A method of welding a steel structure

Abstract

The application discloses a steel structure welding method, and relates to the technical field of 3D visual welding, and comprises the following steps: step S1, scanning point cloud data of a steel structure; step S2, based on the scanned data in step S1, a random sample consensus method is used to find a plane; step S3, according to a visual coordinate, a normal line of the plane obtained in step S2 is unified; step S4, a boundary between planes is calculated according to the plane obtained in step S2; step S5, a nearest neighbor algorithm is used to search for the nearest neighbor points of each two groups of boundaries in step S4, and a common boundary is formed; step S6, straight line fitting is performed on the common boundary obtained in step S5, and end points are extracted as welds; and step S7, path planning is performed based on a distance method. The welding method provided by the application automatically solves the plate plate welding problem in a steel structure, uses a 3D visual system to identify welds, automatically plans a welding path, saves labor cost, improves welding precision stability, and shortens a welding cycle.

Description

Technical Field

[0001] This invention relates to the field of 3D vision welding technology, and more specifically to a steel structure welding method. Background Technology

[0002] Steel structures are structures mainly composed of steel materials and are one of the main types of building structures. The structure is mainly composed of steel beams, steel columns, steel frames and other components made of steel sections and steel plates. The components or parts are usually connected by welds, bolts or rivets. Due to its light weight and simple construction, it is widely used in large factories, stadiums, super high-rise buildings and other fields. The large amount of manual intervention significantly increases operating costs, and due to human factors, there are problems such as unstable welding accuracy and long welding cycle.

[0003] Therefore, it is necessary to propose a steel structure welding method to solve the above problems. Summary of the Invention

[0004] The purpose of this invention is to solve the problems of unstable precision and long welding cycle in the automated steel structure welding process.

[0005] To achieve the above objectives, the present invention specifically adopts the following technical solution:

[0006] A steel structure welding method includes the following steps:

[0007] Step S1: Scan the point cloud data of the steel structure;

[0008] Step S2: Based on the data scanned in step S1, the plane is found using the random sampling consistency method;

[0009] Step S3: Unify the normal of the plane obtained in step S2 according to the visual coordinates;

[0010] Step S4: Calculate the inter-plane boundaries based on the planes obtained in step S2;

[0011] Step S5: Use the nearest neighbor algorithm to search for the nearest neighbor points of every two sets of boundaries in step S4, forming a common boundary;

[0012] Step S6: Perform straight line fitting on the common boundary obtained in step S5, and extract the endpoints as welds;

[0013] Step S7: Perform path planning based on distance methods.

[0014] Furthermore, the specific steps for finding the plane in step S2 are as follows:

[0015] Step S21, randomly select the spatial coordinate system of at least three samples from the point cloud data in step S1, and calculate the related plane parameters, including the plane normal and a point on the plane, to generate the plane equation

[0016] Ax+By+Cz+D=0

[0017] where A, B, C, and D are known constants, and A, B, and C are not zero at the same time;

[0018] Step S22, calculate the distance from the point cloud to the plane in step S21 according to the point-to-plane distance formula, when the number of randomly selected samples reaches the set threshold of 1000 times, the plane recognition is completed, and the found plane data is filtered out;

[0019] Step S23, repeat the above process until all planes are found.

[0020] Further, the specific steps of unifying the plane normal in step S3 are:

[0021] Step S31, form a spatial vector CiO with the plane point cloud centroid Ci identified in step S2 and the visual coordinate origin O, and calculate the scalar product of the vector CiO(x1, y1, z1) and the plane normal n(a, b, c)

[0022] d=x1×a+y1×b+z1×c

[0023] If it is negative, the new plane equation is updated as

[0024] (-A)x+(-B)y+(-C)z+(-D)=0.

[0025] Further, the specific steps of calculating the plane boundary in step S4 are:

[0026] Step S41, use the improved alpha-step Shape algorithm to extract the boundary of the plane point cloud data found in step S2, denoted as the plane point cloud data set Q.

[0027] Further, the specific calculation steps of the improved alpha-step Shape algorithm in step S41 are:

[0028] Step S411, first calculate the minimum circumscribed rectangle containing the plane point cloud P in step S41;

[0029] Step S412, then construct a super triangle containing the rectangle, create an empty triangle set T, and add the super triangle to the triangle set T;

[0030] Step S413, for each point cloud P∈Q, find the triangle containing point P from the triangle set T, connect point P with the vertices of the triangle, construct a new triangle, and add it to the triangle set T;

[0031] Step S414, delete the super triangle from the triangle set T to obtain the final triangle set T, that is, the constructed triangular mesh;

[0032] Step S415, keep the triangle in the triangular mesh whose three edge lengths are less than or equal to twice the radius of the alpha circle, and there is no point in at least one circle with a radius of alpha passing through two points on the edge of the triangle, then the edge is the boundary to be found.

[0033] Further, in step S5, the two points with a distance less than the set distance of 1mm are taken as a group of point pairs by using the nearest neighbor search of each two groups of boundary nearest neighbors in step S4.

[0034] Further, the straight line fitting in step S6 is specifically as follows:

[0035] Step S61, the least square method is used to fit a straight line to the candidate common boundary obtained in step S5, and two maximum distance point pairs in the direction of the straight line are taken as the welding seam end points.

[0036] Further, the method for path planning in step S7 is as follows:

[0037] Step S71, the welding seam end points obtained in step S61 are used as a space vector v, and the angle θ between the vector v and a reference direction vector r is calculated according to the vector dot product formula v·r=|v||r|cosθ.

[0038] Step S72, the straight line obtained in step S61 is classified into horizontal welding lines and vertical welding lines according to the angle θ calculated in step S71.

[0039] Step S73, all welding line end points obtained in step S72 are labeled.

[0040] Step S74, then all welding line end points in step S73 are grouped according to the distance, that is, the end points are classified into the same group if the distance between them meets the set distance of 1mm, the number of end points in each group is counted, and as long as the number of end points is greater than 1, the label of the straight line end point in the group is recorded as the straight line connection point, that is, the straight line is connected to form a welding path.

[0041] Compared with the prior art, the beneficial effects of the present application are as follows:

[0042] 1. The welding method of the present application extracts the welding line and plans the welding path by analyzing the geometric features in the point cloud data and the spatial relationship of each plane, has the characteristics of stable precision, strong expansibility, low cost and high adaptability, and effectively improves the production efficiency.

[0043] 2. The welding method provided by the present application automatically solves the plate welding problem in steel structure, uses a 3D vision system to identify the weld, automatically plans the welding path, saves labor cost, improves welding precision stability, and shortens the welding cycle. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 The figure is a flowchart of the welding method in the present application. DETAILED DESCRIPTION

[0045] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0046] Please refer to Figure 1 A steel structure welding method, comprising the following steps:

[0047] Step S1, align the welds of the two steel structure plates at the positions to be welded, and then scan the point cloud data of the steel structure using a vision sensor;

[0048] Step S2, based on the data scanned in step S1, use the random sample consensus method to find the plane, the specific steps are:

[0049] Step S21, randomly select a space coordinate system of at least 3 samples from the point cloud data in step S1, and calculate the related plane parameters, including the plane normal and a point on the plane, to generate the plane equation

[0050] Ax+By+Cz+D=0

[0051] Where A, B, C, and D are known constants, and A, B, and C are not zero at the same time;

[0052] Step S22, calculate the distance from the point cloud to the plane in step S21 according to the point-to-plane distance formula, when the number of randomly selected samples reaches the set threshold of 1000 times, the plane recognition is completed, and the found plane data is filtered out;

[0053] Step S23, repeat the above process until all planes are found.

[0054] Step S3, the plane normal direction obtained in step S2 has uncertainty, but the same spatial coordinate system is used when collecting data by 3D vision, which is used to unify the plane normal direction, and the specific steps are as follows:

[0055] Step S31, the plane point cloud centroid Ci identified in step S2 and the vision coordinate origin O form a spatial vector CiO, and the scalar product of the vector CiO(x1, y1, z1) and the plane normal n(a, b, c) is calculated

[0056] d=x1×a+y1×b+z1×c

[0057] If it is negative, the new plane equation is updated as

[0058] (-A)x+(-B)y+(-C)z+(-D)=0.

[0059] Step S4, calculate the boundary between planes according to the plane obtained in step S2, and the specific steps are as follows:

[0060] Step S41, use the improved alpha-step Shape algorithm to extract the boundary of the plane point cloud data found in step S2, and mark it as a plane point cloud data set Q;

[0061] The specific calculation steps of the improved alpha-step Shape algorithm in step S41 are as follows:

[0062] Step S411, first calculate the minimum circumscribed rectangle containing the plane point cloud P in step S41;

[0063] Step S412, then build a super triangle containing the rectangle, create an empty triangle set T, and add the super triangle to the triangle set T;

[0064] Step S413, for each point cloud P∈Q, find the triangle containing point P from the triangle set T, connect point P with the vertices of the triangle, construct a new triangle, and add it to the triangle set T;

[0065] Step S414, delete the super triangle from the triangle set T to obtain the final triangle set T, which is the constructed triangular mesh;

[0066] Step S415, keep the triangles in the triangular mesh whose three edge lengths are less than or equal to twice the alpha circle radius, and the two circles with a radius of alpha passing through the two points of the triangle edge have at least one circle without points, then the edge is the boundary to be found.

[0067] Step S5, using the nearest neighbor search step S4 of each two groups of boundary points, the distance between two points less than the set distance 1mm as a group of points, all pairs of points are combined into a candidate common boundary, the nearest neighbor search method is Kd-tree search, which can effectively speed up the search speed, the specific search process is:

[0068] A: Construct kd-tree

[0069] (a) Select a suitable dimension d, and sort the boundary data according to the dimension.

[0070] (b) Find the median as the division point of the current sorting node, and divide the data into two subsets with the point as the boundary.

[0071] (c) Recursively perform the same operation on the left and right subsets, until each node contains only one data point or no data point.

[0072] (d) Recursively build the entire kd-tree.

[0073] B: Search of kd-tree

[0074] (a) Start from the root node, compare the size relationship of the query point and the division point of the current node in the dimension.

[0075] (b) According to the result, select the left or right subtree for recursive search until the target data or nearest neighbor data is found.

[0076] (c) In the search process, record the nearest neighbor data and its distance, and update its value in the backtracking process, and finally search all two points with a distance less than the set distance 1mm.

[0077] Step S6, straight line fitting is performed on the common boundary obtained in step S5, and the end points are extracted as the weld, the specific steps are as follows:

[0078] Step S61, the least square method is used to fit a straight line to the candidate common boundary obtained in step S5, and two maximum distance points are taken along the straight line direction as the weld end points.

[0079] Step S7, distance-based path planning, the specific method is:

[0080] Step S71, using the weld end points obtained in step S61 as a space vector v, the angle θ between the vector v and the reference direction vector r is calculated according to the vector dot product formula v·r=|v||r|cosθ;

[0081] Step S72, according to the angle θ calculated in step S71, the straight line obtained in step S61 is classified into horizontal weld line and vertical weld line;

[0082] Step S73, label all the welding line endpoints obtained in step S72;

[0083] Step S74, then group all the welding line endpoints in step S73 by distance, that is, endpoints are classified into the same group if the distance between them meets the set distance of 1mm, count the number of endpoints in each group, and as long as the number of endpoints is greater than 1, record the linear connection point of the linear endpoint label in this group, that is, connect the obtained straight line to form a welding path.

[0084] For example, three welding lines are obtained, which are respectively named as straight line A: 1-2, straight line B: 3-4, and straight line C: 5-6. When the 1 endpoint of straight line A and the 6 endpoint of straight line C meet the requirements of step S74, straight line A and straight line C are connected as 2-1-6-5. When the 5 endpoint of straight line C and the 3 endpoint of straight line B meet the requirements of step S74, the connection mode of straight line A, straight line B and straight line C is 2-1-6-5-3-4. When the 5 endpoint of straight line C and the 4 endpoint of straight line B meet the requirements of step S74, the connection mode of straight line A, straight line B and straight line C is 2-1-6-5-4-3.

[0085] In summary, the main steps of the method are: first, input the point cloud data of the steel structure; identify the plane of the point cloud data; then, unify the normal of the identified plane; calculate the boundary of the identified plane; calculate the common boundary using the nearest neighbor method; perform straight line fitting on the common boundary to extract the welding line endpoints.

[0086] The above is only a preferred embodiment of the present application, and is not used to limit the present application. The patent protection scope of the present application is subject to the claims, and any equivalent structural changes made by using the content of the specification and drawings of the present application should also be included in the protection scope of the present application.

Claims

1. A method for welding steel structures, characterized in that: Includes the following steps: Step S1: Scan the point cloud data of the steel structure; Step S2: Based on the data scanned in step S1, the plane is found using the random sampling consistency method; Step S3: Unify the normal of the plane obtained in step S2 according to the visual coordinates; Step S4: Calculate the inter-plane boundaries based on the planes obtained in step S2; The specific steps for calculating the planar boundary in step S4 are as follows: Step S41: Use the improved alpha-step Shape algorithm to extract the planar point cloud data boundary found in step S2, denoted as the planar point cloud data set. ; The specific calculation steps of the improved alpha-step Shape algorithm in step S41 are as follows: Step S411: First, calculate the planar point cloud containing the data from step S41. The minimum bounding rectangle; Step S412: Then construct a super triangle containing the rectangle, and create an empty triangle set. Add the super triangle to the triangle collection. middle; Step S413: For each point cloud ∈ Find the triangle containing point p from the triangle set T, connect point p to the vertex of the triangle to construct a new triangle, and add it to the triangle set T; Step S414: From the set of triangles Remove the super triangle to obtain the final set of triangles. This is the constructed triangular network; Step S415: Keep the triangle in the triangulation where the length of each of the three sides is less than or equal to twice the radius of the circle alpha. And among the two circles with radius alpha passing through two points on the sides of the triangle, at least one circle has no points inside it. This side is the boundary to be found. Step S5: Use the nearest neighbor search in step S4 to find the nearest neighbor points of every two sets of boundaries. Two points with a distance of less than 1 mm are considered as a pair of points, and all pairs of points are combined into a candidate common boundary. Step S6: Perform straight line fitting on the common boundary obtained in step S5, and extract the endpoints as welds; The specific steps for line fitting in step S6 are as follows: Step S61: Fit a straight line to the candidate common boundary obtained in step S5 using the least squares method, and take the two maximum distance points along the direction of the straight line as the weld endpoints. Step S7: Perform path planning based on distance methods; The method for path planning in step S7 is as follows: Step S71: Use the weld endpoints obtained in step S61 as spatial vectors. According to the dot product formula: ； Calculate vectors With reference direction vector Angle between ; Step S72: Based on the angle calculated in step S71 The straight lines obtained in step S61 are classified into horizontal bonding lines and vertical bonding lines; Step S73: Label all the wire bonding endpoints obtained in step S72; Step S74: Then, group all the welding wire endpoints in step S73 according to their distance. That is, if the distance between the endpoints meets the set distance of 1mm, they are classified into the same group. Count the number of endpoints in each group. As long as the number of endpoints is greater than 1, record the label of the straight line endpoint in this group as the straight line connection point. That is, connect the obtained straight lines to form the welding path.

2. The steel structure welding method according to claim 1, characterized in that: The specific steps for finding the plane in step S2 are as follows: Step S21: Randomly select at least 3 spatial coordinate systems from the point cloud data in Step S1, and calculate the relevant plane parameters, including the plane normal and a point on the plane, to generate the plane equation: ； Where A, B, C, and D are known constants, and A, B, and C are not all zero at the same time; Step S22: Calculate the distance from the point cloud (excluding the selected samples) to the plane in step S21 according to the point-to-plane distance formula. When the number of random sample selections reaches the set threshold of 1000 times, the plane recognition is completed, and the found plane data is filtered out. Step S23: Repeat the above process until all planes are found.

3. The steel structure welding method according to claim 1, characterized in that: The specific steps for unifying the plane normal in step S3 are as follows: Step S31: The centroid of the planar point cloud identified in step S2 Together with the visual coordinate origin O, form a spatial vector CiO. Calculate the vector CiO( ) and plane normal Dot product: ； If it is negative, the new plane equation is updated to 。

Images