A method and system for splicing steel components using three-dimensional recognition technology

Abstract

The present invention discloses a method and a system for splicing steel members by using three-dimensional recognition technology. The method includes the following steps: S1. Scanning the steel member model by a three-dimensional scanning device to collect the point cloud data of the steel member model; S2. Calculating its dimension data and recording it in a database; S3. Scanning each splicing component of the steel member to obtain the point cloud data of each splicing component of the steel member; S4. Calculating its dimension data according to the point cloud data of each splicing component of the steel member; S5. Retrieving the dimension data of the steel member model in the database and comparing it with each splicing component of the steel member to check whether there is a dimension deviation; S6. Determining that the splicing component of the steel member is an unqualified splicing component; S7. Continuing to verify whether there is a dimension deviation in other splicing components of the entire steel member; S9. Completing the splicing of each splicing component of the steel member. By pre-storing the point cloud data of the steel member model, the present invention realizes the verification of the dimensions and production qualification of the steel member before splicing the components to be spliced of the steel member.

Description

Technical Field

[0001] The present invention belongs to the field of steel structure component manufacturing. Specifically, it relates to a method and system for splicing steel components using three-dimensional recognition technology. Background Art

[0002] Due to the convenience of construction and good controllability of the building style after completion, steel structures are widely used. Steel structure construction is carried out with steel structure components as the basic unit. Generally, the production process of steel structure component manufacturing, that is, the flow operation production process, is carried out according to the following steps: steel plate cutting, assembly spot welding, welding, and straightening to obtain finished steel components.

[0003] Three-dimensional laser scanners have basically realized the deformation monitoring function in the field of civil engineering, making full use of the advantages of three-dimensional laser scanners. However, at the same time, the use of high-speed and large-area scanning methods has generated a large amount of point cloud data. In terms of data processing, most use reverse engineering software for point cloud data processing, which has basically realized the point cloud data processing function and has a certain degree of generality, but lacks pertinence and is not applicable to the data processing of all scanned components. In terms of visualization, a simple point cloud visualization function can be realized, and at the same time, colors can be assigned to the point cloud data without color features. However, each view window can only display one point cloud processing result. If you want to compare the data before and after processing, not only do you need to call the window display separately, but the display angles adjusted manually can only be approximately the same, and the discrimination effect is poor; in terms of point cloud denoising, in the reverse engineering software, point cloud denoising is divided into three parts: non-connected items, outlier points, and noise reduction. When selecting non-connected items, there are two parameter settings: separation and size. Among them, the separation includes three options: low, medium, and high, but it is difficult to determine the demarcation points of low, medium, and high, and there is no unified standard; the size is the proportion of the number of selected points. When setting the parameters, there is no reference object and it is relatively random, and it is impossible to judge whether the setting is excessive and causes feature loss or the setting is too small and the non-connected items are not completely removed; outlier points are found by setting the sensitivity value. The larger this value is, the more outlier points are found. Similar to the size parameter setting of non-connected items, there are certain defects. The noise reduction parameter has three types: free-form surface shape, prismatic conservative, and prismatic active. However, in actual operation, it is generally necessary to manually select and delete the obviously irrelevant points by frame, which reduces the processing efficiency; in terms of point cloud sampling and compression, there are mainly four sampling methods, namely curvature sampling, equidistant sampling, uniform sampling, and random sampling. It is mainly based on the curvature percentage parameter, spacing parameter, remaining point number parameter, and directly sampling the original point cloud data by percentage. This software can realize the compression of point cloud, but most of them directly sample data according to the percentage and point spacing, without recalculating the distribution of point cloud, and it is impossible to set parameters with reference to its specific sampling principle.

[0004] In summary, the defect of the three-dimensional recognition technology in the prior art when used in steel members is that although the point position data can be recognized during the splicing process, there is a lack of verification on whether the splicing of the point position data is in place, resulting in poor splicing effect of high-precision steel members. Summary of the Invention

[0005] Aiming at the problem that the prior three-dimensional recognition technology lacks verification on whether the splicing of point position data is in place when splicing steel members, the present invention provides a method and system for splicing steel members using three-dimensional recognition technology.

[0006] To achieve the above technical objectives, the technical solutions adopted by the present invention are as follows:

[0007] A method for splicing steel members using three-dimensional recognition technology includes the steps of:

[0008] S1. Scanning the steel member model through a three-dimensional scanning device to collect the point cloud data of the steel member model;

[0009] S2. Restoring the steel member model according to the point cloud data of the steel member model at a ratio of 1:1, calculating its dimension data, and recording it in the database;

[0010] S3. Scanning each splicing part of the steel member to obtain the point cloud data of each splicing part of the steel member;

[0011] S4. Calculating the dimension data according to the point cloud data of each splicing part of the steel member;

[0012] S5. Retrieving the dimension data of the steel member model in the database and comparing it with the dimension data of each splicing part of the steel member to check whether there is a dimension deviation. If there is, go to step S6; if not, go to step S7;

[0013] S6. Determining that the splicing part of the steel member is an unqualified splicing part;

[0014] S7. Continuing to verify whether there is a dimension deviation in other splicing parts of the entire steel member. If there is, go to step S8; if not, go to step S9;

[0015] S8. Replacing the unqualified splicing part of the steel member with the splicing part of the same part, and returning to step S5;

[0016] S9. Completing the splicing of each splicing part of the steel member.

[0017] Further, the splicing described in step S9 includes various splicing methods: snap connection, riveting, and welding.

[0018] Further, the point cloud data of the steel member model is consistent with or scaled proportionally to the point cloud data of the actual production standard steel member.

[0019] Further, the dimension data is calculated using the normal vector algorithm of point cloud data. On the premise of judging whether the K-nearest neighbors of the point cloud data are evenly distributed, the point cloud topological relationship constructed by the KD-Tree is used to construct the least squares plane of each point in the point cloud, calculate the normal vector of this plane, and use this normal vector as the calculation result of the normal vector of the data point.

[0020] Further, the point cloud boundary detection and hole detection are also included in step S2;

[0021] Assume that the input point cloud data is S, and the output boundary feature set is F(S). The specific steps of point cloud boundary detection are as follows:

[0022] S21. Extract the data point P from the point cloud data set S in sequence as the current data to be processed;

[0023] S22. Project the K-nearest neighbors of the data point P into the corresponding normal vector plane, and connect the data point P and the projection points of the neighborhood to form a vector set;

[0024] S23. Calculate the angles between all the vectors corresponding to the vector set and the reference vector, sort the calculation results in ascending order to obtain the angle set S', calculate the angle difference of S' and then obtain the angle set A of adjacent vectors in the vector set;

[0025] S24. If the maximum angle difference Amax of the angle set A > u, put the data point P into the set F(S); if all the data in the point cloud data set S have been processed, the algorithm ends, otherwise return to step S21.

[0026] Further, the specific steps of hole detection are as follows:

[0027] S201. Judge whether there are holes on the splicing parts of the steel members;

[0028] S202. If there are holes, perform triangular meshing on the hole boundary points;

[0029] S203. Calculate the hole size by calculating the grid coverage area, and judge whether the hole is a bolt connection hole or a welding boundary hole;

[0030] S204. If it is any of the holes in step S203, it is determined as a normal hole, and the splicing of the splicing parts of the steel members is completed according to the appropriate connection method matching the hole size;

[0031] S205. If it is other holes, it is determined as an abnormal hole, and this splicing part is a non-conforming product.

[0032] A system for splicing steel components using three-dimensional recognition technology, comprising a three-dimensional scanning device, a steel component size calculation unit, a size verification unit for splicing components of steel components, a hole detection unit, and a database;

[0033] The three-dimensional scanning device is used to collect point cloud data of the steel component model;

[0034] The steel component size calculation unit calculates the size of the components to be spliced of the steel component using the point cloud data normal vector algorithm;

[0035] The size verification unit for splicing components of steel components is used to verify whether the size of the components to be spliced of the steel component is the standard size of the steel component model;

[0036] The hole detection unit is used to detect the hole size of the entire component to be spliced of the steel component, determine whether the hole is a normal hole, and at the same time determine whether the hole size of the component to be spliced is a qualified product;

[0037] The database stores the point cloud data and size data of various steel component models in real time.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] By pre-storing the point cloud data and size data of the components to be spliced of the steel component in the point cloud data of the steel component model; it is convenient to verify the steel component size and whether it is qualified in production before splicing the components to be spliced of the steel component. Ensure that the sizes of the splicing components of the steel component are qualified and completely spliced, and the generation of unqualified products; realize the verification before splicing the steel component. Description of the Drawings

[0040] Figure 1 It is a flowchart of a method for splicing steel components using three-dimensional recognition technology of the present invention;

[0041] Figure 2 It is a structural block diagram of a system for splicing steel components using three-dimensional recognition technology of the present invention;

[0042] Figure 3 It is a flowchart of the specific steps of point cloud boundary detection of the present invention;

[0043] Figure 4 It is a flowchart of the specific steps of hole detection of the present invention.

[0044] Explanation of the markings in the figure: 10 - three-dimensional scanning device, 20 - steel component size calculation unit, 30 - size verification unit for splicing components of steel components, 40 - hole detection unit, 50 - database. Detailed Embodiment

[0045] For the convenience of those skilled in the art, the present invention will be further described below in conjunction with embodiments and the accompanying drawings. The content mentioned in the embodiments does not limit the present invention.

[0046] As Figure 1 shown, a method for splicing steel members using three-dimensional recognition technology includes the steps of: S1. Scanning the steel member model through the three-dimensional scanning device 10 to collect the point cloud data of the steel member model; S2. Restoring the steel member model according to the point cloud data of the steel member model at a ratio of 1:1, calculating its dimension data, and recording it in the database 50; S3. Scanning each splicing part of the steel member to obtain the point cloud data of each splicing part of the steel member; S4. Calculating the dimension data according to the point cloud data of each splicing part of the steel member; S5. Retrieving the dimension data of the steel member model in the database 50 and comparing it with each splicing part of the steel member to check whether there is a dimension deviation. If there is, go to step S6; if not, go to step S7; S6. Determining that the splicing part of the steel member is an unqualified splicing part; S7. Continuing to verify whether there is a dimension deviation in other splicing parts of the entire steel member. If there is, go to step S8; if not, go to step S9; S8. Replacing the unqualified splicing part of the steel member with the splicing part of the same part and returning to step S5; S9. Completing the splicing of each splicing part of the steel member.

[0047] The splicing described in step S9 includes various splicing methods: snap connection, riveting, and welding. The holes for snap connection and riveting are internal holes, and the specific splicing method is identified by hole detection; various splicing methods are suitable for the splicing of different splicing parts.

[0048] The point cloud data of the steel member model is consistent with or scaled proportionally to the point cloud data of the actual production standard steel member. The point cloud data with the same proportion ensures that when verifying whether the dimensions or point cloud data of the splicing parts are without deviation, the point cloud data of the steel member model and the splicing parts are exactly the same. The point cloud data with proportional scaling is convenient for verifying steel members with the same structural shape but different sizes, reducing the design of the steel member model. For example, the sizes of large and small screws are different, but the structural shapes are exactly the same.

[0049] The dimension data calculation uses the point cloud data normal vector algorithm. On the premise of judging whether the K-nearest neighbors of the point cloud data are evenly distributed, through the point cloud topological relationship constructed by KD-Tree, the least squares plane of each point in the point cloud is constructed, the normal vector of this plane is calculated, and this normal vector is used as the calculation result of the normal vector of the data point. The selection of the K value is very important. When the curvature changes greatly, the selection of the K value needs to be carefully considered to ensure single convex or single concave, so that the obtained least squares plane can better approximate the original point cloud and also make the projection points on the local surface.

[0050] As Figure 3As shown, the point cloud boundary detection and hole detection are also included in step S2. Assume that the input point cloud data is S, and the output boundary feature set is F(S). The specific steps of point cloud boundary detection are as follows: S21. Extract the data point P from the point cloud data set S in sequence as the current data to be processed; S22. Project the K-nearest neighbor points of the data point P into the corresponding normal vector plane, and connect the data point P and the projected points in the neighborhood to form a vector set; S23. Calculate the angles between all vectors corresponding to the vector set and the reference vector, sort the calculation results in ascending order to obtain the angle set S', calculate the angle difference of S' and then obtain the angle set A of adjacent vectors in the vector set; S24. If the maximum angle difference Amax of the angle set A > u (in general, the threshold u is taken as π / 2), then put the data point P into the set F(S); if all the data in the point cloud data set S have been processed, the algorithm ends, otherwise return to step S21.

[0051] As Figure 4 shown, the specific steps of hole detection are as follows: S201. Judge whether there are holes in the splicing parts of the steel members; S202. If there are holes, triangulate the hole boundary points; S203. Calculate the size of the hole by the grid coverage area, and judge whether the hole is a bolt connection hole or a welding boundary hole; The splicing hole point cloud data and size data of the steel member model are pre-stored in the database 50 to facilitate the verification of whether it is a normal hole for splicing in the subsequent steps. S204. If it is any of the holes in step S203, it is determined as a normal hole, and the appropriate connection method is matched according to the hole size to complete the splicing of the splicing parts of the steel members; S205. If it is other holes, it is determined as an abnormal hole, and the splicing part is a non-conforming product.

[0052] As Figure 2 shown, a system for splicing steel members using three-dimensional recognition technology includes a three-dimensional scanning device 10, a steel member size calculation unit 20, a size verification unit 30 for the splicing parts of the steel members, a hole detection unit 40, and a database 50; The three-dimensional scanning device 10 is used to collect the point cloud data of the steel member model; The steel member size calculation unit 20 calculates the size of the parts to be spliced of the steel member by using the point cloud data normal vector algorithm; The size verification unit 30 for the splicing parts of the steel members is used to verify whether the size of the parts to be spliced of the steel member is the standard size of the steel member model; The hole detection unit 40 is used to detect the size of the holes in the whole parts to be spliced of the steel member, judge whether the holes are normal holes, and at the same time judge whether the size of the holes in the parts to be spliced is a qualified product; The database 50 stores the point cloud data and size data of various steel member models in real time.

[0053] The above has introduced in detail a method and system for splicing steel components using three-dimensional recognition technology. The description of specific embodiments is only used to help understand the method and its core idea of the present application. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present application, several improvements and modifications can still be made to the present application, and these improvements and modifications also fall within the protection scope of the claims of the present application.

Claims

1. A method for splicing steel components using three-dimensional recognition technology, characterized in that, Including the steps: S1. Scanning the steel component model by a 3D scanning device (10) to collect the point cloud data of the steel component model; S2. Restoring the steel component model according to the point cloud data of the steel component model at a ratio of 1:1, calculating its dimension data, and recording it in the database (50); S3. Scanning each splicing part of the steel component to obtain the point cloud data of each splicing part of the steel component; S4. Calculating the dimension data according to the point cloud data of each splicing part of the steel component; S5. Retrieving the dimension data of the steel component model in the database (50) and comparing it with the dimension data of each splicing part of the steel component, and determining whether there is a dimension deviation in each splicing part of the steel component. If there is, go to step S6; if not, go to step S7; S6. Determining that the splicing part of the steel component is a non-conforming splicing part; S7. Continuing to verify whether there is a dimension deviation in other connecting parts of the entire steel component. If there is, go to step S8; if not, go to step S9; S8. Replacing the non-conforming splicing part of the steel component with the splicing part of the same part, and returning to step S5; S9. Completing the splicing of each splicing part of the steel component; The dimension data calculation uses the point cloud data normal vector algorithm. First, it is judged whether the K-nearest neighbor points of the point cloud data are evenly distributed. Through the point cloud topological relationship constructed by the KD-Tree, the least square plane of each point in the point cloud is constructed, the normal vector of this plane is calculated, and this normal vector is used as the calculation result of the normal vector of the data point.

2. The method for splicing steel members using three-dimensional recognition technology according to claim 1, wherein The splicing described in step S9 includes various splicing methods: snap connection, riveting, and welding.

3. A method for splicing steel members using three-dimensional recognition technology according to claim 2, characterized in that, The point cloud data of the steel component model is consistent with or scaled proportionally to the point cloud data of the actual production standard steel component.

4. A method for splicing steel members using three-dimensional recognition technology according to claim 3, characterized in that Step S2 also includes point cloud boundary detection and hole detection; Let the input point cloud data be S, and the output boundary set be F(S). The specific steps of point cloud boundary detection are as follows: S21. Extracting the data point P from the point cloud data set S in sequence as the currently processed data; S22. Projecting the K-nearest neighbor points of the data point P into the corresponding normal vector plane, and connecting the data point P and the projection points of the neighborhood to form a vector set; S23. Calculating the included angles between all the vectors corresponding to the vector set and the reference vector, sorting the calculation results in ascending order to obtain the included angle set S', calculating the included angle difference of S' to obtain the included angle set A of adjacent vectors in the vector set; S24. If the maximum angle difference Amax of the included angle set A > u, where u is equal to π / 2, then put the data point P into the set F(S); if all the data in the point cloud data set S have been processed, the algorithm ends, otherwise return to step S21.

5. A method for splicing steel members using three-dimensional recognition technology according to claim 4, characterized in that, The specific steps of hole detection are as follows: S201. Judging whether there are holes on the splicing part of the steel component; S202. If there are holes, performing triangular meshing on the hole boundary points; S203. Calculating the size of the hole by calculating the grid coverage area, and judging whether the hole is a bolt connection hole or a welding boundary hole; S204. If it is any of the holes in step S203, it is determined as a normal hole, and the splicing of the splicing part of the steel component is completed according to the appropriate connection method matching the hole size; S205. If it is other holes, it is determined as an abnormal hole, and the splicing part is a non-conforming product.

6. A system for splicing steel members that implements the method for splicing steel members using three-dimensional recognition technology according to any one of claims 1-5, characterized in that, It includes a three-dimensional scanning device (10), a steel component dimension calculation unit (20), a splicing component dimension verification unit (30) for steel components, a hole detection unit (40), and a database (50); The three-dimensional scanning device (10) is used to collect the point cloud data of the steel component model; The steel component dimension calculation unit (20) calculates the dimensions of the components to be spliced of the steel component by using the point cloud data normal vector algorithm; The splicing component dimension verification unit (30) for steel components is used to verify whether the dimensions of the components to be spliced of the steel component are the standard dimensions of the steel component model; The hole detection unit (40) is used to detect the hole sizes of the entire components to be spliced of the steel component, judge whether the holes are normal splicing holes, and at the same time judge whether the hole sizes of the components to be spliced are qualified products; The database (50) stores the point cloud data and dimension data of various steel component models in real time.

Images