A steel rail chamfering path generation method and chamfering and polishing system

Abstract

This invention discloses a method for generating a chamfering machining path for steel rails and a chamfering and grinding system, solving the problem of inaccurate automatic chamfering of variable cross-section machining edges in current rail processing. The chamfering process is performed synchronously with the gantry milling machine machining process, offering advantages such as high efficiency and good accuracy. The system includes a robot mounted inverted on a gantry milling machine. The robot's end effector has a radially floating high-speed electric spindle and an automatic tool changer, capable of automatically retrieving and placing chamfering and grinding tools and robot calibration tools from the tool magazine. The path generation method of this chamfering and grinding system includes 3D laser acquisition of workpiece point cloud data, point cloud data processing, feature extraction, and registration of the blank point cloud with the theoretical model point cloud based on process features, resulting in machining trajectory generation. This invention features a compact structure, strong versatility, and high processing efficiency.

Description

Technical Field

[0001] This invention belongs to the field of chamfering and grinding of complex spatial edges of railway tracks, specifically involving a method for generating chamfering processing paths for railway tracks and a chamfering processing and grinding system. Background Technology

[0002] After the heavy railway rails are processed, the joint surfaces need to be rounded and deburred.

[0003] The aforementioned rail processing is all done by gantry milling machines, while rounding and deburring are usually done manually. Manual chamfering is performed on the curved edges of complex intersecting surfaces using tools such as angle grinders. With increasing demands for product precision and production capacity, manual chamfering exhibits disadvantages such as low processing efficiency, long processing time, and poor accuracy. Furthermore, rail processing requires a large workstation area, affecting workshop space utilization. Regarding point cloud processing technologies, existing mainstream point cloud simplification methods result in the loss of some point cloud features after simplification. In point cloud feature extraction, existing general algorithms randomly sample and extract certain features from the point cloud, failing to accurately pinpoint specific features. For point cloud registration, the ICP algorithm, used for fine registration, often gets stuck in local optima on complex point clouds, failing to achieve the expected accurate registration results. Developing point cloud feature extraction and registration methods based on processing technology features to guide processing operations will significantly improve manufacturing precision and efficiency. Summary of the Invention

[0004] This invention addresses the problems in existing technologies by providing a method for generating rail chamfering processing paths and a chamfering processing and grinding system, which can better improve the accuracy and efficiency of rail chamfering, and is applicable to different types of rails and the rounding of complex intersections with different complex curved surfaces.

[0005] To solve the above technical problems, the present invention provides the following technical solution: a method for generating a rail chamfering processing path, comprising the following steps:

[0006] S1. Determine the tool coordinate system of the chamfering and grinding tool, and then perform a 3D model scan of the long rail to be processed to obtain the point cloud data of the long rail to be processed.

[0007] S2. Fit the feature surface of the intersection point of the fillet and the adjacent surface on the cross section of the theoretical model to obtain the point cloud normal vector at the intersection of the fillet surface. Then, obtain the path curve of the trajectory of the center of the fillet edge in the theoretical model, which is the theoretical processing trajectory.

[0008] S3. The point cloud data of the long railway track to be processed and the point cloud data of the theoretical model are simplified using an improved point cloud simplification algorithm. Specifically, the simplification method of non-substitute points is used to simplify the massive point cloud data, while retaining the original point cloud features, to obtain the simplified point cloud data of the long railway track to be processed and the point cloud data of the theoretical model.

[0009] S4. Extract the point cloud feature surface based on the processing technology from the simplified theoretical model point cloud data to obtain the processing technology feature plane;

[0010] S5. Perform processing technology feature registration on the theoretical model point cloud and the long rail point cloud data to be processed. Add constraints during registration and use a layered solution method. After registration, transform the theoretical processing trajectory and direction vector into the actual processing coordinate system.

[0011] S6. Obtain the corresponding three-dimensional rotation matrix based on the tool coordinate system and the actual machining coordinate system of the chamfering grinding tool. After obtaining a unique set of quaternions, the axis direction of the tool coordinate system of the chamfering grinding tool is perpendicular to the normal vector of the fillet surface of the theoretical model under the machining coordinate system, thus obtaining the machining path.

[0012] Furthermore, the aforementioned step S1 includes the following sub-steps:

[0013] S101. Perform tool coordinate system calibration on the robot's fillet tool to obtain the tool coordinate system. The tool coordinate system of the robot deburring tool is calibrated to obtain the tool coordinate system. The tool coordinate system is calibrated using the 3D laser scanner at the robot's end effector to obtain the tool coordinate system. ;

[0014] S102. The robot drives a 3D laser scanner to perform a 3D scan of the intersecting surfaces of the long rail to be processed, and obtains the point cloud data of the long rail to be processed.

[0015] Furthermore, the aforementioned step S2 specifically involves: [The text abruptly shifts to a different topic] ...the intersection of the fillet on the cross-section of the theoretical model and the adjacent surface... , By fitting feature surfaces, the normal vector of the point cloud at the intersection of rounded corners is obtained. , According to the fillet radius is The coordinates of the center of the circle are The curve of the center path of the rounded corner surface is fitted, and this path is the path of the machining tool.

[0016] Furthermore, the improved point cloud simplification algorithm in step S3 mentioned above includes the following sub-steps:

[0017] S301. Calculate the maximum value of the coordinates of the entire point cloud data. and minimum value Then, construct a bounding cube based on the maximum values ​​in each direction, with bounding side lengths... Calculate using the following formula:

[0018] ;

[0019] S302. Divide the enclosing cube into smaller cubes according to the preset side length l;

[0020] S303. Determine the number of point clouds within each small cube. Is it greater than the set threshold? If the number of point clouds inside the cube Greater than this threshold Then, random samples are taken from the points within the cube, with the number of samples being... One, of which Rounding down, if If so, then all points in the cube are retained;

[0021] S304. Save the point cloud obtained from the sampling.

[0022] Furthermore, the aforementioned step S4 includes the following sub-steps:

[0023] S401. Calculate the maximum and minimum values ​​of the point cloud in each direction. ;

[0024] S402. Divide the point cloud data into cubes. The amount of point cloud data in each small cube is represented as follows:

[0025] ,

[0026] Among them, It is the size of the cube. It is the total number of points in the point cloud. This is the cube occupancy rate, representing the ratio of the number of cubes with points to the total number of cubes. It is the expected number of points in the cube;

[0027] S403. Fitting the planar features yields the following plane equation:

[0028] ,

[0029] The coefficients of the plane equation are as follows: ,

[0030] in , The index of the point set.

[0031] , Represents a point;

[0032] S404, the distances between the midpoints of each cube are as follows:

[0033] ,

[0034] Where j is the index of the point set, and r is the number of points.

[0035] S405. Based on the plane equation, use a known open-source library to fit the data and obtain the processing feature surface.

[0036] Furthermore, the aforementioned step S5 includes the following sub-steps:

[0037] S501. Perform preliminary constraints on point cloud data registration. The point cloud of the long railway track to be processed will be enveloped by the point cloud of the CAD model, i.e., the distance between corresponding points. or normal distance Always positive, that is:

[0038] ;

[0039] In the formula, R is the rotation matrix of the point cloud spatial transformation, and T is the translation matrix of the point cloud spatial transformation.

[0040] S502. Establish a mathematical model for the point cloud data, as follows:

[0041]

[0042] In the formula, (R,T) is the least squares constraint function;

[0043] S503, Minimum margin constraint, as shown in the following formula:

[0044]

[0045] in, This refers to the minimum machining allowance required for machining the feature surfaces. for The set of subscripts; M j A class representing feature surfaces with different margin requirements;

[0046] S504. There is a tolerance constraint between the theoretical model surface and the workpiece surface. The following constraint conditions apply.

[0047]

[0048] in, This represents the lower limit of the permissible variation. This represents the lower limit of the permissible variation. for Given a set of subscripts, assume there are N classes of surfaces with different allowable variation constraints;

[0049] S505, unifies the minimum margin constraint and tolerance constraint into a single expression. The mathematical model for registration constraints is expressed as follows:

[0050] ;

[0051] S506. Point cloud registration based on process features, that is, obtaining the correspondence between the theoretical model and the long rail to be processed, and obtaining a set of spatial rotation and translation transformation matrices:

[0052]

[0053]

[0054] Parameter explanation: Where R is the spatial rotation transformation matrix, They are respectively Rotation transformation matrices in three directions, They are respectively The angular changes in three directions, where T is the spatial translation transformation matrix. These represent the changes in movement in three directions;

[0055] The spatial transformation relationship between the two coordinate systems is as follows:

[0056]

[0057] After registration according to the process characteristics, the coordinates of the theoretical model are transformed into the actual machining coordinate system.

[0058] Furthermore, the aforementioned step S6 includes the following sub-steps:

[0059] S601. Based on the pose relationship between the tool coordinate system and the machining coordinate system of the fillet tool, the corresponding rotation matrix is ​​obtained as follows:

[0060] ,

[0061] S602. Obtain the unique set of quaternions corresponding to the rotation matrix using the following formula:

[0062]

[0063] In the formula, symbols and Same number, symbols and Same number, symbols and Same number.

[0064] Furthermore, the aforementioned method for generating rail chamfering processing paths extracts point cloud feature surfaces based on processing technology from the simplified theoretical model point cloud data, including but not limited to planar and cylindrical surfaces.

[0065] Another aspect of this invention proposes a chamfering and grinding system based on a method for generating chamfering machining paths for rails, comprising: a robot, and a gantry milling machine and a tool changer connected to the robot via Ethernet; a robot end effector is installed at the end of the robot, the robot end effector including a radially floating high-speed electric spindle and a 3D laser scanner, the high-speed electric spindle having an automatic tool changer mechanism.

[0066] As the robot follows the gantry milling machine, it drives a 3D laser scanner to complete a 3D scan of the long rail. The robot drives an automatic tool changer to automatically pick up and put in chamfering tools, grinding heads, and calibration devices from the tool magazine. The chamfering and grinding of the rail are completed according to the path generated by the rail chamfering processing path generation method proposed in this invention.

[0067] Furthermore, the aforementioned chamfering and grinding system based on the rail chamfering processing path generation method also includes a chip removal pneumatic nozzle, used to clean up iron filings generated during chamfering and grinding, as well as during rounding and deburring processes.

[0068] Compared to existing technologies, the beneficial technical effects of the present invention using the above technical solution are as follows: The chamfering and grinding system based on the rail chamfering processing path generation method provided by the present invention has a compact structure, low investment, and is applicable to the rounding of complex junction surfaces of various types of rails. The 3D laser scanner can store a large amount of data and can acquire 3D point cloud data of various long rails. In addition, the robot end has a high-speed radial floating electric spindle, which reduces the risk of damage from rigid removal by utilizing floating. The robot is mounted upside down on a gantry milling machine, without occupying extra space. The rounding of corners and rail processing are completed simultaneously using the rail chamfering processing path generation method of the present invention, without occupying extra time and with high efficiency. The chamfering and grinding processing accuracy is high, and there are no obvious junction marks between the two surfaces after the rounding is completed. The point cloud processing algorithm is efficient and has high registration accuracy. The processing trajectory and robot posture parameters during processing are automatically generated to guide the robot to process according to the corresponding posture. This technology addresses the challenges of determining datum values, controlling allowances, and low efficiency in traditional rail machining. By utilizing digital processing technology for raw materials, it quickly registers allowance constraints with CAD models, outputting and executing robot-readable machining trajectory programs. This improves the accuracy, reliability, and efficiency of rail machining, achieving the goals of eliminating manual labor, increasing labor productivity, comprehensively enhancing the automation and integration of manufacturing equipment, improving the intelligence level of common basic processes, reducing product production cycles, controlling production costs, and playing a positive role in promoting and ensuring the production and equipment of various rail models. Attached Figure Description

[0069] Figure 1 This is a schematic diagram of the chamfering and polishing system according to an embodiment of this application.

[0070] Figure 2 This is a schematic diagram of the end effector of the robot according to an embodiment of this application.

[0071] Figure 3 This is a schematic diagram of the calibration device structure according to an embodiment of this application.

[0072] Figure 4 This is a schematic diagram of rail processing according to an embodiment of this application.

[0073] Figure 5 This is a comparison diagram of the effects of chamfering and grinding on the workpiece in the embodiments of this application.

[0074] Figure 6 This is a schematic diagram of the point cloud simplification algorithm in an embodiment of this application.

[0075] Figure 7 This is a schematic diagram of the algorithm for solving the fillet center and normal vector of the theoretical model in the embodiment of this application.

[0076] Figure 8This is a schematic diagram of the registration of the theoretical model point cloud and the scanned blank point cloud based on process features in an embodiment of this application.

[0077] Figure 9 This is a schematic diagram of the point cloud layer-by-layer registration process in an embodiment of this application.

[0078] Figure 10 This is a schematic diagram of the anti-collision communication between a gantry milling machine and a robot.

[0079] Figure 11 This is a schematic diagram of the overall processing method.

[0080] In the diagram: 1-Robot; 2-Gantry milling machine; 3-Tool changer; 4-Robot end effector; 5-Rail workpiece;

[0081] 401-Radial floating high-speed electric spindle; 402-Chip removal pneumatic nozzle; 403-Chamfering tool; 404-3D laser scanner; 405-Calibration device; 406-Grinding head. Detailed Implementation

[0082] To better understand the technical content of the present invention, specific embodiments are described below in conjunction with the accompanying drawings.

[0083] In this invention, various aspects of the invention are described with reference to the accompanying drawings, in which numerous illustrative embodiments are shown. Embodiments of the invention are not limited to those depicted in the drawings. It should be understood that the invention is implemented through any of the various concepts and embodiments described above, as well as the concepts and embodiments described in detail below, because the concepts and embodiments disclosed herein are not limited to any particular implementation. Furthermore, some aspects of the invention disclosed may be used alone or in any suitable combination with other aspects of the invention disclosed.

[0084] This invention is applied to railway rails, specifically to rail workpiece 5. The rail workpiece is typically machined on a 25m gantry milling machine 2, generally involving two-stage planing, three-stage planing, or five-stage planing. Figure 4 As shown.

[0085] refer to Figure 11 This invention provides a method for generating a rail chamfering processing path, comprising the following steps:

[0086] S1. Determine the tool coordinate system of the chamfering and grinding tool, and then perform a 3D model scan of the long rail to be processed to obtain the point cloud data of the long rail to be processed.

[0087] S2. Obtain the path curve of the center of the fillet edge of the theoretical model, i.e. the theoretical machining trajectory, and obtain the direction vector of the machining tool of the theoretical model using the normal vector of the intersecting surface.

[0088] S3. The point cloud data of the long railway track to be processed and the point cloud data of the theoretical model are simplified using an improved point cloud simplification algorithm. Specifically, the simplification method of non-substitute points is used to simplify the massive point cloud data, while retaining the original point cloud features, to obtain the simplified point cloud data of the long railway track to be processed and the point cloud data of the theoretical model.

[0089] S4. Extract the point cloud feature surfaces based on the processing technology from the simplified theoretical model point cloud data to obtain the processing technology feature surfaces;

[0090] S5. Based on the processing technology feature surface, register the theoretical model point cloud and the long rail point cloud data to be processed. Add constraints during registration and adopt a layered solution method. After registration, transform the theoretical processing trajectory and direction vector into the actual processing coordinate system.

[0091] S6. Obtain the corresponding three-dimensional rotation matrix based on the tool coordinate system and the actual machining coordinate system of the chamfering grinding tool. After obtaining a unique set of quaternions, the axis direction of the tool coordinate system of the chamfering grinding tool is perpendicular to the normal vector of the fillet surface of the theoretical model under the machining coordinate system, thus obtaining the machining path.

[0092] Preferably, step S1 includes the following sub-steps:

[0093] S101. Take calibration device 405 from tool changer 3 for robot 1, and calibrate the tool coordinate system of the fillet tool using the 6-point calibration method to obtain the tool coordinate system. Robot 1 retrieves calibration device 405 from tool changer 3 and uses a 6-point calibration method to calibrate the tool coordinate system of the grinding and deburring tool, thus obtaining the tool coordinate system. Robot 1 retrieves calibration device 405 from tool changer 3 and uses a 6-point calibration method to calibrate the tool coordinate system of the 3D laser scanner, thus obtaining the tool coordinate system. .

[0094] like Figure 4 As shown, during the secondary planing process of the steel rail, the gantry milling machine 5 drives the robot 1 to move synchronously. The 3D scanner installed at the end of the robot 1 performs a 3D scan on the semi-finished product after the secondary planing process, and obtains the point cloud data of the two faces adjacent to the edge line on the left side of the secondary planing process that needs to be rounded.

[0095] S102. The robot drives a 3D laser scanner to perform a 3D scan of the intersecting surfaces of the long rail to be processed, and obtains the point cloud data of the long rail to be processed.

[0096] Since the theoretical model is designed using 3D software, the coordinate positions of the feature points on the rail in the 3D model are fixed and can be directly obtained, such as... Figure 7As shown, the intersection point of the fillet of the cross-section and the adjacent face is fixed. Therefore, in the point cloud data of the theoretical model, the spatial surface can be directly fitted using the points of the adjacent faces, and the normal vector on the cross-section can be obtained. The center of the circle can then be calculated using the radius data of the fillet. Coordinate data. For example... Figure 5 As shown, the intersection of the fillet on the cross section and the adjacent surface. , Given points, using MATLAB and the PCL library, NURBS surface fitting can be performed on adjacent point cloud data to obtain the normal vectors of the two points. , The radius of the fillet is The coordinates of the center of the circle are Finally, a series of points with the center coordinates are obtained. During actual machining, the tool's radius offset along the center trajectory is the theoretical machining trajectory.

[0097] Point cloud simplification is performed on the semi-finished rail model obtained from scanning. To preserve the original features of the rail, the traditional point cloud simplification algorithm is modified, and a non-substitute point simplification method is used to simplify the massive point cloud data. For example... Figure 6 As shown, preferably, the improved point cloud simplification algorithm in step S3 includes the following sub-steps:

[0098] S301. Calculate the maximum value of the coordinates of the entire point cloud data. and minimum value Then, construct a bounding cube based on the maximum values ​​in each direction, with bounding side lengths... Calculate using the following formula:

[0099] ;

[0100] S302. Divide the enclosing cube into smaller cubes according to the preset side length l;

[0101] S303. Determine the number of point clouds within each small cube. Is it greater than the set threshold? If the number of point clouds inside the cube Greater than this threshold Then, random samples are taken from the points within the cube, with the number of samples being... One, of which Rounding down, if If so, then all points in the cube are retained;

[0102] S304. Save the point cloud obtained from the sampling.

[0103] The improved point cloud simplification algorithm performs one iteration of point cloud simplification. For massive point cloud data, multiple simplifications are performed by modifying the simplification parameters. The point cloud data of each simplification is saved, and the final transformation matrix can be solved by performing layer-by-layer registration in the subsequent feature-based registration process.

[0104] A schematic diagram of cloud plane feature extraction. Preferably, step S4 includes the following sub-steps:

[0105] S401. Calculate the maximum and minimum values ​​of the point cloud in each direction. ;

[0106] S402. Divide the point cloud data into cubes. The amount of point cloud data in each small cube is represented as follows:

[0107] ,

[0108] Among them, It is the size of the cube. It is the total number of points in the point cloud. This is the cube occupancy rate, which is the ratio of the number of cubes with points to the total number of cubes. Empirically, it should be around 0.3. This is the expected number of points in the cube, which is empirically estimated to be around 50 points.

[0109] S403. Fitting the planar features yields the following plane equation:

[0110] ,

[0111] The coefficients of the plane equation are as follows: ,

[0112] in , Point set index

[0113] , Represents a point;

[0114] S404, the distances between the cubes are as follows:

[0115] ,

[0116] Where j is the point index and r is the total number of points.

[0117] S405. For the three-dimensional surface features, use known open-source libraries such as MATLAB and PCL to fit the data, obtain the fitted plane equation, and then obtain the machining process feature plane.

[0118] In step S5, the point cloud model of the rail to be processed is registered with the theoretical 3D model based on the processing technology features. Since the point cloud data obtained by scanning is semi-processed 3D data, some surfaces can coincide with the surfaces in the theoretical model, such as... Figure 8 As shown, the breakpoint line represents the point cloud of the workpiece cross-section contour, and the solid line represents the theoretical cross-section contour point cloud data. Specifically, it includes the following sub-steps:

[0119] S501. Perform preliminary constraints on point cloud data registration. The point cloud of the long railway track to be processed will be enveloped by the point cloud of the CAD model, i.e., the distance between corresponding points. or normal distance Always positive, that is:

[0120] ;

[0121] In the formula, R is the rotation matrix of the point cloud spatial transformation, and T is the translation matrix of the point cloud spatial transformation.

[0122] S502. Establish a mathematical model for the point cloud data, as follows:

[0123]

[0124] In the formula, (R,T) is the least squares constraint function;

[0125] S503, Minimum margin constraint, as shown in the following formula:

[0126]

[0127] in, This refers to the minimum machining allowance required for machining the feature surfaces. for The set of subscripts; M j A class representing feature surfaces with different margin requirements.

[0128] To improve registration efficiency, the registration process is divided into two steps: coarse registration and fine registration. Coarse registration uses principal component analysis, which can quickly align two point cloud data in one direction. However, the final result of coarse registration cannot guarantee the accuracy of point cloud registration, so it is necessary to continue to constrain the registration by processing features.

[0129] S504. There is a tolerance constraint between the theoretical model surface and the workpiece surface. The following constraint conditions apply.

[0130]

[0131] in, This represents the lower limit of the permissible variation. This represents the lower limit of the permissible variation. for Given a set of subscripts, assume there are N classes of surfaces with different allowable variation constraints;

[0132] S505, unifies the minimum margin constraint and tolerance constraint into a single expression. The mathematical model for registration constraints is expressed as follows:

[0133] ;

[0134] S506. Point cloud registration based on process features, that is, obtaining the correspondence between the theoretical model and the long rail to be processed, and obtaining a set of spatial rotation and translation transformation matrices:

[0135]

[0136]

[0137] Parameter explanation: Where R is the spatial rotation transformation matrix, They are respectively Rotation transformation matrices in three directions, They are respectively The angular changes in three directions, where T is the spatial translation transformation matrix. These represent the changes in movement in three directions.

[0138] The spatial transformation relationship between the two coordinate systems is as follows:

[0139]

[0140] After registration according to the process characteristics, the coordinates of the theoretical model are transformed into the actual machining coordinate system.

[0141] After the initial simplification of the massive point cloud data, if the registration of a smaller point cloud does not meet the requirements, the workpiece is directly judged to be scrapped. If the registration of a smaller point cloud is successful, then the registration of a larger point cloud is performed, and so on, until the final complete point cloud meets the registration requirements, at which point the registration process ends.

[0142] The hierarchical registration method in this invention has three scenarios: First, if the simplest point cloud registration does not meet the conditions, the process terminates directly. Second, if the simplest point cloud meets the conditions, registration is performed layer by layer until all conditions are met, at which point the process ends. Finally, if the simplest point cloud meets the conditions, but some points do not meet the conditions during layer-by-layer registration, these points are added to the point set and registered again until the conditions are finally met, at which point the process ends. Clearly, the first two scenarios can quickly terminate the registration process; only the last scenario requires multiple verifications, but the overall registration efficiency is significantly improved. The flowchart of the hierarchical registration method is shown below. Figure 9As shown.

[0143] Preferably, step S6 includes the following sub-steps:

[0144] S601. Based on the pose relationship between the tool coordinate system and the machining coordinate system of the fillet tool, the corresponding rotation matrix is ​​obtained as follows:

[0145] ,

[0146] S602. Obtain the unique set of four elements corresponding to the rotation matrix using the following formula:

[0147]

[0148] In the formula, symbols and Same number, symbols and Same number, symbols and Same number.

[0149] During the movement of robot 1, the robot's posture is controlled using quaternions. While the rail is being processed on the gantry milling machine 2, robot 1 synchronizes with the forward direction of the gantry milling machine 2 according to the conversion path and posture. After the rail is processed, robot 1 also completes the rounding of the rail.

[0150] like Figure 1 and Figure 2 As shown, another aspect of the present invention provides a chamfering and grinding system based on a method for generating chamfering machining paths for rails, comprising: a robot 1, and a gantry milling machine 2 and a tool changer 3 connected to the robot 1 via Ethernet; a robot end effector 4 is installed at the end of the robot 1, the robot end effector 4 including a radially floating high-speed electric spindle 401 and a three-dimensional laser scanner 404, the high-speed electric spindle 401 having an automatic tool changer mechanism. The robot 1 is connected to the gantry milling machine 2 via Ethernet, and the communication diagram between the two is shown in the figure. Figure 10 Establish real-time communication for information exchange and mutual cooperation to ensure that the robot does not collide when it continues to complete the previous processing step and then moves to the next processing step after the gantry milling machine finishes a single processing step.

[0151] As robot 1 follows the movement of gantry milling machine 2, it drives 3D laser scanner 404 to complete 3D scanning of long rail 5. Robot 1 utilizes a rail chamfering machining path generation method of this invention to drive an automatic tool changer to automatically retrieve and place chamfering tools 403, grinding heads 406, and calibration devices 405 from tool magazine 3, completing the chamfering and grinding of the rail according to the path. This system also includes a chip removal pneumatic nozzle 402 for cleaning up iron filings generated during machining, rounding, and deburring processes, including chamfering and grinding. The calibration device includes... Figure 3 Using the system of this invention, the chamfering and grinding effect of long railway rails is significantly improved. Figure 5 As shown. Although the present invention has been described above with reference to preferred embodiments, it is not intended to limit the invention. Those skilled in the art can make various modifications and refinements without departing from the spirit and scope of the invention. Therefore, the scope of protection of the present invention shall be determined by the claims.

Claims

1. A method for generating a rail chamfering processing path, characterized in that, Includes the following steps: S1. Determine the tool coordinate system of the chamfering and grinding tool, and then perform a 3D model scan of the long rail to be processed to obtain the point cloud data of the long rail to be processed. S2. Fit the feature surface of the intersection point of the fillet and the adjacent surface on the cross section of the theoretical model to obtain the point cloud normal vector at the intersection of the fillet surface. Then, obtain the path curve of the trajectory of the center of the fillet edge in the theoretical model, which is the theoretical processing trajectory. S3. The point cloud data of the long railway track to be processed and the point cloud data of the theoretical model are simplified using an improved point cloud simplification algorithm. The improved point cloud simplification algorithm uses a non-substitute point simplification method to simplify the massive point cloud data while retaining the original point cloud features, resulting in the simplified point cloud data of the long railway track to be processed and the point cloud data of the theoretical model; including the following sub-steps: S301. Calculate the maximum value of the coordinates of the entire point cloud data. and minimum value Then, construct a bounding cube based on the maximum values ​​in each direction, with bounding side lengths... Calculate using the following formula: ； S302. Divide the enclosing cube into smaller cubes according to the preset side length l; S303. Determine the number of point clouds within each small cube. Is it greater than the set threshold? If the number of point clouds inside the cube Greater than this threshold Then, random samples are taken from the points within the cube, with the number of samples being... One, of which Rounding down, if If so, then all points in the cube are retained; S304. Save the point cloud obtained from the sampling; S4. Extract the point cloud feature surface based on the processing technology from the simplified theoretical model point cloud data to obtain the processing technology feature plane; S5. Register the theoretical model point cloud and the long railway track point cloud data to be processed using processing technology features. Add constraints during registration and use a layered solution method. After registration, transform the theoretical processing trajectory and direction vector into the actual processing coordinate system. Step S5 includes the following sub-steps: S501. Perform preliminary constraints on point cloud data registration. The point cloud of the long railway track to be processed will be enveloped by the point cloud of the CAD model, i.e., the distance between corresponding points. or normal distance Always positive, that is: ； In the formula, R is the rotation matrix of the point cloud spatial transformation, and T is the translation matrix of the point cloud spatial transformation. S502. Establish a mathematical model for the point cloud data, as follows: ， In the formula, (R,T) is the least squares constraint function; S503, Minimum margin constraint, as shown in the following formula: ， in, This refers to the minimum machining allowance required for machining the feature surfaces. for The set of subscripts; M j A class representing feature surfaces with different margin requirements; S504. There is a tolerance constraint between the theoretical model surface and the workpiece surface. The following constraint conditions apply: ， in, This represents the lower limit of the permissible variation. This represents the upper limit of the allowed variation. for The set of subscripts, assuming there are N types of surfaces with different allowable variation constraints; S505, unifies the minimum margin constraint and tolerance constraint into a single expression. The mathematical model for registration constraints is expressed as follows: ； S6. Obtain the corresponding three-dimensional rotation matrix based on the tool coordinate system and the actual machining coordinate system of the chamfering grinding tool. After obtaining a unique set of quaternions, the axis of the tool coordinate system of the chamfering grinding tool is perpendicular to the normal vector of the fillet surface of the theoretical model under the machining coordinate system, thus obtaining the machining path.

2. The method for generating a rail chamfering processing path according to claim 1, characterized in that, Step S1 includes the following sub-steps: S101. Perform tool coordinate system calibration on the robot's fillet tool to obtain the tool coordinate system. The tool coordinate system of the robot deburring tool is calibrated to obtain the tool coordinate system. The tool coordinate system is calibrated using the 3D laser scanner at the robot's end effector to obtain the tool coordinate system. ; S102. The robot drives a 3D laser scanner to perform a 3D scan of the intersecting surfaces of the long rail to be processed, and obtains the point cloud data of the long rail to be processed.

3. The method for generating a rail chamfering processing path according to claim 1, characterized in that, Step S2 includes: [The text abruptly ends here, likely due to an incomplete sentence or a formatting error.] , By fitting feature surfaces, the normal vector of the point cloud at the intersection of rounded corners is obtained. , Given a fillet radius of r', the coordinates of the center are... ; The curve of the center path of the fitted fillet surface is the path of the machining tool.

4. The method for generating a rail chamfering processing path according to claim 3, characterized in that, Step S4 includes the following sub-steps: S401. Calculate the maximum and minimum values ​​of the point cloud in each direction. ; S402. Divide the point cloud data into cubes. The amount of point cloud data in each small cube is represented as follows: ， Among them, It is the size of the cube. It is the total number of points in the point cloud. This is the cube occupancy rate, representing the ratio of the number of cubes with points to the total number of cubes. It is the expected number of points in the cube; S403. Fitting the planar features yields the following plane equation: ， The coefficients of the plane equation are as follows: , in , For point set subscripts; , Represents a point; S404, the distances between the midpoints of each cube are as follows: ， Where j is the index of the point set, and r is the number of points; S405. Based on the plane equation, use a known open-source library to fit the data and obtain the processing feature surface.

5. The method for generating a rail chamfering processing path according to claim 4, characterized in that, Step S5 also includes: S506. Point cloud registration based on process features, that is, obtaining the correspondence between the theoretical model and the long rail to be processed, and obtaining a set of spatial rotation and translation transformation matrices: ， ， Parameter explanation: Where R is the spatial rotation transformation matrix, They are respectively Rotation transformation matrices in three directions, They are respectively The angular changes in three directions, where T is the spatial translation transformation matrix. These represent the changes in movement in three directions; The spatial transformation relationship between the two coordinate systems is as follows: ， After registration according to the process characteristics, the coordinates of the theoretical model are transformed into the actual machining coordinate system.

6. The method for generating a rail chamfering processing path according to claim 5, characterized in that, Step S6 includes the following sub-steps: S601. Based on the pose relationship between the tool coordinate system and the machining coordinate system of the fillet tool, the corresponding rotation matrix is ​​obtained as follows: ， S602. Obtain the unique set of quaternions corresponding to the rotation matrix using the following formula: ， In the formula, symbols and Same number, symbols and Same number, symbols and Same number.

7. The method for generating a rail chamfering processing path according to claim 1, characterized in that, The simplified theoretical model point cloud data is used to extract point cloud feature surfaces, including planar and cylindrical surfaces, based on the processing technology.

8. A chamfering and grinding system based on a rail chamfering path generation method, characterized in that, include: Robot (1), and gantry milling machine (2) and tool changer (3) connected to robot (1) via Ethernet; robot end effector (4) is installed at the end of robot (1), the robot end effector (4) includes a radial floating high-speed electric spindle (401) and a three-dimensional laser scanner (404), the high-speed electric spindle (401) is equipped with an automatic tool changer mechanism, As the robot (1) moves along the gantry milling machine (2), it drives the three-dimensional laser scanner (404) to complete the three-dimensional scanning of the long rail (5). The robot (1) drives the automatic tool changer to automatically pick up and put in the chamfering tool (403), grinding head (406), and calibration device (405) from the tool magazine (3). The path is generated according to the method described in claim 1 to complete the chamfering and grinding of the rail.

9. A chamfering and grinding system based on a rail chamfering path generation method according to claim 8, characterized in that, It also includes a chip removal pneumatic nozzle (402) for cleaning up metal chips generated during machining, rounding, and deburring processes, as well as chamfering and grinding.

Images