Method and system for detecting missing welds of railway wagon underframe accessories

Abstract

A method and system for detecting missing or incomplete welds on railway freight car underframe accessories are disclosed, solving the problem that manual inspection of underframe accessories is not easy to detect. The method includes: stitching together point clouds of the welding area and adjacent features of the underframe accessories obtained through multi-pose photography to obtain a point cloud model; establishing a theoretical coordinate system C1 based on the theoretical design model of the welding area and adjacent features of the accessories, and establishing a point cloud coordinate system C2 based on the point cloud model; matching C1 and C2 to achieve registration between the point cloud model and the theoretical design model; extracting feature contour lines of the area to be welded on the theoretical design model and storing them as reference units; using a feature extraction algorithm to extract the corresponding feature contour lines projected onto the welding surface point cloud from the registered point cloud model and storing them as comparison units; performing registration iterations between the reference unit and the comparison unit, and determining whether there are missing or incomplete welds based on the overlap of the edge contour lines.

Description

Technical Field

[0001] This invention belongs to the field of measurement of large and complex surface objects, specifically relating to a method and system for detecting missing or incomplete welds on railway freight car underframe accessories. Background Technology

[0002] The manufacturing process of railway freight cars is extremely complex and involves numerous details. Among these, whether the manufacturing of the underframe accessories meets the standards is a crucial factor affecting the safety of train operation. Manual inspection of underframe accessories is not easy to detect quality problems such as missing or incomplete welds, and it is also inefficient and costly. Summary of the Invention

[0003] This invention provides a method and system for detecting missing or incomplete welds on railway freight car underframe accessories, solving the problems of low efficiency and difficulty in detecting missing or incomplete welds during manual inspection of underframe accessories.

[0004] Firstly, a method for detecting missing or incomplete welds on railway freight car underframe accessories is provided, comprising: stitching together point clouds of the welding area and adjacent features of the underframe accessories obtained through multi-pose photography to obtain a point cloud model; establishing a theoretical coordinate system C1 based on the theoretical design model of the welding area and adjacent features of the accessories, and establishing a point cloud coordinate system C2 based on the point cloud model; matching the theoretical coordinate system C1 and the point cloud coordinate system C2 to achieve registration between the point cloud model and the theoretical design model; extracting feature contour lines of the area to be welded on the theoretical design model and storing them as reference units; using a feature extraction algorithm to extract the feature contour lines corresponding to the welding area projected onto the welding surface point cloud on the point cloud model after registration between the point cloud model and the theoretical design model, and storing them as comparison units; performing registration iteration between the reference unit and the comparison unit, and determining whether there are missing or incomplete welds based on the overlap of their edge contour lines.

[0005] In some examples, if the outlines coincide, there is a missing or incomplete weld; if the outlines do not coincide, the weld is normal.

[0006] In some examples, a global constraint method for multi-viewpoint stitching is established based on the principle of point cloud stitching and coordinate transformation. The accumulated error generated during continuous stitching is reduced or eliminated based on the stitching feature points. Finally, the stitched point cloud model is output and displayed.

[0007] In some examples, a theoretical coordinate system C1 is established based on three types of features: points, lines, and surfaces, in the welding area and adjacent feature locations of the theoretical design model. A point cloud feature fitting algorithm is used to fit the corresponding points, lines, and surfaces of the theoretical coordinate system C1 in the point cloud model and establish corresponding features. A point cloud coordinate system C2 is then established based on the established features.

[0008] In some examples, the collected point cloud is first denoised and filtered to eliminate the influence of impurity points; a coarse model matching is performed by combining the theoretical coordinate system C1 and the point cloud coordinate system C2; based on the coarse matching, a registration algorithm based on the principle of minimizing variance is used to construct an objective function based on the sum of the squares of the distance between the measured points and their mean, and the motion spinor of each iteration is linearly solved to achieve a fine matching between the point cloud model and the theoretical design model.

[0009] Secondly, a system for detecting missing or incomplete welds on railway freight car underframe accessories is provided, comprising: a three-dimensional structured light vision sensor configured to capture images of the welding area and adjacent features of the underframe accessories; a computer including: a processor; and a memory including one or more program modules; wherein the one or more program modules are stored in the memory and configured to be executed by the processor, and the one or more program modules include instructions for implementing the method for detecting missing or incomplete welds on railway freight car underframe accessories. Attached Figure Description

[0010] Figure 1 This is a schematic diagram of a railway freight car underframe accessory detection system based on binocular structured light from a single perspective.

[0011] Figure 2 This is a schematic diagram of a railway freight car underframe accessory detection system based on binocular structured light from another perspective.

[0012] Figure 3 This is a schematic flowchart of a method for detecting missing or incomplete welds on railway freight car underframe accessories according to an embodiment of the present invention.

[0013] Figure 4 This is a high-precision point cloud-model registration effect diagram according to an embodiment of the present invention. Detailed Implementation

[0014] Figure 1 , Figure 2 A system for detecting missing or incomplete welds on railway freight car underframe accessories is shown. The system includes an industrial robot 10, a three-dimensional structured light vision sensor 11, and a computer 12.

[0015] The 3D structured light vision sensor 11 uses a binocular structured light camera, such as the PowerScan 2.3M from Visionox. The 3D structured light vision sensor 11 is mounted on the robotic arm 14 of the robot 10. The seventh axis 15 of the robot 10 is positioned below the component under test 13. In this invention, the component under test 13 is an accessory of a truck chassis. The industrial robot 10 and the 3D structured light vision sensor 11 are respectively connected to the computer 12. The industrial robot 10 and the computer 12 communicate using the IP protocol.

[0016] Figure 3A flowchart illustrating a method for detecting missing or incomplete welds on railway freight car underframe accessories is provided. The following describes the five steps involved. Figure 3 The method shown will be explained in detail.

[0017] Step 1: The computer issues a seventh axis movement command to control the industrial robot to move to the point cloud acquisition area of ​​the truck chassis accessories, so that the three-dimensional structured light vision sensor can take multi-pose pictures of the welding area and adjacent features.

[0018] In the process of binocular structured light point cloud acquisition, since a single point cloud data cannot fully describe the information carried by the welding area and adjacent feature areas, a global constraint method for multi-viewpoint stitching is established based on the principle of point cloud stitching and coordinate transformation. Based on the stitching feature points, the cumulative error generated during continuous stitching is reduced or eliminated. Finally, the point cloud stitching result is transmitted to the host computer for output display.

[0019] Step 2: Import the theoretical model into the point cloud registration system. Establish a model coordinate system C1 at the welding area and adjacent feature positions of the theoretical design model. Combine the method of extracting actual workpiece features by point cloud fitting to establish a coordinate system C2 based on the actual point cloud model features in the point cloud region corresponding to the theoretical model.

[0020] A theoretical coordinate system C1 is established at adjacent feature locations in the welding area based on three types of features: points, lines, and surfaces. Based on the actual point cloud model spliced ​​in step 1, a point cloud feature fitting algorithm is used to fit the corresponding points, lines, and surfaces in the theoretical coordinate system C1 in the point cloud model and establish corresponding features. Based on the established features, an actual point cloud model coordinate system C2 is established.

[0021] Step 3: In the registration process between the actual truck chassis accessories and the corresponding truck chassis accessories point cloud data, a combination of coarse matching and fine matching is adopted to achieve high-precision point cloud-model registration.

[0022] The complete point cloud data of the truck chassis accessories acquired by binocular structured light vision is first subjected to point cloud noise reduction and filtering to eliminate the influence of impurity points. Then, a coarse model matching is performed by combining the established coordinate system of the actual workpiece with the coordinate system of the theoretical design model.

[0023] Building upon the aforementioned coarse matching, a high-precision point cloud-model registration algorithm based on the principle of variance minimization is developed. This algorithm constructs an objective function using the sum of squared distances between measured points and their mean values, and linearly solves for the motion spinor in each iteration to achieve point cloud and model registration. Figure 4 A high-precision point cloud-model registration result is shown.

[0024] Step 4: First, extract the feature contour lines of the area to be welded on the theoretical design model and save them as a reference unit; second, use the feature extraction algorithm to extract the feature contour lines of the welding area projected onto the welding surface point cloud on the point cloud model after point cloud-model registration, and save them as a comparison unit.

[0025] Step 5: Register and iterate the reference unit and the comparison unit. If the edge contours of the reference unit and the comparison unit in the extracted welding area overlap, it indicates missing or incomplete welding. If they do not overlap, the welding is normal. Display the detection results on the host computer. Specifically, place the comparison unit and the reference unit in the same coordinate system and perform contour size comparison analysis. If the contours are basically overlapped and the deviation value is small, it proves that there is missing or incomplete welding in this area. If the contours do not overlap at all and the deviation values ​​differ greatly, it proves that the welding in this area is normal.

[0026] The computer in the detection system includes a processor and a memory. The memory stores non-transitory instructions (e.g., one or more program modules). The processor executes these non-transitory instructions, which, when run by the processor, can perform one or more steps of the aforementioned method for detecting missing or incomplete welds in railway freight car underframe accessories. The memory and processor can be interconnected via a bus system and / or other forms of connection. For example, the processor can be a central processing unit (CPU), a graphics processing unit (GPU), or other processing units with data processing capabilities and / or program execution capabilities. For example, the CPU can be an x86 or ARM architecture. The processor can be a general-purpose processor or a special-purpose processor, capable of controlling other components in the computer to perform desired functions. For example, the memory can be volatile memory and / or non-volatile memory. Volatile memory can include, for example, random access memory (RAM) and / or cache memory. Non-volatile memory can include, for example, read-only memory (ROM), hard disk, erasable programmable read-only memory (EPROM), compact optical disc read-only memory (CD-ROM), USB storage, flash memory, etc. One or more program modules can be stored in the memory, and the processor can run one or more program modules to perform various functions of the computer.

Claims

1. A method for detecting missing or incomplete welds on railway freight car underframe accessories, characterized in that, include: A point cloud model is obtained by stitching together the point clouds of the welding area and adjacent features of the chassis accessories obtained by multi-pose photography. A theoretical coordinate system C1 is established based on the theoretical design model of the welding area and adjacent features of the accessory, and a point cloud coordinate system C2 is established based on the point cloud model. The theoretical coordinate system C1 and the point cloud coordinate system C2 are matched to realize the registration of the point cloud model and the theoretical design model. In the theoretical design model, the feature contour lines of the area to be welded are extracted and stored as reference units; After the point cloud model is registered with the theoretical design model, the feature extraction algorithm is used to extract the feature contour lines corresponding to the welding area projected onto the point cloud of the welding surface, and these contour lines are stored as comparison units. The reference unit and the comparison unit are registered and iterated. The overlap of their edge contours is used to determine whether there are missing or missing welds: if the contours overlap, there are missing or missing welds; if the contours do not overlap, the welding is normal.

2. The method for detecting missing or incomplete welds in railway freight car underframe accessories according to claim 1, characterized in that, A global constraint method for multi-viewpoint stitching is established based on the principles of point cloud stitching and coordinate transformation. The cumulative error generated during continuous stitching is reduced or eliminated by stitching feature points. Finally, the stitched point cloud model is output and displayed.

3. The method for detecting missing or incomplete welds in railway freight car underframe accessories according to claim 1, characterized in that, A theoretical coordinate system C1 is established based on three types of features: points, lines, and surfaces in the welding area and adjacent feature locations of the theoretical design model. A point cloud feature fitting algorithm is used to fit the corresponding points, lines, and surfaces in the point cloud model and establish corresponding features. A point cloud coordinate system C2 is then established based on the established features.

4. The method for detecting missing or incomplete welds in railway freight car underframe accessories according to claim 3, characterized in that, The collected point cloud is first subjected to noise reduction and filtering to eliminate the influence of impurity points; then, a coarse model matching is performed by combining the theoretical coordinate system C1 and the point cloud coordinate system C2; based on the coarse matching, a registration algorithm based on the principle of minimizing variance is used to construct an objective function based on the sum of the squares of the distance between the measurement points and their mean, and the motion screw of each iteration is linearly solved to achieve a fine matching between the point cloud model and the theoretical design model.

5. A system for detecting missing or incomplete welds on railway freight car underframe accessories, characterized in that, include: A three-dimensional structured light vision sensor is configured to capture images of the welding area and adjacent features of the chassis attachments. A computer includes: a processor; a memory including one or more program modules; wherein the one or more program modules are stored in the memory and configured to be executed by the processor, and the one or more program modules include instructions for implementing the method for detecting missing or incomplete welds in railway freight car underframe accessories according to any one of claims 1-4.

Images