Intelligent grinding method for steel rail welding seam

Abstract

The invention relates to an intelligent steel rail welding seam grinding method which comprises the following steps: S1, conveying a steel rail: conveying the welded steel rail to a grinding station, moving a welding seam into a grinding range, and locking the steel rail by adopting a steel rail positioning device; s2, setting a steel rail type; s3, the robots on the two sides of the steel rail adopt 3D cameras to conduct visual recognition on welding seams in the grinding area, the type of the steel rail is intelligently recognized and compared with the set type of the steel rail, and if the types are inconsistent, prompt information is given out, and grinding is stopped; if yes, the welding seam position, the steel rail deviation angle and the welding seam thickness are recognized, and the grinding speed and the grinding path are calculated; s4, the robots on the two sides of the steel rail drive grinding devices connected with the robots correspondingly to grind the steel rail welding seam according to the grinding path and the grinding speed; and S5, after polishing is completed, the robot returns to the initial position, and next polishing operation is prepared.

Description

Technical Field

[0001] The present invention relates to the technical field of rail grinding, and in particular to an intelligent grinding method for rail welds. Background Art

[0002] After rail welding, rough grinding is required to remove protruding weld beads and weld reinforcements, ensuring that the weld's external dimensions meet the requirements of subsequent process steps. This rough grinding process requires grinding the weld beads and weld reinforcements on the rail head, rail web, and rail foot, meeting the requirements of the work instructions. Due to the irregular cross-section of the rail, the grinding process is extremely demanding. Simple contour-based automatic grinding equipment cannot identify and determine the grinding position, let alone accurately control the grinding position and grinding thickness. Over- and under-grinding often occurs, and the grinding quality fails to meet the production standards of the welded rail base. Traditional grinding wheel grinding methods are limited by tool structure, requiring frequent tool changes to achieve the grinding standard. Furthermore, grinding wheels cannot cover complex curved surfaces such as rail jaws, resulting in blind spots. The control accuracy of the overlapping working areas between robotic arms is insufficient, which can easily lead to over- and under-grinding. Therefore, the use of belt grinding is gaining popularity. However, the design of a reasonable grinding path for belt grinding, especially the combination of two belts to achieve high-precision grinding, is an urgent issue. Summary of the Invention

[0003] The present invention provides a rail weld intelligent grinding method, comprising the following steps: S1: Rail conveying: The welded rail is conveyed to the grinding station and the weld is moved into the grinding range. The rail is locked using the rail positioning device. S2: Set rail type; S3: The robots on both sides of the rail use 3D cameras to visually identify the welds in the grinding area, intelligently identify the rail type, and compare it with the set rail type. If they are inconsistent, a prompt message is given and grinding stops. If they are consistent, the robots identify the weld position, rail offset angle, and weld thickness, and calculate the grinding speed and grinding path. S4: The robots on both sides of the rail drive their connected grinding devices to grind the rail welds according to the grinding path and grinding speed; S5: After polishing is completed, the robot returns to its initial position and prepares for the next polishing operation; Among them, in the above step S3, the method for intelligently identifying the rail type includes: establishing a rail database model for various rail models, extracting point clouds from pictures taken by a 3D camera, and matching them with the rail database model, thereby intelligently identifying the rail type.

[0004] Furthermore, in the above step S3, the polishing path planning method is: The outer surface of the rail is divided into several areas: rail bottom area AB, left rail foot bottom area AJ, right rail foot bottom area BC, left rail foot upper surface and triangular area JI, right rail foot upper surface and triangular area CD, left rail waist and rail jaw area IH, right rail waist and rail jaw area DE, left rail top side area GH, right rail top side area EF, and rail top area FG; The left robot grinds the rail top area FG and the left rail top side area GH, while the right robot is on standby; The right robot grinds the rail bottom area AB, while the left robot is on standby; The left robot polishes the left rail foot area AJ, and the right robot polishes the right rail foot area BC; The right robot polishes the upper surface of the right rail foot and the triangular area CD, while the left robot is on standby; The right robot grinds the right rail waist and rail jaw area DE and the right rail top side area EF, and the left robot grinds the left rail foot upper surface and triangular area JI. The left robot grinds the left rail waist and rail jaw area IH, while the right robot is on standby.

[0005] Furthermore, the grinding points of the left and right rail waists are determined by the following method: the center position O of the rail is selected as the base point, where the coordinate of O is, starting from the base point, multiple reference points Oi are selected upward and downward along the surface of the rail, i=1,2…n, where n is determined according to the type of rail, the coordinate of Oi is, the center position of the rail is obtained by visual recognition as Oc, the coordinate of Oc is, the offset dx=OcX-OX, dy=OcY-OY, dz=OcZ-OZ between the center position Oc of the rail obtained by visual recognition and the center position O of the rail is calculated, according to the above offset, the coordinates of multiple reference points Oi are adjusted in turn, the coordinate of the grinding point Oci is, the above grinding points Oci and Oc constitute the grinding path.

[0006] Furthermore, after the above step S4, the grinding quality is inspected. If the inspection fails, re-grinding or manual re-grinding is performed; after the inspection passes, the rail positioning device releases the rail and the rail is moved out of the grinding station; The method for inspecting the grinding quality includes: the robots on both sides use 3D cameras to scan the weld area again to detect the residual thickness of the weld nodule. If it exceeds the set value, it will be judged as unqualified.

[0007] The present invention also provides an intelligent rail grinding device, including a rail positioning device, a left robot, and a right robot. The front ends of the left robot and the right robot are equipped with a sand belt grinding device and a 3D camera, and the above-mentioned grinding method is adopted.

[0008] Furthermore, the intelligent rail grinding equipment also includes a recipe system, which stores the grinding force and the speed of the sanding belt machine corresponding to different rail types. In step S3, after the rail type is determined to be consistent, the corresponding grinding force and the speed of the sanding belt machine are called to control the operation of the sanding belt grinding device.

[0009] Furthermore, a plurality of rail positioning devices are arranged along the extension direction of the rail, and are used to lock and position the rail to prevent the rail from shifting during the grinding process.

[0010] Furthermore, the 3D camera is a laser 3D industrial camera, which is also used to take photos of the weld positions after the rails are locked and before they are released after grinding.

[0011] The rail grinding technology of the present invention uses 3D visual recognition technology to intelligently identify rail type, weld position, rail offset angle, and weld thickness. It automatically plans the grinding path according to the set grinding process, realizes automatic intelligent grinding of rail welds, and improves the grinding operation efficiency and quality. By setting a reasonable grinding path: (1) While avoiding interference between the two robotic arms, the operation is carried out synchronously in the non-interference area, thus shortening the operation time as much as possible. (2) It ensures that the grinding tool and the rail surface maintain good contact, thereby ensuring the grinding effect. (3) The grinding path is allowed to operate continuously as much as possible to ensure the grinding effect and avoid the occurrence of misalignment between different paths. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is a schematic diagram of the rail grinding path; Figure 2 This is a schematic diagram of the coordinate point distribution of the rail waist grinding path; Figure 3 This is a schematic diagram of the structure of the intelligent rail grinding equipment; Figure 4 It is a schematic diagram of the belt grinding device at work. DETAILED DESCRIPTION

[0013] The rail weld intelligent grinding method of the present invention mainly includes the following steps: S1: Rail conveying: The welded rail is conveyed to the grinding station and the weld is moved into the grinding range. The rail is locked using the rail positioning device. S2: Set rail type and grinding intensity; S3: The robots on both sides of the rail use 3D cameras to visually identify the welds in the grinding area, intelligently identify the rail type, and compare it with the set rail type. If they are inconsistent, a prompt message is given and grinding stops. If they are consistent, the robots identify the weld position, rail offset angle, and weld thickness, and calculate the grinding speed and grinding path. S4: The grinding devices connected to the robots on both sides of the rail grind the rail weld according to the grinding path and grinding speed; S5: After grinding is completed, the robot returns to its initial position and prepares for the next grinding operation.

[0014] In step S2, the rail model is set manually. In step S3, the method for intelligently identifying rail types includes: establishing a rail database model for various rail models, extracting point clouds from images taken by a 3D camera, and matching them with the rail database model to intelligently identify rail types.

[0015] The grinding force and material determine the specific amount of grinding, which in turn determines the grinding speed. The grinding material and force are determined based on on-site experience. Visual recognition is primarily used to identify processes such as weld position, rail offset angle, and weld height, which require relatively complex manual measurement. It can also identify rail type. This dual matching prevents human input errors while combining manual experience with visual recognition to quickly determine all key process parameters required for subsequent operations.

[0016] In the above step S3, the planning method of the grinding path is: See also Figure 1 First, the outer surface of the rail is divided into several areas: rail bottom area AB, left rail foot bottom area AJ, right rail foot bottom area BC, left rail foot upper surface and triangular area JI, right rail foot upper surface and triangular area CD, left rail waist and rail jaw area IH, right rail waist and rail jaw area DE, left rail top side area GH, right rail top side area EF, and rail top area FG.

[0017] Because the two robots' arms have interference areas, some areas cannot be operated synchronously and need to be avoided. Therefore, some paths can be operated synchronously, while some paths need to be operated separately. In the present invention, the specific grinding order is as follows: The left robot grinds the rail top area FG and the left rail top side area GH, while the right robot is on standby; The right robot grinds the rail bottom area AB, while the left robot is on standby; The left robot polishes the left rail foot area AJ, and the right robot polishes the right rail foot area BC; The right robot polishes the upper surface of the right rail foot and the triangular area CD, while the left robot is on standby; The right robot grinds the right rail waist and rail jaw area DE and the right rail top side area EF, and the left robot grinds the left rail foot upper surface and triangular area JI. The left robot grinds the left rail waist and rail jaw area IH, while the right robot is on standby.

[0018] The above path design ensures effective grinding, avoids misalignment, and maximizes efficiency. The rail top and bottom are the primary contact surfaces, so the optimal approach is to process the rails in a non-segmented manner. Specifically, the rail top area FG and the rail bottom area AB should not be further divided. Otherwise, misalignment may occur at the intersection of the segments, affecting the grinding effect. Therefore, one robotic arm is responsible for the rail top and the other for the rail bottom. The rail bottom corners, rail waist, and rail jaws are each handled by the corresponding robotic arm.

[0019] Due to the robot arm's posture and the grinding tool's own gravity, convex and horizontal surfaces (FG, AB) are moved from far to near (pull back) to ensure good contact between the grinding tool and the rail surface. For concave surfaces (CDE, HIJ), the movement is from near to far (push forward).

[0020] For the GH and EF areas, the two areas are perpendicular to the ground, so the principle of fewer segments is maintained, GH and FG operate continuously, and EF and DE operate continuously.

[0021] Further, for rail waist grinding, see Figure 2 Taking the right rail waist as an example, the method for determining the coordinate points on the path is as follows: select the center position O of the rail as the base point, where the coordinates of O are (OX, OY, OZ). Starting from the base point, select multiple reference points Oi along the surface of the rail in sequence upward and downward, i=1,2…n, where n is determined according to the type of rail. The coordinates of Oi are (Oix, OiY, OiZ). The center position of the rail is obtained by visual recognition as Oc, and the coordinates of Oc are (OcX, OcY, OcZ). The offset dx=OcX-OX, dy=OcY-OY, dz=OcZ-OZ between the center position Oc of the rail obtained by visual recognition and the center position O of the rail is calculated. According to the above offsets, the coordinates of multiple reference points Oi are adjusted in sequence. The coordinates of the grinding point Oci are (OiX+dx, OiY+dy, OiZ+dz). The above grinding points Oci and Oc constitute the grinding path.

[0022] O and Oi are preset positions, and Oc is a position obtained by visual recognition.

[0023] Grinding speed=grinding amount / weld thickness, wherein the grinding amount is determined in step S2.

[0024] Furthermore, after the above step S4, the grinding quality is inspected. If the inspection fails, re-grinding or manual re-grinding is performed. After the inspection passes, the rail positioning device releases the rail and the rail moves out of the grinding station; The method for inspecting the grinding quality includes: the robots on both sides use 3D cameras to scan the weld area again to detect the residual thickness of the weld nodule. If it exceeds the set value (0.3mm), it will be judged as unqualified.

[0025] See also Figure 3-Figure 4 , is a schematic diagram of the structure of the intelligent rail grinding equipment of the present invention, which includes a rail positioning device 1, a left robot 2, and a right robot 3. The front ends of the left robot 2 and the right robot 3 are equipped with a belt grinding device 4 and a 3D camera 5. The rail positioning device 1 is arranged in multiples along the extension direction of the rail, and is used to lock and position the rail to prevent the rail from shifting during the grinding process. Its functions include hydraulic clamping and hydraulic pressing of the rail. It is generally required that each time the rail is transmitted and locked, the position reached by the weld is guaranteed to be within the range of ±20cm. The rail positioning device should first perform the hydraulic clamping action, and then perform the hydraulic pressing action, so as to achieve the locking and fixation of the rail and prevent the rail from shifting during the grinding process.

[0026] In the present invention, based on the characteristics of the dual robots, the two belt grinding devices 4 cooperate with each other, combined with 3D visual recognition technology, and realize high-precision grinding of the top, waist and bottom of the rail without dead angles through dynamic path planning.

[0027] Furthermore, the 3D camera is a laser 3D industrial camera, which can also be used to take photos of the weld positions after the rails are locked and before they are released after grinding.

[0028] Furthermore, the intelligent grinding equipment also includes a recipe system, which stores the grinding force and the speed of the sanding belt machine corresponding to different rail types. In step S3, after the rail type is determined to be consistent, the corresponding grinding force and the speed of the sanding belt machine are called to control the operation of the sanding belt grinding device.

[0029] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A rail weld intelligent grinding method, characterized in that: The following steps are involved: S1: Rail conveying: The welded rail is conveyed to the grinding station and the weld is moved into the grinding range. The rail is locked using the rail positioning device. S2: Set rail type; S3: The robots on both sides of the rail use 3D cameras to visually identify the welds in the grinding area, intelligently identify the rail type, and compare it with the set rail type. If they are inconsistent, a prompt message is given and grinding stops. If they are consistent, the robots identify the weld position, rail offset angle, and weld thickness, and calculate the grinding speed and grinding path. S4: The robots on both sides of the rail drive their connected grinding devices to grind the rail welds according to the grinding path and grinding speed; S5: After polishing is completed, the robot returns to its initial position and prepares for the next polishing operation; Among them, in the above step S3, the method for intelligently identifying the rail type includes: establishing a rail database model for various rail models, extracting point clouds from pictures taken by a 3D camera, and matching them with the rail database model, thereby intelligently identifying the rail type.

2. The rail weld intelligent grinding method according to claim 1, characterized in that: In the above step S3, the planning method of the grinding path is: The outer surface of the rail is divided into several areas: rail bottom area AB, left rail foot bottom area AJ, right rail foot bottom area BC, left rail foot upper surface and triangular area JI, right rail foot upper surface and triangular area CD, left rail waist and rail jaw area IH, right rail waist and rail jaw area DE, left rail top side area GH, right rail top side area EF, and rail top area FG; The left robot grinds the rail top area FG and the left rail top side area GH, while the right robot is on standby; The right robot grinds the rail bottom area AB, while the left robot is on standby; The left robot polishes the left rail foot area AJ, and the right robot polishes the right rail foot area BC; The right robot polishes the upper surface of the right rail foot and the triangular area CD, while the left robot is on standby; The right robot grinds the right rail waist and rail jaw area DE and the right rail top side area EF, and the left robot grinds the left rail foot upper surface and triangular area JI. The left robot grinds the left rail waist and rail jaw area IH, while the right robot is on standby.

3. The rail weld intelligent grinding method according to claim 2, characterized in that: The grinding points of the left and right rail waists are determined by the following method: The center position O of the rail is selected as the base point, where the coordinates of O are (OX, OY, OZ). Starting from the base point, multiple reference points Oi are selected upward and downward along the surface of the rail, where i=1,2…n, where n is determined according to the type of rail. The coordinates of Oi are (Oix, OiY, OiZ). The center position of the rail is obtained by visual recognition as Oc, and the coordinates of Oc are (OcX, OcY, OcZ). The offset dx=OcX-OX, dy=OcY-OY, dz=OcZ-OZ between the center position Oc of the rail obtained by visual recognition and the center position O of the rail is calculated. According to the above offset, the coordinates of multiple reference points Oi are adjusted in turn. The coordinates of the grinding point Oci are (OiX+dx, OiY+dy, OiZ+dz). The above grinding points Oci and Oc constitute the grinding path.

4. The rail weld intelligent grinding method according to claim 1, characterized in that: After the above step S4, the grinding quality is inspected. If the inspection fails, the rail is re-grinded or manually re-grinded. After the inspection passes, the rail positioning device releases the rail and the rail is moved out of the grinding station. The method for inspecting the grinding quality includes: the robots on both sides use 3D cameras to scan the weld area again to detect the residual thickness of the weld nodule. If it exceeds the set value, it will be judged as unqualified.

5. An intelligent rail grinding device, comprising a rail positioning device (1), a left robot (2), and a right robot (3), wherein the front ends of the left robot (2) and the right robot (3) are equipped with a belt grinding device (4) and a 3D camera (5), characterized in that: The polishing method according to any one of claims 1 to 4 is adopted.

6. The intelligent rail grinding equipment according to claim 5, characterized in that: The intelligent grinding device further includes a recipe system, which stores grinding forces and sanding machine rotation speeds corresponding to different rail types. In step S3, after the rail type is determined to be consistent, the corresponding grinding force and sanding machine rotation speed are retrieved to control the operation of the sanding belt grinding device (4).

7. The intelligent rail grinding equipment according to claim 5, characterized in that: The rail positioning devices are arranged in plurality along the extending direction of the rail and are used to lock and position the rail to prevent the rail from shifting during the grinding process.

8. The intelligent rail grinding equipment according to claim 7, characterized in that: The 3D camera is a laser 3D industrial camera, and is also used to take photos of the weld positions after the rails are locked and before they are released after grinding.

Images