CN113793344B - Impeller weld joint positioning method based on three-dimensional point cloud - Google Patents
Impeller weld joint positioning method based on three-dimensional point cloud Download PDFInfo
- Publication number
- CN113793344B CN113793344B CN202111018065.5A CN202111018065A CN113793344B CN 113793344 B CN113793344 B CN 113793344B CN 202111018065 A CN202111018065 A CN 202111018065A CN 113793344 B CN113793344 B CN 113793344B
- Authority
- CN
- China
- Prior art keywords
- point
- pcd
- point cloud
- vector
- cloud
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 56
- 238000003466 welding Methods 0.000 claims abstract description 39
- 230000011218 segmentation Effects 0.000 claims abstract description 14
- 238000012545 processing Methods 0.000 claims abstract description 7
- 238000005070 sampling Methods 0.000 claims abstract description 7
- 238000001914 filtration Methods 0.000 claims abstract description 6
- 239000013598 vector Substances 0.000 claims description 74
- 239000011159 matrix material Substances 0.000 claims description 18
- 230000009466 transformation Effects 0.000 claims description 12
- 238000000354 decomposition reaction Methods 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 4
- 238000007781 pre-processing Methods 0.000 claims description 4
- 241001061225 Arcos Species 0.000 claims description 3
- 230000001419 dependent effect Effects 0.000 claims description 3
- 238000000638 solvent extraction Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 239000000428 dust Substances 0.000 description 1
- 239000012535 impurity Substances 0.000 description 1
- 238000000746 purification Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000009423 ventilation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G06T3/06—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration by the use of local operators
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10028—Range image; Depth image; 3D point clouds
Abstract
The invention discloses an impeller weld positioning method based on three-dimensional point cloud, which is used for acquiring impeller three-dimensional point cloud data comprising two blades from the front of a weld start point to the rear of the weld end point based on a pre-scanning mode; dividing the vertical plate blade point cloud and the bottom plate blade point cloud by using a region growing and dividing algorithm; processing the blade point cloud of the bottom plate to obtain a partial point set close to the welding line area, grouping the partial point set at self-adaptive intervals and fitting a plane; projecting a set of riser points within a certain distance from the plane within the segmentation range of each fitting plane onto the plane; and finally, filtering, fitting and uniformly sampling the projection point sets to obtain a final weld point set. The method has the characteristics of high precision and high efficiency, meets the wrap angle requirement, and has certain anti-interference capability on the point welding point and the reflection.
Description
Technical Field
The invention relates to the technical field of weld joint positioning, in particular to an impeller weld joint positioning method based on three-dimensional point cloud.
Background
Industrial fans are widely applied to the fields of environmental protection purification, industrial dust removal, building ventilation and the like, the demand of the fans is increased along with the development of the environmental protection industry and the building industry in recent years, and the quality of welding of impellers serving as key parts of the fans directly influences the service life and the running condition of the fans.
The impeller has a simple structure and consists of blades and a base, and the blades are required to be fixed on a base bracket through welding. Traditional impeller welding relies on the manual work to weld the seam piece by piece, and this mode welding efficiency is low, with high costs, welding quality is uneven, and can cause harm to welding personnel's health.
At present, most of impeller welding adopts a welding system based on laser vision, the system fixes a vision system at the tail end of a mechanical arm, the laser vision system at the tail end of the mechanical arm is manually taught to scan the area where a welding line is located, a camera shoots a plurality of images containing laser lines, the coordinates of inflection points of the welding line are obtained through an image processing algorithm and a calibration algorithm, and finally a fitting algorithm is used for fitting the track of the welding line. The method improves the automation degree of impeller welding, has the advantages of high efficiency and high positioning precision, but has higher requirements on impeller materials and assembly.
The welding system based on laser vision utilizes an image processing algorithm to extract the center of a laser line in an image and calculate an inflection point coordinate to realize a positioning function, and if the impeller is special in material or unreasonable in shooting angle or oil dirt and impurities exist, the laser image shot by a camera has a reflection phenomenon, so that the positioning algorithm cannot accurately extract the center of the laser line and calculate the inflection point coordinate; moreover, the consistency of each blade is poor when the impeller is assembled and spot-welded, a section of missing condition possibly occurs in a welding line track which is guided to take a picture and is finally calculated through a teaching mode, and a good welding line wrap angle function (namely, a welding line track from a welding line starting point to a welding line finishing point) cannot be stably realized; in addition, the presence of spot welds on the weld may cause weld localized failure in the spot weld area resulting in a weld trajectory shift or loss.
Disclosure of Invention
The weld joint positioning method based on the three-dimensional point cloud has certain anti-interference capability on local reflection and spot welding points, has high positioning precision, and can realize the wrap angle function; the adopted technical scheme is as follows:
an impeller weld positioning method based on three-dimensional point cloud, comprising the following steps:
s1, acquiring three-dimensional point cloud data of an impeller, wherein the three-dimensional point cloud data of the impeller comprise point cloud data of a vertical plate blade and point cloud data of a bottom plate blade;
s2, dividing the three-dimensional point cloud of the impeller by a region growing and dividing algorithm, and distinguishing a vertical plate blade point cloud PCD_v and a bottom plate blade point cloud PCD_b according to the characteristics of the point cloud;
s3, extracting point cloud PCD_bs on a bottom plate blade close to a vertical plate blade, resolving a first principal vector PCA_vec1 of the point cloud PCD_bs by using a PCA algorithm, determining an axis1 coordinate axis of a certain coordinate axis of the nearest xyz according to the first principal vector PCA_vec1, calculating a transformation matrix by the first principal vector of the point cloud PCD_bs and the axis1 coordinate axis unit vector through axial angle transformation, and converting the point cloud PCD_bs and the point cloud PCD_v into a new coordinate system, wherein the transformed point cloud is the point cloud PCD_bs_t and the point cloud PCD_v_t respectively, and the first principal vector of the transformed point cloud PCD_bs_t is the axis1 coordinate axis vector; solving a second main vector of the point cloud PCD_bs_t, and determining an axis2 coordinate axis closest to the second main vector according to the second main vector of the point cloud PCD_bs_t;
s4, equally dividing the point cloud PCD_bs_t along the axis1 axis, extracting point sets along the axis2 axis to obtain a plurality of groups of divided point sets, respectively fitting planes to the plurality of groups of divided point sets, and obtaining the boundary range of each group of point sets along the axis1 axis direction;
s5, projecting the point set of the point cloud PCD_v_t within a proper distance d3 from each fitting plane onto the corresponding fitting plane to obtain a projection point set PCD_v_t_p;
s6, converting the projection point set PCD_v_t_p into a point set PCD_contour under an original coordinate system through an inverse matrix of the transformation matrix calculated in the step S3, preprocessing the point set PCD_Contour, fitting a weld track by a least square method, and uniformly sampling to obtain a final weld point set.
Preferably, the specific method for segmenting the three-dimensional point cloud of the impeller by using the region growing segmentation algorithm in the step S2 includes:
(a) Selecting a point with the minimum curvature value from the residual point set list A as a seed point Pmin, and adding the seed point Pmin into a seed list Sc; sequentially processing each point in the seed list Sc, and searching a neighboring point set Bc of each seed point Pmin within the r range by using a neighboring point searching method omega (Sc); judging whether points in each adjacent point set Bc are in the residual point set list A, if so, calculating the angles between the normals of the points in the residual point set list A and the adjacent point set Bc and the normals of the seed points Pmin, adding the points in the adjacent point set Bc with the angles smaller than an angle threshold value θth into a cluster Rc, and removing the points in the adjacent point Bc from the residual point set list A; judging the curvature value of points in the adjacent points Bc, and adding the points in the adjacent point set Bc with the curvature smaller than a threshold Cth into a seed list Sc; sequentially judging the adjacent point set of each seed point until the seed list Sc is empty, and adding the cluster Rc into the cluster list R;
(b) Clearing the seed list Sc and the cluster Rc, reselecting the point with the smallest curvature value from the residual point set list A as a seed point, and repeating the step (a) until the residual point set list A is empty;
the final clustering list R is the segmentation result of the algorithm, and the two clusters with the largest extraction quantity are the vertical plate blade point cloud and the bottom plate blade point cloud.
The calculation method of the normal line of the point in the point cloud comprises the following steps: and (3) searching a neighboring point set Bc with the radius within the range of r by using a neighboring point searching method for each point in the point cloud, fitting a plane to the neighboring point set Bc, taking a normal vector of the plane as a normal line of the point, and calculating the normal line of each point in the neighboring point set Bc and the seed point Pmin in the step (a) according to the method.
Initially, the rest point set list a is all point clouds. Preferably, the adjacent point set Bc of each seed point Pmin within the range of 5mm is searched by using the adjacent point searching method Ω (Sc).
Preferably, in the step S2, the point cloud feature includes a curvature value, where the curvature value is small is a riser blade point cloud pcd_v, and the curvature value is large is a bedplate blade point cloud pcd_b.
Preferably, in the step S3, the method for extracting the point cloud pcd_bs on the bottom plate blade close to the riser blade includes the following steps:
calculating a centroid point pt_v of the vertical plate blade point cloud PCD_v and a centroid point pt_b of the bottom plate blade point cloud PCD_b, wherein a direction vector from the centroid point pt_v to the centroid point pt_b is dir_v2b; fitting a plane to the point cloud of the vertical plate blade, wherein the plane coefficient is plane_v, and adjusting a plane normal vector plane_dir to be in the same direction as the dir_v2b vector; extracting a point set within a directional distance range from d1 to d2 from the plane fitted by the vertical plate blade point cloud from the bottom plate blade point cloud PCD_b, setting a point on the plane fitted by the vertical plate blade point cloud as ptinPlane, and setting the i-th point in the bottom plate blade point cloud PCD_b as PCD_bi, wherein the directional distance from the i-th point to the plane fitted by the vertical plate point cloud is
dis＝(PCD_bi-ptInPlane) T * The plane_dir is the point set meeting the condition is PCD_bs; the d1 ranges from-15 mm to 0mm, and the d2 ranges from 10mm to 20mm.
Preferably, in the step S3, the method for converting the point cloud pcd_bs and the point cloud pcd_v into a new coordinate system includes the following steps:
calculating the mean value of the point cloud PCD_bs as PCD_bs_ave, and calculating covariance as
Cov(PCD_bs)＝(PCD_bs-PCD_bs_ave)(PCD_bs-PCD_bs_ave) T SVD decomposition is carried out on the covariance matrix, a feature vector corresponding to the maximum feature value is taken as a first principal vector PCA_vec1, meanwhile, a coordinate axis unit vector with the minimum included angle with the first principal vector is obtained as axis1_vec, and the included angle theta=arcos (PCA_vec1) of the two vectors is calculated T * axis1_vec), solving for the two-vector cross-product vector n=cross (pca_vec1, axis1_vec), the transformation matrix can be calculated from pcd_bs_ave, N and θ as:
transVector＝-rotationMatrix*PCD_bs_ave
converting the homogeneous coordinates of PCD_bs and PCD_v by multiplying the homogeneous coordinates by TransMatrix into a new coordinate system, wherein the transformed two-blade point clouds are PCD_bs_t and PCD_v_t respectively.
According to the first main vector and the nearest axis unit vector of the bottom plate blade point cloud, converting the impeller point cloud into a new coordinate system taking the welding line direction as the nearest axis direction, and conveniently grouping the bottom plate point cloud along the nearest axis direction.
Preferably, in the step S4, the method for partitioning the point cloud pcd_bs_t includes the following steps: equally dividing the point cloud PCD_bs_t into a plurality of groups of divided point sets at interval intervals along the axis direction of axis1, checking the number of each group of divided point sets after grouping, and increasing interval width and then re-dividing the point cloud until the point number meets the requirement if the point number is smaller than pts_num 1; wherein 10< interval <20, 50< pts_num1<150.
Preferably, in the step S5, d3 is more than or equal to 4mm and less than or equal to 6mm.
Preferably, the step S6 includes the steps of: carrying out statistical filtering on the point set PCD_Contours, constructing a covariance matrix, carrying out SVD decomposition, extracting a feature vector corresponding to the maximum feature value to obtain a weld point set main vector concour_vec, and determining a coordinate axis concour_axis closest to the weld point set main vector; constructing two polynomials by taking concour_axis as independent variables and taking the other two dimensions as dependent variables respectively, and solving polynomial coefficients by using a least square method; and uniformly sampling the welding seam by using a start point and a stop point of the welding seam point set and two polynomial equations, wherein the sampled point set is the final welding seam point set.
Preferably, in the step S1, the three-dimensional point cloud data of the impeller is filtered, and the number of point clouds is reduced through downsampling.
The invention has the beneficial effects that: the method for positioning the welding line based on the point cloud segmentation, projection and fitting method in three dimensions has high positioning accuracy, certain anti-interference performance on the point welding point and reflection and a wrap angle function. The weld joint point set is verified point by a high-precision three-coordinate measuring machine, and verification results show that the positioning error of the weld joint positioning method based on the three-dimensional point cloud provided by the invention is less than or equal to 1mm, and the technical requirements of impeller welding are met.
Drawings
FIG. 1 is a flow diagram of a three-dimensional point cloud based weld positioning method according to the present invention;
FIG. 2 is a schematic diagram of a point cloud PCD_bs extraction;
FIG. 3 is a schematic illustration of a set of projection points;
FIG. 4 is a schematic view of an impeller construction and weld;
FIG. 5 is a number one weld point cloud segmentation map and weld positioning effect map;
FIG. 6 is a weld point cloud segmentation map number two and a weld positioning effect map;
FIG. 7 is a third weld point cloud segmentation map and a weld positioning effect map;
fig. 8 is a fourth weld point cloud segmentation map and a weld positioning effect map.
Detailed Description
The invention is further described with reference to a flow chart of a weld positioning method based on three-dimensional point cloud shown in fig. 1:
an impeller weld positioning method based on three-dimensional point cloud, comprising the following steps:
s1, acquiring three-dimensional point cloud data of an impeller, preprocessing the three-dimensional point cloud of the impeller, filtering discrete points by a statistical filtering method, and reducing the number of the point cloud by downsampling to obtain low-noise and sparse three-dimensional point cloud data; the three-dimensional point cloud data of the impeller comprise point cloud data of the vertical plate blades and point cloud data of the bottom plate blades;
s2, calculating a point cloud normal and curvature, dividing the three-dimensional point cloud of the impeller by a region growing and dividing algorithm, and distinguishing a vertical plate blade point cloud PCD_v and a bottom plate blade point cloud PCD_b according to the characteristics of the point cloud;
s3, extracting point cloud PCD_bs on a bottom plate blade close to a vertical plate blade, resolving a first main vector of the point cloud PCD_bs by using a PCA algorithm, wherein the first main vector is basically consistent with a weld joint main vector, determining an axis1 coordinate axis of a certain coordinate axis of the closest xyz according to the first main vector, calculating a transformation matrix by the first main vector of the point cloud PCD_bs and the axis1 coordinate axis unit vector through axial angle transformation, and converting the point cloud PCD_bs and the point cloud PCD_v into a new coordinate system, wherein the transformed point cloud is the point cloud PCD_bs_t and the point cloud PCD_v_t respectively, and the transformed first main vector of the point cloud PCD_bs_t is the axis1 coordinate axis vector; solving a second principal vector of the point cloud PCD_bs_t, and determining an axis2 coordinate axis (perpendicular to the welding line direction) closest to the second principal vector according to the second principal vector of the point cloud PCD_bs_t;
the coordinate conversion is to convert the impeller point cloud into a new coordinate system taking the welding line direction as the nearest axis direction according to the first main vector and the nearest axis unit vector of the point cloud PCD_bs, and the function is to facilitate grouping processing of the point cloud PCD_bs along the nearest axis direction (namely the welding line direction);
s4, equally dividing the point cloud PCD_bs_t along the axis1 axis, extracting point sets along the axis2 axis to obtain a plurality of groups of divided point sets, respectively fitting planes to the plurality of groups of divided point sets, and obtaining the boundary range of each group of point sets along the axis1 axis direction;
s5, projecting a point set within a proper distance from the point cloud PCD_v_t to each fitting plane to the corresponding fitting plane to obtain a projection point set PCD_v_t_p;
s6, converting the projection point set PCD_v_t_p into a point set PCD_contour under an original coordinate system through an inverse matrix of the transformation matrix calculated in the step S3, preprocessing the point set PCD_Contour, fitting a weld track by a least square method, and uniformly sampling to obtain a final weld point set.
The specific method for segmenting the three-dimensional point cloud of the impeller by the region growing segmentation algorithm in the step S2 comprises the following steps:
setting the point cloud as P, the point cloud normal as N, the point cloud curvature as C, the adjacent point searching method as omega, the clustering result list as R, the rest point set list as A, the curvature threshold of the segmentation algorithm as Cth and the angle threshold as theta th;
(a) Selecting a point with the minimum curvature value from the residual point set list A as a seed point Pmin, and adding the seed point Pmin into a seed list Sc; sequentially processing each point in the seed list Sc, and searching a neighboring point set Bc of each seed point Pmin within the r range by using a neighboring point searching method omega (Sc); judging whether points in each adjacent point set Bc are in the residual point set list A, if so, calculating the normal angles between the normals of the points in the residual point set list A and the adjacent point set Bc and the normal angles of the seed points Pmin, adding the points in the adjacent point set Bc with the included angles smaller than an angle threshold value θth (θth=4.5°) into a cluster Rc, and removing the points in the adjacent points Bc from the residual point set list A; judging the curvature value of the points in the adjacent points Bc, and adding the points in the adjacent point set Bc with the curvature smaller than a threshold Cth (Cth=5) into a seed list Sc; sequentially judging the adjacent point set of each seed point until the seed list Sc is empty, and adding the cluster Rc into the cluster list R;
(b) Clearing the seed list Sc and the cluster Rc, reselecting the point with the smallest curvature value from the residual point set list A as a seed point, and repeating the step (a) until the residual point set list A is empty;
the final clustering list R is the segmentation result of the algorithm, and the two clusters with the largest extraction quantity are the vertical plate blade point cloud and the bottom plate blade point cloud.
The calculation method of the normal line of the point in the point cloud comprises the following steps: searching a neighboring point set Bc with the radius within the range of r by using a neighboring point searching method for each point in the point cloud, fitting a plane to the neighboring point set Bc, taking a normal vector of the plane as a normal line of the point, and calculating the normal line of each point in the neighboring point set Bc and a seed point Pmin in the step (a) according to the method;
in the step S2, the point cloud features include curvature values, and the plate blade point cloud pcd_v and the base plate blade point cloud pcd_b may be distinguished according to the curvature values, where the curvature values are small for the plate blade point cloud pcd_v and the curvature values are large for the base plate blade point cloud pcd_b.
Referring to fig. 2, in step S3, the method for extracting the point cloud pcd_bs on the bottom plate blade close to the riser blade includes the following steps:
calculating a centroid point pt_v of the vertical plate blade point cloud PCD_v and a centroid point pt_b of the bottom plate blade point cloud PCD_b, wherein a direction vector from the centroid point pt_v to the centroid point pt_b is dir_v2b; fitting a plane to the point cloud of the vertical plate blade, wherein the plane coefficient is plane_v, and adjusting a plane normal vector plane_dir to be in the same direction as the dir_v2b vector; extracting a point set within a directional distance range from d1 to d2 from the plane fitted by the vertical plate blade point cloud from the bottom plate blade point cloud PCD_b, setting a point on the plane fitted by the vertical plate blade point cloud as ptinPlane, and setting the i-th point in the bottom plate blade point cloud PCD_b as PCD_bi, wherein the directional distance from the i-th point to the plane fitted by the vertical plate point cloud is
dis＝(PCD_bi-ptInPlane) T * The plane_dir is the point set meeting the condition is PCD_bs; the d1 is-5 mm, and the d2 is 15mm.
In step S3, the method for converting the point cloud pcd_bs and the point cloud pcd_v into a new coordinate system includes the following steps:
calculating the mean value of the point cloud PCD_bs as PCD_bs_ave, and calculating covariance as
Cov(PCD_bs)＝(PCD_bs-PCD_bs_ave)(PCD_bs-PCD_bs_ave) T SVD decomposition is carried out on the covariance matrix, a feature vector corresponding to the maximum feature value is taken as a first principal vector PCA_vec1, meanwhile, a coordinate axis unit vector with the minimum included angle with the first principal vector is obtained as axis1_vec, and the included angle theta=arcos (PCA_vec1) of the two vectors is calculated T * axis1_vec), solving for the two-vector cross-product vector n=cross (pca_vec1, axis1_vec), the transformation matrix can be calculated from pcd_bs_ave, N and θ as:
transVector＝-rotationMatrix*PCD_bs_ave
converting the homogeneous coordinates of PCD_bs and PCD_v by multiplying the homogeneous coordinates by TransMatrix into a new coordinate system, wherein the transformed two-blade point clouds are PCD_bs_t and PCD_v_t respectively.
In step S4, the method for partitioning the point cloud pcd_bs_t includes the following steps: equally dividing the point cloud PCD_bs_t into a plurality of groups of divided point sets at interval intervals along the axis direction of axis1, checking the number of each group of divided point sets after grouping, and increasing interval width and then re-dividing the point cloud until the point number meets the requirement if the point number is smaller than pts_num 1; wherein interval=15 mm, pts_num1=100.
Referring to fig. 3, in the step S5, a set of points pcd_v_t within 5mm from each fitting plane is projected onto each corresponding fitting plane, so as to obtain a set of projected points pcd_v_t_p.
The step S6 includes the steps of: carrying out statistical filtering on the point set PCD_Contours, constructing a covariance matrix, carrying out SVD decomposition, extracting a feature vector corresponding to the maximum feature value to obtain a weld point set main vector concour_vec, and determining a coordinate axis concour_axis closest to the weld point set main vector; constructing two polynomials by taking concour_axis as independent variables and taking the other two dimensions as dependent variables respectively, and solving polynomial coefficients by using a least square method; and uniformly sampling the welding seam by using a start point and a stop point of the welding seam point set and two polynomial equations, wherein the sampled point set is the final welding seam point set.
The invention locates the weld joint on the three-dimensional basis based on the point cloud segmentation, projection and fitting method, the invention collects 4 different weld joint point cloud data of the impeller to test the locating method described in the embodiment, wherein FIG. 4 is a schematic diagram of the impeller structure and the weld joint, and FIGS. 5 to 8 are respectively point cloud segmentation of 4 weld joints and weld joint locating effect diagrams, and from the weld joint locating effect diagrams, it can be seen that the locating method has good locating precision, has certain anti-interference capability on the point welding joint and reflection, and has the function of edge wrapping; the weld joint point set is verified point by a high-precision three-coordinate measuring machine, and verification results show that the positioning error of the weld joint positioning method based on the three-dimensional point cloud provided by the invention is less than or equal to 1mm, and the technical requirements of impeller welding are met.
The values of the parameters are more than 0 and less than 10 Cth, more than 0 and less than 6 DEG, more than 15mm and less than 1 and less than 0mm, more than 10mm and less than 2 and less than 20mm, more than 50 and less than pts_num1 and less than 150, more than 10mm and less than 20mm.
Claims (9)
1. The impeller weld joint positioning method based on the three-dimensional point cloud is characterized by comprising the following steps of:
s1, acquiring three-dimensional point cloud data of an impeller, wherein the three-dimensional point cloud data of the impeller comprise point cloud data of a vertical plate blade and point cloud data of a bottom plate blade;
s2, dividing the three-dimensional point cloud of the impeller by a region growing and dividing algorithm, and distinguishing a vertical plate blade point cloud PCD_v and a bottom plate blade point cloud PCD_b according to the characteristics of the point cloud;
s3, extracting point cloud PCD_bs on a bottom plate blade close to a vertical plate blade, calculating a first main vector PCA_vec1 of the point cloud PCD_bs by using a PCA algorithm, determining an axis1 coordinate axis of a certain coordinate axis of the nearest xyz according to the first main vector PCA_vec1, calculating a transformation matrix by the first main vector of the point cloud PCD_bs and the axis1 coordinate axis unit vector through axial angle transformation, and converting the point cloud PCD_bs and the point cloud PCD_v into a new coordinate system, wherein the transformed point clouds are point cloud PCD_bs_t and point cloud PCD_v_t respectively, and the first main vector of the transformed point cloud PCD_bs_t is the axis1 coordinate axis vector; solving a second main vector of the point cloud PCD_bs_t, and determining an axis2 coordinate axis closest to the second main vector according to the second main vector of the point cloud PCD_bs_t;
s4, equally dividing the point cloud PCD_bs_t along the axis1 axis, extracting point sets along the axis2 axis to obtain a plurality of groups of divided point sets, respectively fitting planes to the plurality of groups of divided point sets, and obtaining the boundary range of each group of point sets along the axis1 axis direction;
s5, projecting a point set in a proper distance d3 from the point cloud PCD_v_t to each fitting plane to the corresponding fitting plane to obtain a projection point set PCD_v_t_p;
s6, converting the projection point set PCD_v_t_p into a point set PCD_Contour by the inverse matrix of the transformation matrix calculated in the step S3, preprocessing the point set PCD_Contour, fitting a weld joint track by a least square method, and uniformly sampling to obtain a final weld joint point set;
in the step S3, the method for converting the point cloud pcd_bs and the point cloud pcd_v into a new coordinate system includes the following steps:
calculating the mean value of the point cloud PCD_bs as PCD_bs_ave, and calculating the covariance as Cov (PCD_bs) = (PCD_bs-PCD_bs_ave) (PCD_bs-PCD_bs+ave) T SVD decomposition is carried out on the covariance matrix, a feature vector corresponding to the maximum feature value is taken as a first principal vector PCA_vec1, meanwhile, a coordinate axis unit vector with the minimum included angle with the first principal vector is obtained as axis1_vec, and the included angle theta=arcos (PCA_vec1) of the two vectors is calculated T * axis1_vec), solving for the two-vector cross-product vector n=cross (pca_vec1, axis1_vec), the transformation matrix can be calculated from pcd_bs_ave, N and θ as:
transVector＝-rotationMatrix*PCD_bs_ave
converting the homogeneous coordinates of PCD_bs and PCD_v by multiplying the homogeneous coordinates by TransMatrix into a new coordinate system, wherein the transformed two-blade point clouds are PCD_bs_t and PCD_v_t respectively.
2. The three-dimensional point cloud-based impeller weld positioning method of claim 1, wherein: the specific method for dividing the three-dimensional point cloud of the impeller by the region growing and dividing algorithm in the step S2 comprises the following steps:
(a) Selecting a point with the minimum curvature value from the residual point set list A as a seed point Pmin, and adding the seed point Pmin into a seed list Sc; sequentially processing each point in the seed list Sc, and searching a neighboring point set Bc of each seed point Pmin within the r range by using a neighboring point searching method omega (Sc); judging whether points in each adjacent point set Bc are in the residual point set list A, if so, calculating angles between normals of points in the residual point set list A and the adjacent point set Bc and normals of the seed points Pmin, adding the points in the adjacent point set Bc with the angles smaller than an angle threshold value θth into a cluster Rc, and removing the points in the adjacent point Bc from the residual point set list A; judging the curvature value of points in the adjacent points Bc, and adding the points in the adjacent point set Bc with the curvature smaller than a threshold Cth into a seed list Sc; sequentially judging the adjacent point set of each seed point until the seed list Sc is empty, and adding the cluster Rc into the cluster list R;
(b) Clearing the seed list Sc and the cluster Rc, reselecting the point with the smallest curvature value from the residual point set list A as a seed point, and repeating the step (a) until the residual point set list A is empty;
the final clustering list R is the segmentation result of the algorithm, and the two clusters with the largest extraction quantity are the vertical plate blade point cloud and the bottom plate blade point cloud.
3. The three-dimensional point cloud-based impeller weld positioning method of claim 1, wherein: in the step S2, the point cloud features include curvature values, where the curvature values are small and the curvature values are respectively the riser blade point cloud pcd_v and the floor blade point cloud pcd_b.
4. The three-dimensional point cloud-based impeller weld positioning method of claim 1, wherein: in the step S3, the method for extracting the point cloud pcd_bs on the bedplate blade close to the riser blade includes the following steps:
calculating a centroid point pt_v of the vertical plate blade point cloud PCD_v and a centroid point pt_b of the bottom plate blade point cloud PCD_b, wherein a direction vector from the centroid point pt_v to the centroid point pt_b is dir_v2b; fitting a plane to the point cloud of the vertical plate blade, wherein the plane coefficient is plane_v, and adjusting a plane normal vector plane_dir to be in the same direction as the dir_v2b vector; extracting the vertical plate blade from the bottom plate blade point cloud PCD_bSetting a point set of a point cloud fitting plane within a directional distance range from d1 to d2, wherein one point on the vertical plate blade point cloud fitting plane is ptInPlane, and the ith point in the bottom plate blade point cloud PCD_b is PCD_bi, and the directional distance from the ith point to the vertical plate point cloud fitting plane is dis= (PCD_bi-ptInPlane) T * The plane_dir is the point set meeting the condition is PCD_bs; the d1 ranges from-15 mm to 0mm, and the d2 ranges from 10mm to 20mm.
5. The three-dimensional point cloud-based impeller weld positioning method of claim 4, wherein: in the step S4, the method for partitioning the point cloud pcd_bs_t includes the following steps: equally dividing the point cloud PCD_bs_t into a plurality of groups of divided point sets at interval intervals along the axis direction of axis1, checking the number of each group of divided point sets after grouping, and increasing interval width and then re-dividing the point cloud until the point number meets the requirement if the point number is smaller than pts_num 1; wherein 10< interval <20, 50< pts_num1<150.
6. The three-dimensional point cloud-based impeller weld positioning method of claim 5, wherein the method comprises the steps of: in the step S5, d3 which is more than or equal to 4mm and less than or equal to 6mm.
7. The three-dimensional point cloud-based impeller weld positioning method of claim 6, wherein: the step S6 includes the steps of: carrying out statistical filtering on the point set PCD_Contours, constructing a covariance matrix, carrying out SVD decomposition, extracting a feature vector corresponding to the maximum feature value to obtain a weld point set main vector concour_vec, and determining a coordinate axis concour_axis closest to the weld point set main vector; constructing two polynomials by taking concour_axis as independent variables and taking the other two dimensions as dependent variables respectively, and solving polynomial coefficients by using a least square method; and uniformly sampling the welding seam by using a start point and a stop point of the welding seam point set and two polynomial equations, wherein the sampled point set is the final welding seam point set.
8. The three-dimensional point cloud-based impeller weld positioning method of claim 1, wherein: in the step S1, the three-dimensional point cloud data of the impeller is filtered, and the number of point clouds is reduced through downsampling.
9. The three-dimensional point cloud-based impeller weld positioning method of claim 2, wherein: r is 5mm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111018065.5A CN113793344B (en) | 2021-08-31 | 2021-08-31 | Impeller weld joint positioning method based on three-dimensional point cloud |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111018065.5A CN113793344B (en) | 2021-08-31 | 2021-08-31 | Impeller weld joint positioning method based on three-dimensional point cloud |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113793344A CN113793344A (en) | 2021-12-14 |
| CN113793344B true CN113793344B (en) | 2023-09-29 |
Family
ID=78879409
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111018065.5A Active CN113793344B (en) | 2021-08-31 | 2021-08-31 | Impeller weld joint positioning method based on three-dimensional point cloud |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113793344B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114227054B (en) * | 2022-01-05 | 2023-01-31 | 南昌大学 | Automatic detection method for tube plate welding seam based on 3D point cloud |
| CN115408549B (en) * | 2022-08-31 | 2024-04-12 | 深圳前海瑞集科技有限公司 | Workpiece point cloud filtering method and device, computer readable medium and electronic equipment |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110455187A (en) * | 2019-08-21 | 2019-11-15 | 哈尔滨工业大学 | A kind of detection method of the box body workpiece weld seam based on 3D vision |
| CN111814888A (en) * | 2020-07-14 | 2020-10-23 | 南京航空航天大学苏州研究院 | Three-dimensional scanning line point cloud gap step extraction method for aircraft skin butt joint |
| CN113177983A (en) * | 2021-03-25 | 2021-07-27 | 埃夫特智能装备股份有限公司 | Fillet weld positioning method based on point cloud geometric features |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE102007008598A1 (en) * | 2007-02-19 | 2008-08-21 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Automatic programming of robots to weld stapled profiles onto micropanels using digital image capture |
| US10755132B2 (en) * | 2016-01-15 | 2020-08-25 | Wuhan Wuda Zoyon Science And Technology Co., Ltd | Methods for extracting surface deformation feature of object based on linear scanning three-dimensional point cloud |
-
2021
- 2021-08-31 CN CN202111018065.5A patent/CN113793344B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110455187A (en) * | 2019-08-21 | 2019-11-15 | 哈尔滨工业大学 | A kind of detection method of the box body workpiece weld seam based on 3D vision |
| CN111814888A (en) * | 2020-07-14 | 2020-10-23 | 南京航空航天大学苏州研究院 | Three-dimensional scanning line point cloud gap step extraction method for aircraft skin butt joint |
| CN113177983A (en) * | 2021-03-25 | 2021-07-27 | 埃夫特智能装备股份有限公司 | Fillet weld positioning method based on point cloud geometric features |
Non-Patent Citations (1)
| Title |
|---|
| 斜率表征与卡尔曼滤波的焊缝跟踪方法研究;马波;林少铎;张南峰;高向东;;机电工程(第02期) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113793344A (en) | 2021-12-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113793344B (en) | Impeller weld joint positioning method based on three-dimensional point cloud | |
| CN110986788B (en) | Automatic measurement method based on three-dimensional point cloud livestock phenotype body size data | |
| US5521847A (en) | System and method for determining airfoil characteristics from coordinate measuring machine probe center data | |
| CN106407502B (en) | Blade section profile parameter evaluation method based on optimal matching | |
| CN107516098B (en) | Target contour three-dimensional information extraction method based on edge curvature angle | |
| CN110455187B (en) | Three-dimensional vision-based box workpiece weld joint detection method | |
| EP1760427B1 (en) | Methods for measuring the nozzle flow area between gas turbine engine vanes | |
| CN111336990A (en) | Tunnel section convergence rapid analysis method and device | |
| CN110823077A (en) | Wall surface guiding rule detection method and system based on three-dimensional point cloud | |
| CN113744351B (en) | Underwater structure light measurement calibration method and system based on multi-medium refraction imaging | |
| CN108846374B (en) | Power equipment inspection method and device | |
| CN112964172B (en) | Aviation blade surface measuring method and measuring equipment based on structured light camera | |
| CN113192116A (en) | Aviation blade thickness parameter measuring method based on structured light camera | |
| CN115578398A (en) | Weld point cloud segmentation method based on region growing method | |
| CN116402866A (en) | Point cloud-based part digital twin geometric modeling and error assessment method and system | |
| CN112697388A (en) | Method for measuring attitude angle of hypersonic wind tunnel model based on schlieren image | |
| CN108267104A (en) | A kind of axial workpiece radius size measuring method based on binocular vision | |
| CN108335332A (en) | A kind of axial workpiece central axes measurement method based on binocular vision | |
| CN112440011B (en) | Automatic positioning machining method for opposite-impact air film holes of guide blades | |
| CN116909208B (en) | Shell processing path optimization method and system based on artificial intelligence | |
| CN116309556B (en) | Steel member finished product quality management method based on machine vision technology | |
| CN117058411A (en) | Method, device, medium and equipment for identifying edge appearance flaws of battery | |
| Huang et al. | A global-to-local registration and error evaluation method of blade profile lines based on parameter priority | |
| CN116721144A (en) | Cone hole size measurement method based on point cloud slicing | |
| CN115578363A (en) | Butt weld seam deviation and welding interruption judgment method based on three-dimensional point cloud |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |