectors and P[s] to create plane P ⊥ and plane P || When using formula (4) and formula (5) respectively:

[0103]

[0104]

[0105] In the formula are the components of the RN vector in the X, Y, and Z directions respectively, x[s], y[s], and z[s] are the coordinate values ​​of the point P[s] in the X, Y, and Z directions respectively; X, Y, and Z are the three axes in the three-dimensional space.

[0106] This step uses vector RN and point P[s] to create a plane P perpendicular to vector RN. ⊥ and the plane P parallel to the vector RN || , used to measure the characteristic orientation of point clouds, and the orientation description is more accurate.

[0107] In step S2, θ and When using formula (6) and formula (7):

[0108]

[0109]

[0110] Where θ is the plane P ⊥ The angle between the XOZ plane and are the components of the RN vector in the X, Y, and Z directions respectively; x[s], y[s], and z[s] are the coordinate values ​​of the point P[s] in the X, Y, and Z directions respectively.

[0111] In step S2, the rotation axis vector V used to calculate the rotation matrix is ​​calculated. ⊥ and V || When using formula (8) and formula (9) respectively:

[0112]

[0113]

[0114] Where V ⊥x 、V ⊥y 、V ⊥z V ⊥ Components in the X, Y, and Z directions; V ||x 、V ||y 、V ||z V || Components in the X, Y, and Z directions; are the components of the RN vector in the X, Y, and Z directions respectively; x[s], y[s], and z[s] are the coordinate values ​​of the point P[s] in the X, Y, and Z directions respectively.

[0115] In step S2, the rotation matrix T is calculated ⊥ and T || When using formula (10) and formula (11) respectively:

[0116]

[0117]

[0118] Where θ is the plane P ⊥ The angle between the XOZ plane and For plane P || The angle between the plane and the YOZ plane, V ⊥x 、V ⊥y 、V ⊥z V ⊥ Components in the X, Y, and Z directions; V ||x 、V ||y 、V ||z V || Components in the X, Y, and Z directions;

[0119] Example 4

[0120] like Figure 2 As shown, the application of a trajectory point cloud posture correction method adopted in this embodiment includes the following steps:

[0121] A. Correcting the posture of 3DPC by any of the posture correction methods of the track point cloud in Examples 1 to 3 to obtain a corrected three-dimensional point cloud 3DPCⅡ;

[0122] B. According to the type of track 1 identified by 3DPCⅡ, the corresponding standard track point cloud model is selected to perform deviation analysis on 3DPCⅡ to obtain the deviation analysis result of 3DPCⅡ.

[0123] Before step A, a track inspection plan can be formulated, including the formulation of inspection locations and inspection parameters; for example, for the No. 18 point rail with a speed of 350 kilometers per hour, the reduction value, vertical wear, and side wear of the straight point rail and the curved point rail can be inspected; for the No. 18 switch with a speed of 350 kilometers per hour, the throat width and stroke of the wing rail can be inspected, the reduction value and wear of the long center rail can be inspected, and the reduction value and wear of the short center rail can be inspected; for the rail, its inspection parameters include rail head wear value, rail head tread wear depth, rail head height, rail head width, rail height, rail waist thickness, rail bottom width, rail head vertical straightness and horizontal straightness, wheel-rail contact radius and other geometric dimension parameters; for the joint rail, its inspection parameters include rail joint width, left and right tooth misalignment, upper and lower tooth misalignment and straightness; the preparation of the inspection data file matches the setting of the inspection location and inspection parameters.

[0124] In step A, the specific acquisition method of 3DPC can refer to the following steps:

[0125] An image of track 1 with grating stripes is acquired by projecting surface structured light onto the surface of track 1; and the image of track 1 is resolved into 3DPC.

[0126] Specifically, the surface structured light includes two groups of grating stripes, and the two groups of grating stripes are arranged perpendicular to each other; when projected onto the surface of track 1, one group of grating stripes is parallel to the longitudinal direction of the track, and one group of grating stripes is parallel to the height direction of the track; the color of the grating stripes is blue, so that the shorter wavelength of blue light can be used to reduce the bending and deviation of light on the surface of track 1, thereby improving the measurement accuracy; the visual sensor uses an industrial camera that includes at least two cameras arranged at intervals, so that the distance between each pixel point and the visual sensor can be calculated through binocular or multi-eye vision, and then the three-dimensional coordinates of each point on the image are solved to obtain 3DPC.

[0127] When selecting the standard track point cloud model corresponding to 3DPCⅡ in step B, the similarity between 3DPCⅡ and different standard track point cloud models can be used as a basis; the similarity comparison between 3DPCⅡ and different standard track point cloud models can refer to the existing technology and can be evaluated by indicators such as mean square error, maximum common point set, land movement distance, chamfer distance, etc.; after obtaining the deviation analysis result in step B, the track geometric parameter state conclusion can be obtained in combination with the existing geometric parameter state judgment basis, such as the wear data of the track including vertical wear, side wear and total wear, and whether the geometric dimension parameters including rail bottom width, rail head vertical straightness and horizontal straightness are qualified.

[0128] Specifically, step B comprises the following steps:

[0129] B1. Roughly align 3DPCⅡ with the corresponding standard track point cloud model;

[0130] B2. Accurately align 3DPCⅡ with the corresponding standard track point cloud model;

[0131] B3. Calculate the deviation between 3DPCⅡ and the corresponding standard track point cloud model.

[0132] Coarse registration is used to quickly align 3DPCⅡ and the standard track point cloud model, which can be achieved in a variety of ways, such as obtaining the centroid of the specified structure on 3DPCⅡ and aligning it with the centroid of the structure corresponding to the specified structure on the standard track point cloud model; or using the SAC-IA algorithm (Sample Consensus Initial Alignment) for coarse registration; fine registration is used to further optimize the alignment effect of 3DPCⅡ and the standard track point cloud model, which can be achieved in a variety of ways, such as the ICP algorithm (Iterative Closest Point) or the NDT algorithm (Normal Distributions Transform);

[0133] Adding step B1 before step B2 to perform hierarchical registration of 3DPCⅡ and the corresponding standard track point cloud model can reduce the amount of calculation for fine registration in step B2, thereby speeding up the calculation speed of fine registration and improving the efficiency of the registration process; at the same time, coarse registration can also eliminate most of the errors of the initial position of 3DPCⅡ relative to the standard track point cloud model, making it easier for the point cloud to converge to the optimal solution in the fine registration stage, thereby increasing the stability of the registration;

[0134] Furthermore, before step B1, a noise removal step is also included to further eliminate interference information, such as point cloud information of other structures such as the ground, plywood, bolts, etc., and extract the required track point cloud information; noise removal can be carried out in various ways, such as using a Euclidean distance clustering algorithm to remove noise points with a small number of points; using a spherical radius filtering algorithm or a statistical filtering algorithm to remove outliers; by obtaining the three-dimensional coordinates of the 3DPCⅡ centroid, setting the direction and threshold, and using a straight-through filtering algorithm, remove all noise points away from track 1 in a certain direction.

[0135] The deviation between 3DPCⅡ and the corresponding standard track point cloud model in step B3, that is, the distance between each point on 3DPCⅡ and the nearest neighbor point in the standard point cloud model, can be calculated by various methods, such as using an octree nearest neighbor search algorithm or a kdtree nearest neighbor search algorithm.

[0136] Furthermore, when performing deviation analysis on 3DPCⅡ in step B, 3DPCⅡ is divided into m sections 3DPCS[i] along the Z axis, 0≤i<m; the corresponding section on the standard track point cloud model is selected for each 3DPCS[i] to perform similarity calculation, and the control section is obtained, and the corresponding standard track point cloud model is selected according to the control section. Deviation analysis is performed on each section 3DPCS[i] of 3DPCⅡ along the Z axis, which is suitable for detecting variable-section track 1, and it is necessary to select the corresponding section on the standard track point cloud model for each section of track 1 for deviation analysis.

[0137] This embodiment further segments the rail head and rail waist in 3DPCⅡ before step B1; correspondingly, in step B1, the centroid of the rail head of 3DPCⅡ is aligned with the centroid of the rail head of the standard track point cloud model, or the centroid of the rail waist of 3DPCⅡ is aligned with the centroid of the rail waist of the standard track point cloud model; various algorithms can be used to segment the rail head and rail waist, such as the Euclidean clustering segmentation algorithm or the region growing segmentation algorithm.

[0138] Example 5

[0139] like Figure 3 As shown, on the basis of Example 4, it further includes establishing a track database; the track database stores at least the deviation analysis results of 3DPCⅡ; specifically, the track database uses a relational database such as SQLite or ORACLE to store track data in the form of rows, columns, and a table composed of rows and columns, which is convenient for users to understand and query;

[0140] Specifically, through the deviation analysis results of 3DPCⅡ, we can obtain the track geometric dimension parameters including wheel flange groove width, throat width, rail height, rail head width, rail bottom width, rail waist thickness, and reduction value, as well as wear data including vertical wear, side wear, and total wear; further, we can traverse the control section along the longitudinal direction of the track to obtain wear data curves, such as curves of vertical wear, side wear, and total wear along the longitudinal direction of the track, so as to more intuitively display the wear of the track along the longitudinal direction of the track.

[0141] In addition to the deviation analysis results of 3DPCⅡ, the track database of this embodiment also stores the 3DPC obtained in step A as original data, and the intermediate data 3DPCⅠ, which can facilitate historical data backtracking and verification.

[0142] Correspondingly, the track database also needs to have corresponding track detection data management means and track geometric state parameter management means; specifically, these two management means include the management of detection parameters and parameter curves obtained after data processing, including the addition, deletion, modification and query of geometric parameter state judgment basis, and the addition, deletion, modification and query of detection parameters and parameter curves. Addition, deletion, modification and query means adding new judgment basis, deleting old judgment basis, modifying judgment basis that needs to be updated, and querying whether the judgment basis is correct.

[0143] This embodiment also formulates a maintenance plan based on the information in the track database; the maintenance plan includes a routine maintenance plan, a major and medium maintenance plan, and a re-inspection plan;

[0144] The routine maintenance and repair plan is mainly preventive and repair work. For example, when the deviation analysis result of the point rail finds that the point rail reduction value does not meet the requirements, it is necessary to insert a special iron gasket of a certain thickness between the bottom of the point rail and the slide bed according to the point rail reduction value data, so that the point rail reduction value meets the design standard; or according to the point rail reduction value error, adjust the thickness of the rubber pad under the base rail, so that the relative height of the base rail and the point rail, that is, the point rail reduction value, meets the design allowable error range;

[0145] The major and medium repair maintenance plan mainly involves replacement operations. For example, for rails, when the head wear of the rail reaches the light damage standard, that is, the vertical wear reaches 8mm, the side wear reaches 10mm, and the total wear reaches 9mm, the damaged part is marked according to the degree of damage, and the wear development of the damaged part is monitored; when the head wear of the rail reaches the severe damage standard, that is, the vertical wear reaches 10mm and the side wear reaches 12mm, the severely damaged part is promptly removed, welded, or even replaced or re-laid at the corresponding part of the rail; for turnouts, when the following situations occur in the turnout, it is necessary to organize repairs or replacements in a timely manner: the deviation of the point rail relative to the basic rail reduction value and the heart rail relative to the wing rail reduction value exceeds 1mm, and it affects the driving stability; the point rail and the heart rail cause abnormal light bands due to torsion or wear, and it affects the driving stability; other damages reach the rail light damage standard;

[0146] The re-inspection plan is mainly to evaluate the status of the maintenance and repair work after it is completed, so as to conduct acceptance and check whether the maintenance and repair work has achieved its goals.

[0147] Furthermore, a track geometry parameter trend graph can be established based on the deviation analysis results of 3DPCⅡ in the track database, and the geometry parameters of track 1 can be predicted based on the track geometry parameter trend graph; specifically, a track geometry parameter line graph can be established based on the historical geometry parameter data in the track database, and the geometric parameters can be predicted using a grey prediction model or a support vector machine prediction model to establish a track geometry parameter trend graph, thereby assisting in formulating a maintenance plan based on the track geometry parameter trend graph.

[0148] The specific formulation of the maintenance and repair plan can refer to the "High-speed Railway Ballasted Track Line Maintenance Rules" (Tieyun (2013) No. 29), "High-speed Railway Ballastless Track Line Maintenance Rules (Trial)" (Tieyun (2012) No. 83), "Conventional Railway Line Repair Rules", "Conventional Railway Line Repair Rules Supplementary Provisions" (Tiegongdian

[2022] No. 78), and "High-speed Railway Turnout and Rail Expansion Adjuster Maintenance Implementation Rules".

[0149] Example 6

[0150] A device for measuring the three-dimensional geometric state of a track comprises a processor and a memory; the memory and the processor are communicatively connected; the memory stores instructions that can be executed by the processor, and the processor can execute any one of the track point cloud posture correction methods in Examples 1 to 3 through the instructions.

[0151] The memory may adopt existing technologies, such as a solid-state drive or a mechanical hard drive; the processor may also adopt existing technologies, such as a central processing unit and a graphics processing unit; the communication connection between the memory and the processor may be wired communication or wireless communication.

[0152] Example 7

[0153] A three-dimensional track geometric state measurement system comprises a surface structured light emitter, a visual sensor and a three-dimensional track geometric state measurement device in Example 6; the surface structured light emitter is used to project surface structured light onto track 1; the visual sensor is used to obtain an image of track 1 with grating stripes; the visual sensor is communicatively connected to a processor.

[0154] The surface structured light emitter can be in various forms, such as an emitter that can emit two groups of orthogonally distributed grating stripes, or a combination of multiple unidirectional grating stripe emitters; the surface structured light emitter can emit various lights, such as yellow light and green light. This embodiment uses an emitter that can emit blue unidirectional grating stripes. When in use, the first group of grating stripes is first projected toward track 1 along the longitudinal direction of the track or the height direction of the track and photographed, and then the second group of grating stripes is projected toward track 1 along the height direction of the track or the longitudinal direction of the track and photographed.

[0155] The visual sensor can be a visual sensor capable of measuring distance, such as a binocular vision camera, a multi-vision camera or a structured light camera.

[0156] Further, such as Figures 4 to 7 As shown, it also includes a crossbeam 2 and a wheel assembly 3; the wheel assembly 3 is arranged on both sides of the crossbeam 2 along the horizontal direction of the track; at least one of the wheel assemblies 3 is connected to the crossbeam 2 through an adaptive device 5 so that it can swing around the vertical axis relative to the crossbeam 2; the spacing of the wheel assemblies 3 corresponds to the gauge of the track 1; the wheel assembly 3 can move along the track 1; the surface structured light emitter and the visual sensor are connected to the crossbeam 2.

[0157] The wheel assembly 3, the crossbeam 2 and the adaptive device 5 are combined into a track inspection trolley. When the staff takes pictures of the track 1 along the longitudinal direction of the track, they can first move the track inspection trolley equipped with a surface structured light emitter and a visual sensor to the position where the picture needs to be taken, thereby eliminating the need to manually carry the surface structured light emitter and the visual sensor, which can save physical effort.

[0158] Furthermore, the spacing between the wheel assemblies 3 on both sides is adjustable, and the change in the track gauge of the track 1 can drive the spacing between the wheel assemblies 3 on both sides to change accordingly, so that the wheel assemblies 3 on both sides maintain contact with the track 1 on the corresponding side;

[0159] The spacing between the wheel assemblies 3 on both sides is adjustable, and can be various structures, such as at least one wheel assembly 3 is movably connected to the crossbeam 2, or the two wheel assemblies 3 are connected to each other by an adjustable mechanism such as a slide rail slider mechanism or a cylinder; the spacing between the wheel assemblies 3 on both sides can change with the gauge of the track 1, and a passive structure can be adopted, such as limiting the freedom of movement of the wheel assembly 3 relative to the track 1 along the lateral direction of the track through a limit baffle or a limit pin, or an active structure can be adopted, such as driving the spacing between the wheel assemblies 3 on both sides through a servo motor or a cylinder to change with the gauge of the track 1.

[0160] At least two traveling wheels 31 are arranged on each side of the wheel assembly 3, and the traveling wheels on the wheel assembly 3 on one side are spaced apart longitudinally along the track; specifically, Fig.10 As shown, two traveling wheels 31 are provided on the single-side wheel assembly 3 of the present embodiment, so that the track three-dimensional geometric state measurement system of the present embodiment forms a four-wheel structure, which can run along the track 1 more smoothly.

[0161] The axle of at least one traveling wheel 31 is a stepped shaft 32, and is provided with a position sensor for sensing the number of rotations of the stepped shaft 32, so that the mileage of the traveling wheel 31 can be calculated based on the number of rotations of the stepped shaft 32; when using the visual sensor, the operator can take an image of the track 1 at a certain distance according to the mileage of the traveling wheel 31, so as to avoid missing part of the track 1 when taking the image of the track 1, resulting in incomplete 3DPC data; the axles of the remaining traveling wheels 31 can use optical axes to reduce manufacturing costs.

[0162] At least one guide wheel 34 is also provided on each wheel assembly 3; the axis of the guide wheel 34 is arranged along the track height direction, and the guide wheel 34 is used to abut against the side wall of the track 1, thereby providing a guiding function for the wheel assembly 3 to prevent the wheel assembly 3 from deviating from the track 1 when moving longitudinally along the track; the radius of each horizontal section of the guide wheel 34 along the track height direction is different, that is, the two side surfaces of the guide wheel 34 along its axis are inclined surfaces, which can ensure good contact with the fat edge of the track 1 and improve the restraint performance of the wheel assembly 3; specifically, as Figure 6 , Figure 7 and Fig.10 As shown, the guide wheel 34 of this embodiment is arranged on the side of the wheel assembly 3 facing the crossbeam 2, so as to reduce the horizontal dimension of the track three-dimensional geometric state measurement system along the track, and the cross-section of the guide wheel 34 along the vertical plane is V-shaped.

[0163] The surface structured light emitter and the visual sensor are connected to the crossbeam 2 via a cantilever connecting rod 6; the cantilever connecting rod 6 is used to adjust the position and angle of the surface structured light emitter and the visual sensor; the present embodiment adopts a cantilever connecting rod 6 including at least three degrees of freedom of movement and two degrees of freedom of rotation, specifically three degrees of freedom of movement along the horizontal direction of the track, the longitudinal direction of the track and the height direction of the track, and the axis along the height direction of the track, and the degree of freedom of rotation parallel to the horizontal direction, so that the operator can adjust the distance and angle between the surface structured light emitter and the visual sensor and the track 1 at will when operating the surface structured light emitter and the visual sensor, and it is easier to obtain high-quality images;

[0164] Specifically, Figure 4 and Fig.11 As shown, the cantilever connecting rod 6 includes at least two connecting rods 61; each connecting rod 61 is connected end to end, and the connection between two adjacent connecting rods 61 includes at least two rotating pairs connected in series, wherein the axis of at least one rotating pair is arranged along the track height direction, and the axis of at least one rotating pair is parallel to the horizontal direction, so that the two adjacent connecting rods 61 have rotational freedom with the axis along the track height direction and parallel to the horizontal direction respectively; one end of the whole formed by the combination of each connecting rod 61 is fixed relative to the crossbeam 2, and the other end is connected to the equipment platform 62 through at least two rotating pairs connected in series, and the surface structure light emitter and the visual sensor are placed on the equipment platform 62.

[0165] At least one wheel assembly 3 is also connected to a track gauge adjustment mechanism 4; the track gauge adjustment mechanism 4 comprises an elastic element, one end of the elastic element is connected to the wheel assembly 3 on one side, and the other end of the elastic element is connected to the crossbeam 2 or the wheel assembly 3 on the other side; the elastic element is used to drive the wheel assemblies 3 on both sides to respectively press against the rail 1 on the corresponding side; the elastic element can be in various forms, such as a rubber block, a disc spring; the track gauge adjustment mechanism 4 is used to drive the spacing between the wheel assemblies 3 on both sides to change accordingly when the gauge of the rail 1 changes, such as when the gauge increases, so that the wheel assemblies 3 on both sides can always keep in contact with the rail 1 on the corresponding side, prevent the wheel assemblies 3 from falling off the rail 1 due to the change in the gauge of the rail 1, and ensure the stability of the wheel assemblies 3 traveling on the rail 1;

[0166] Specifically, the track gauge adjustment mechanism 4 of this embodiment includes a plurality of coil springs, one end of the coil spring abuts against the side of one of the wheel assemblies 3 facing the crossbeam 2, and the other end of the coil spring is fixed to the crossbeam 2, thereby being able to drive the two wheel assemblies 3 away from each other.

[0167] Furthermore, if Figure 7 , Figure 8 , Fig. 9 and Fig.10As shown, the adaptive device 5 includes a first mounting seat 21 and a second mounting seat 51, the first mounting seat 21 can rotate relative to the second mounting seat 51 along the vertical axis, one of the first mounting seat 21 and the second mounting seat 51 is connected to the crossbeam 2, and the other of the first mounting seat 21 and the second mounting seat 51 is connected to the wheel assembly 3; for example, the first mounting seat 21 is connected to the crossbeam 2, and the second mounting seat 51 is connected to the wheel assembly 3, or the second mounting seat 51 is connected to the crossbeam 2, and the first mounting seat 21 is connected to the wheel assembly 3.

[0168] The relative rotation of the first mounting seat 21 and the second mounting seat 51 can be achieved in various ways. For example, one of the first mounting seat 21 and the second mounting seat 51 is provided with a recessed portion 211, and the other of the first mounting seat 21 and the second mounting seat 51 is at least partially rotatably disposed in the recessed portion 211, so that; for example, the second mounting seat 51 is provided with a recessed portion 211, and at least a portion of the first mounting seat 21 is rotatably disposed in the recessed portion 211, or the first mounting seat 21 is provided with a recessed portion 211, and at least a portion of the second mounting seat 51 is rotatably disposed in the recessed portion 211.

[0169] like Figure 8 As shown, the adaptive device 5 further includes a connecting member 53 for connecting the first mounting seat 21 and the second mounting seat 51, at least a portion of the connecting member 53 is located in the recessed portion 211 to connect the second mounting seat 51 and the first mounting seat 21, and to rotate with the other of the second mounting seat 51 and the first mounting seat 21. For example, the connecting member 53 is a quick release locker, and the connecting member 53 includes a first portion that cooperates with the second mounting seat 51 and a second portion that cooperates with the first mounting seat 21, so as to ensure that the second mounting seat 51 and the first mounting seat 21 are connected as one, and the second mounting seat 51 and the first mounting seat 21 can rotate relative to each other.

[0170] Further, in the case where one of the first mounting seat 21 and the second mounting seat 51 is provided with a recessed portion 211, a raised portion 511 may be provided on the other of the first mounting seat 21 and the second mounting seat 51, for example, the first mounting seat 21 is provided with a recessed portion 211, and the second mounting seat 51 is provided with a raised portion 511, or the first mounting seat 21 is provided with a raised portion 511, and the second mounting seat 51 is provided with a recessed portion 211. The circumferential surface of the raised portion 511 and the circumferential surface of the recessed portion 211 are both provided with arc surfaces that match each other, so that when the second mounting seat 51 and the first mounting seat 21 rotate relative to each other, the arc surfaces of the two can produce sliding contact and guide the rotation, thereby ensuring the smoothness and stability of the rotation and avoiding abnormalities such as jamming during rotation; the connecting member 53 is sequentially penetrated with the recessed portion 211 and the raised portion 511, thereby connecting the first mounting seat 21 and the second mounting seat 51.

[0171] Optionally, the periphery of the recessed portion 211 is provided with a first curved surface and a second curved surface opposite to each other, and an opening communicating with the inner cavity of the recessed portion 211, the opening being located between the first curved surface and the second curved surface, and two openings can be provided, and the two openings are provided opposite to each other; the curved surface provided on the raised portion 511 cooperates with the first curved surface and the second curved surface. In this way, the effective rotation between the second mounting seat 51 and the first mounting seat 21 can be ensured, while the contact surface between the recessed portion 211 and the raised portion 511 can be reduced, thereby reducing the friction resistance and ensuring smooth rotation.

[0172] Optionally, for the specific structure of the protrusion 511, the periphery of the protrusion 511 can be a continuous and complete annular surface. Figure 8 As shown, the two ends of the raised portion 511 are provided with a third curved surface and a fourth curved surface opposite to each other, and a side plane respectively connecting the third curved surface and the fourth curved surface, the third curved surface is slidably matched with the first curved surface, the fourth curved surface is slidably matched with the second curved surface, the side plane corresponds to the opening, and when the first mounting seat 21 and the second mounting seat 51 rotate relative to each other, the side plane of the raised portion 511 can extend or retract into the opening of the recessed portion 211. In this way, the first mounting seat 21 and the second mounting seat 51 can have a high degree of structural matching, which is conducive to the miniaturization of the adaptive device 5 and improves compactness.

[0173] Optionally, the adaptive device 5 also includes a slide rail 33, which is arranged laterally along the track, and one of the first mounting seat 21 and the second mounting seat 51 is slidably connected to the beam 2 or the wheel assembly 3 through the slide rail 33, thereby realizing the position adjustment of the wheel assembly 3 relative to the beam 2 along the lateral direction of the track; for example, when the second mounting seat 51 is connected to the beam 2 and the first mounting seat 21 is connected to the wheel assembly 3, the second mounting seat 51 can be slidably connected to the beam 2 through the slide rail 33, and the first mounting seat 21 can be slidably connected to the wheel assembly 3 through the slide rail 33.

[0174] Furthermore, a dustproof groove for accommodating the slide rail 33 is opened on one side of the second mounting seat 51 and the other of the first mounting seat 21 toward the slide rail 33, and the slider 52 described later can be set in the dustproof groove, and the slide rail 33 can be at least partially accommodated in the dustproof groove. In this way, the dustproof groove can improve the concealment of the slide rail 33 and the slider 52, so as to improve the appearance performance of the track inspection trolley while providing a shield protection for the slide rail 33 and the slider 52, so as to prevent dust, rainwater and other foreign objects from falling directly on the slide rail 33 to cause wear or corrosion to the slide rail 33, thereby extending the service life of the slide rail 33.

[0175] Furthermore, the width of the dustproof groove is greater than the width of the slide rail 33 to prevent the second mounting seat 51 from interfering with the slide rail 33 when the second mounting seat 51 moves relative to the first mounting seat 21, thereby ensuring the smoothness of the relative movement between the second mounting seat 51 and the first mounting seat 21.

[0176] Furthermore, if Fig. 9 As shown, the adaptive device 5 also includes a slider 52, which is fixed to the other one of the second mounting seat 51 and the first mounting seat 21, and the slider 52 is slidably connected to the slide rail 33, thereby forming a sliding connection between the second mounting seat 51 and the other one of the first mounting seat 21 and the slider 52; when the slider 52 or the slide rail 33 is worn, it is only necessary to replace the slider 52 and the slide rail 33 without replacing the first mounting seat 21 or the second mounting seat 51 as a whole, so that this embodiment has better maintainability.

[0177] Furthermore, the adaptive device 5 also includes a limit block 35 for limiting the position of the slider 52. The limit block 35 is arranged at at least one of the two ends of the slide rail 33 along the transverse direction of the track to limit the travel of the slider 52 in the corresponding direction along the transverse direction of the track, thereby limiting the sliding travel of the second mounting seat 51 and the other of the first mounting seat 21 relative to the slide rail 33, thereby limiting the excessive movement of the wheel assembly 3 relative to the beam 2.

[0178] Specifically in this embodiment, the first mounting seat 21 of this embodiment is arranged at the bottom of the end of the crossbeam 2 along the transverse direction of the track, and a recessed portion 211 is arranged in the first mounting seat 21, and the second mounting seat 51 is arranged at the top of the wheel assembly 3, and a protruding portion 511 is arranged in the second mounting seat 51, and the recessed portion 211 and the protruding portion 511 cooperate with each other, so that the wheel assembly 3 can rotate around the track height direction relative to the crossbeam 2; the slide rail 33 is fixed to the top of the wheel assembly 3, and a slider 52 is provided at the corresponding position of the second mounting seat 51, and the slider 52 and the slide rail 33 are slidably connected; a dustproof groove is also arranged on a side of the second mounting seat 51 facing the slide rail 33, and the slide rail 33 and the slider 52 are both located in the dustproof groove;

[0179] In this embodiment, the adaptive device 5 provides the wheel assembly 3 with a degree of freedom of movement relative to the cross beam 2 and a degree of freedom of rotation relative to the cross beam 2, wherein the degree of freedom of movement is along the horizontal direction of the track, and the axis of the degree of freedom of rotation is along the height direction of the track. When the wheel assembly 3 travels on the track 1 with an uneven surface or a curved deformation, the wheel assembly 3 can move and rotate accordingly according to the contour change of the track 1, so as to always keep the relative position and angle of the wheel assembly 3 and the track 1 unchanged, and then always keep the running wheel and the track 1 in contact. On the one hand, it can eliminate the interference of the relative position and angle change of the wheel assembly 3 and the track 1 on the measurement operation or measurement result, so as to obtain more accurate On the other hand, it can also avoid the situation that the wheel assembly 3 is separated from the track 1 or stuck due to the excessive distance between the wheel assembly 3 and the track 1 or the excessive deflection angle relative to the track 1, so that the present embodiment has higher passability; and the structure of the adaptive device 5 of the present embodiment can realize the swing of the wheel assembly 3 relative to the crossbeam 2 around the track height direction, and can realize the adjustable distance between the wheel assemblies 3 on both sides, and has a compact structure, which can reduce the volume of the track 1 three-dimensional geometric state measurement system of the present embodiment.

[0180] like Figure 4 As shown, a second cantilever connecting rod is also connected to the top surface of the crossbeam 2; the second cantilever connecting rod includes a second connecting rod and a second equipment platform; the second equipment platform can be used to place equipment or tools required for measurement, such as a computer or a notebook; the connection between the second connecting rod and the crossbeam 2 includes at least two rotating pairs, wherein the axis of at least one rotating pair is arranged along the track height direction, and the axis of at least one rotating pair is parallel to the horizontal direction, so that the second connecting rod has a rotational freedom relative to the crossbeam 2, with the axis being respectively along the track height direction and parallel to the horizontal direction; when the second cantilever connecting rod is not needed to place the equipment or tools required for measurement, the second connecting rod can be rotated until the axis is parallel to the crossbeam 2, thereby reducing the space occupied by the second connecting rod; when the second cantilever connecting rod is needed to place the equipment or tools required for measurement, the second connecting rod can be rotated to stand upright relative to the horizontal plane, thereby facilitating the operator to access the second equipment platform;

[0181] The second equipment platform and the second connecting rod may also include a rotating pair; this embodiment includes a rotating pair with an axis parallel to the horizontal direction between the second equipment platform and the second connecting rod, so that the second equipment platform has a rotational freedom along the horizontal direction relative to the crossbeam 2, and when the second equipment platform needs to be stored, the second equipment platform can be rotated to be parallel to the horizontal plane, thereby reducing the space occupied by the second equipment platform.

[0182] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for correcting the position and posture of a track point cloud, characterized in that: The following steps are involved: S1, obtain the three-dimensional point cloud 3DPC of track (1); fit the 3DPC into a plane point cloud PPC; S2. Calculate the RN vector that measures the average statistical value of the normal vector of each point in PPC, and establish the characteristic plane P of PPC based on the RN vector ⊥ and P || ; Calculate P ⊥ The rotation matrix T of the XOZ plane relative to the standard orbit point cloud model ⊥ ; Calculate P || The rotation matrix T of the YOZ plane relative to the standard orbit point cloud model || , use T ⊥ and T || Perform rotation transformation on 3DPC; Calculate the X value of the centroid of the 3DPC. If the X value of the centroid is less than the preset threshold, rotate the 3DPC 180° around the Y axis to obtain the processed three-dimensional point cloud 3DPCⅠ. S3. Determine whether 3DPCⅠ meets the posture correction requirement according to at least one of the X value, Y value, and Z value of the processed three-dimensional point cloud 3DPCⅠ. If the posture correction requirement is met, obtain the three-dimensional point cloud 3DPCⅡ after posture correction.

2. The method for correcting the position and posture of a track point cloud according to claim 1, characterized in that: The step of fitting the 3DPC into the plane point cloud PPC in step S1 includes dividing the 3DPC into t pieces of point clouds 3DPC[i], 0≤i<t, and fitting the 3DPC[i] into t pieces of plane point clouds PPC[i]; In step S2, 3DPC is rotated based on PPC[i] to obtain a processed three-dimensional point cloud 3DPCⅠ[i]; In step S3, it is determined whether 3DPCⅠ[i] meets the posture correction requirements. If so, the three-dimensional point cloud 3DPCⅡ after posture correction is obtained.

3. The method for correcting the position and posture of a track point cloud according to claim 2, characterized in that: Step S3 includes the following steps: S31, comparing the similarity of each 3DPCⅠ[i] with the corresponding part of the standard track point cloud model; S32. According to the similarity between each 3DPCⅠ[i] and the corresponding part of the standard track point cloud model, the pose correction requirement is judged in descending order; if a 3DPCⅠ[i] that meets the pose correction requirement appears, then 3DPCⅠ meets the pose correction requirement and 3DPCⅡ is obtained.

4. The method for correcting the position and posture of a track point cloud according to claim 2, characterized in that: In step S1, PPC[i] is obtained by performing plane model segmentation on 3DPC[i].

5. A method for correcting the position and posture of a track point cloud according to any one of claims 1 to 4, characterized in that: In step S3, the condition that 3DPCⅠ meets the posture correction requirement is that the maximum Y value of 3DPCⅠ is less than or equal to the preset threshold; The condition for 3DPCⅠ[i] to meet the posture correction requirements is that the maximum Y value of at least one 3DPCⅠ[i] is less than or equal to the preset threshold.

6. A method for correcting the position and posture of a track point cloud according to any one of claims 1 to 4, characterized in that: The 3DPC in step S1 is the data of the rail head and rail waist.

7. A method for correcting the position and posture of a track point cloud according to claim 6, characterized in that: In step S1, data of the rail head and rail waist are selected by automatic area selection and / or manual area selection.

8. A method for correcting the position and posture of a track point cloud according to any one of claims 1 to 4, characterized in that: The 3DPC is downsampled before step S1.

9. A method for correcting the position and posture of a track point cloud according to any one of claims 1 to 4, characterized in that: The following formula is used to calculate the RN vector in step S2: In the formula is the RN vector, are the components of the RN vector in the X, Y, and Z directions, respectively, N[i] x 、N[i] y 、N[i] z They are the components of the N[i] vector in the X, Y, and Z directions respectively. N[i] is the normal vector of the i-th point in PPC, 0≤i<n, and n is the number of points in PPC.

10. A method for correcting the position and posture of a track point cloud according to any one of claims 1 to 4, characterized in that: Step S2 comprises the following steps: Calculate the normal vector N[i] of each point P[i] in PPC, 0≤i<n, n is the number of points in PPC; calculate the RN vector that measures the average statistical value of N[i]; calculate the positional relationship d[i] of P[i] to the RN vector respectively, let the maximum value in d[i] be d[s], and the point corresponding to d[s] be P[s]; use RN vector and P[s] to establish plane P ⊥ and plane P || , where P ⊥ Perpendicular to RN vector, P || Parallel to the RN vector; Calculate P ⊥ The angle θ between the standard orbit point cloud model coordinate system plane XOZ and the corresponding rotation axis vector V ⊥ , calculate around V ⊥ The rotation matrix T of the rotation θ ⊥ ; Calculate P || The angle between the YOZ plane and the standard track point cloud model coordinate system and the corresponding rotation axis vector V || , calculate around V || Rotation The rotation matrix T || ; Use T ⊥ and T || Perform rotation transformation on 3DPC; Calculate the X value of the centroid of the 3DPC. If the X value of the centroid is less than the preset threshold, rotate the 3DPC 180° around the Y axis to obtain the processed three-dimensional point cloud 3DPCⅠ.

11. A method for correcting the position and posture of a track point cloud according to claim 10, characterized in that: The following formula is used to calculate d[i] and select d[s] in step S2: d[s]=max(d[i])(0≤i<n) Where x[i], y[i], and z[i] are the coordinates of point P[i] in the X, Y, and Z directions, respectively. are the components of the RN vector in the X, Y, and Z directions respectively.

12. The method for correcting the position and posture of a track point cloud according to claim 10, characterized in that: In step S2, plane P is established. ⊥ and plane P || The following formula is used: P ⊥ ： P || ： In the formula are the components of the RN vector in the X, Y, and Z directions respectively, x[s], y[s], and z[s] are the coordinate values ​​of the point P[s] in the X, Y, and Z directions respectively; X, Y, and Z are the three axes in three-dimensional space.

13. The method for correcting the position and posture of a track point cloud according to claim 10, characterized in that: In step S2, V is calculated ⊥ and V || The following formula is used: Where V ⊥x 、V ⊥y 、V ⊥z V ⊥ Components in the X, Y, and Z directions; V ||x 、V ||y 、V ||z V || Components in the X, Y, and Z directions; are the components of the RN vector in the X, Y, and Z directions respectively; x[s], y[s], and z[s] are the coordinate values ​​of the point P[s] in the X, Y, and Z directions respectively.

14. The method for correcting the position and posture of a track point cloud according to claim 10, characterized in that: In step S2, θ and The following formula is used: Where θ is the plane P ⊥ The angle between the XOZ plane and are the components of the RN vector in the X, Y, and Z directions respectively; x[s], y[s], and z[s] are the coordinate values ​​of the point P[s] in the X, Y, and Z directions respectively.

15. An application of a method for correcting the position and posture of a track point cloud, characterized in that: The following steps are involved: A. Obtaining 3DPCⅡ by a posture correction method of a track point cloud as described in any one of claims 1 to 14; B. According to the type of track (1) identified by 3DPCⅡ, the corresponding standard track point cloud model is selected to perform deviation analysis on 3DPCⅡ to obtain the deviation analysis result of 3DPCⅡ.

16. The application of the method for correcting the position and posture of a track point cloud according to claim 15, characterized in that: Step B includes the following steps: B1. Roughly align 3DPCⅡ with the corresponding standard track point cloud model; B2. Accurately align 3DPCⅡ with the corresponding standard track point cloud model; B3. Calculate the deviation between 3DPCⅡ and the corresponding standard track point cloud model.

17. The application of the method for correcting the position and posture of a track point cloud according to claim 16, characterized in that: Before step B1, the rail head and rail waist in 3DPCⅡ are segmented; in step B1, the centroid of the rail head of 3DPCⅡ is aligned with the centroid of the rail head of the standard rail point cloud model, or the centroid of the rail waist of 3DPCⅡ is aligned with the centroid of the rail waist of the standard rail point cloud model.

18. Application of the method for correcting the position and posture of a track point cloud according to any one of claims 15 to 17, characterized in that: In step B, m sections 3DPCS[i] are cut out from 3DPCⅡ along the Z axis, 0≤i<m, and multiple standard sections corresponding to 3DPCS[i] are cut out from the standard track point cloud model; the similarity between 3DPCS[i] and the standard sections is compared one by one to identify the control section; and the corresponding standard track point cloud model is selected according to the control section.

19. Application of the method for correcting the position and posture of a track point cloud according to any one of claims 15 to 17, characterized in that: The method also includes establishing a track database; the track database at least stores the deviation analysis results of 3DPCⅡ.

20. The application of the method for correcting the position and posture of a track point cloud according to claim 19, characterized in that: The deviation analysis results of the 3DPCⅡ include the track geometry parameter curve obtained by traversing the control section along the longitudinal direction of the track.

21. The application of the method for correcting the position and posture of a track point cloud according to claim 19, characterized in that: According to the deviation analysis results of 3DPCⅡ in the track database, a track geometry parameter trend chart is established, and a maintenance plan is formulated based on the track geometry parameter trend chart.

22. The application of the method for correcting the position and posture of a track point cloud according to claim 21, characterized in that: The following steps are involved in creating a track geometry trend graph: Establish a track geometry parameter line graph based on historical data in the track database; use the prediction model to predict track geometry parameters; and establish a track geometry parameter trend graph.

23. A three-dimensional geometric state measuring device for a track, characterized in that: It comprises a processor and a memory; the memory is communicatively connected to the processor; the memory stores instructions that can be executed by the processor, and the processor can execute a method for correcting the posture of a track point cloud as described in any one of claims 1 to 14 through the instructions.

24. A three-dimensional geometric state measurement system for a track, characterized in that: It comprises a three-dimensional track geometric state measuring device as described in claim 23, a surface structured light emitter and a visual sensor; the surface structured light emitter is used to project surface structured light onto the track; the visual sensor is used to obtain a track image with grating stripes; the visual sensor is communicatively connected to a processor.

25. A track three-dimensional geometric state measurement system according to claim 24, characterized in that: The invention also comprises a crossbeam (2) and a wheel assembly (3); the wheel assembly (3) is arranged on both sides of the crossbeam (2) along the horizontal direction of the track; at least one of the wheel assemblies (3) is connected to the crossbeam (2) through an adaptive device (5) so as to be able to swing around a vertical axis relative to the crossbeam (2); the spacing of the wheel assemblies (3) corresponds to the track gauge of the track (1); the wheel assembly (3) can move along the track (1); and the surface structured light emitter and the visual sensor are connected to the crossbeam (2).

26. A track three-dimensional geometric state measurement system according to claim 25, characterized in that: The adaptive device (5) comprises a first mounting seat (21) and a second mounting seat (51), wherein the first mounting seat (21) is capable of rotating relative to the second mounting seat (51) along an axis in a vertical direction; one of the first mounting seat (21) and the second mounting seat (51) is connected to the crossbeam (2), and the other of the first mounting seat (21) and the second mounting seat (51) is connected to the wheel assembly (3).

27. A track three-dimensional geometric state measurement system according to claim 26, characterized in that: The other end of the adaptive device (5) is slidably connected to the wheel assembly (3) or the crossbeam (2), and the slidable connection comprises a slide rail (33) whose axis is arranged transversely along the track.

28. A track three-dimensional geometric state measurement system according to any one of claims 24 to 27, characterized in that: The surface structured light emitter and the visual sensor are connected to the crossbeam (2) via a cantilever connecting rod (6); the cantilever connecting rod (6) is used to adjust the position and angle of the surface structured light emitter and the visual sensor relative to the track (1).

Images