Search

CN-121798319-B - Machining method suitable for large-scale marine propeller blade root

CN121798319BCN 121798319 BCN121798319 BCN 121798319BCN-121798319-B

Abstract

The invention discloses a processing method suitable for a blade root of a large-sized marine propeller. Firstly, fitting a positioning reference on a machine tool by using a standard ball and a calibration needle, and establishing a workpiece coordinate system and a conversion relation. And identifying the number of blades based on the propeller point cloud and planning an optimal photographing path. And integrating the high-resolution visual sensor into a machine tool spindle, and collecting original three-dimensional point cloud data of each part of the blade root along the path. And converting the original point cloud into a workpiece coordinate system, and directly performing data splicing based on absolute coordinates of a machine tool to obtain the complete three-dimensional point cloud of the blade root. And registering the point cloud with the theoretical model, calculating machining allowance, generating a machining track, and finally finishing machining of the blade root. The invention multiplexes the high-precision processing machine tool in measurement, realizes the blind zone-free high-resolution scanning of the narrow area of the blade root by utilizing the micron-level positioning precision of the high-precision processing machine tool, avoids the accumulation of matching errors by directly splicing frames based on machine tool coordinates, achieves the sub-millimeter-level splicing precision, and has high calculation efficiency and strong robustness.

Inventors

  • LI WENJUN
  • LI REN
  • GUO JIANHUA
  • GE JIANG
  • HU CHANGJIANG
  • QU SHAOJIE

Assignees

  • 大连誉洋工业智能有限公司

Dates

Publication Date
20260505
Application Date
20260309

Claims (8)

  1. 1. The processing method suitable for the blade root of the large-scale marine propeller is characterized by comprising the following specific steps of: s1, installing a screw propeller and a standard ball on a rotating table of a B axis of a machine tool, installing a high-precision calibration needle on a main shaft A, determining a positioning reference of the screw propeller based on the positions of the standard ball and the high-precision calibration needle through a fitting algorithm, and constructing a workpiece coordinate system and a conversion relation between the workpiece coordinate system and the machine tool coordinate system; s2, acquiring propeller point cloud data, and identifying the number of blades based on the propeller point cloud data; S3, manufacturing a single-blade photographing template, and obtaining an optimal photographing path according to the single-blade photographing template and the number of blades; integrating high-resolution visual sensors on a main shaft A and a main shaft B of a machine tool, and enabling the high-resolution visual sensors to acquire original three-dimensional point cloud data of each part of a blade root according to the optimal photographing path; S4, converting original three-dimensional point cloud data of each part of the blade root into point cloud data under a machine tool coordinate system based on the positioning reference, and determining the point cloud data of each part of the blade root under a workpiece coordinate system based on the conversion relation; S5, splicing point cloud data of all parts of the blade root under a workpiece coordinate system, so that complete three-dimensional point cloud data of the blade root are obtained; S6, acquiring a propeller theoretical model, and registering the complete three-dimensional point cloud data of the blade root with the propeller theoretical model; s7, calculating machining allowance of the blade root based on a registration result, generating a machining track of the blade root based on the machining allowance of the blade root, and finishing machining of a blade root area according to the machining track; the specific steps for calculating the machining allowance of the blade root based on the registration result comprise: S71, determining a normal vector of the theoretical point cloud by adopting a principal component analysis method, and correcting the normal vector, wherein the method comprises the following steps: extracting any point in theoretical point cloud Is a local neighborhood point cloud; calculating covariance matrix of local neighborhood point cloud; performing feature decomposition on the covariance matrix to obtain an orthogonal matrix Wherein Describing the principal direction of the neighborhood point cloud distribution, i.e , Description and drawings Orthogonal directions of the next largest distribution, i.e , Description and drawings 、 All in orthogonal directions, i.e. ; Determining a normal vector: for the upper surface of the propeller, normal vector ; For the lower surface of the propeller, normal vector ; Setting a principal direction constraint to correct the normal vector, comprising: Setting the main direction of the root area ; Calculation method vector The included angle between the main direction of the blade root area and the main direction of the blade root area indicates that the normal vector direction accords with the constraint of the main direction of the blade root area if the included angle is smaller than or equal to the set angle threshold value, the normal vector is reserved and used as the corrected normal vector, and if the included angle is larger than the set angle threshold value, the normal vector is determined to be Projected to the main direction of the root area Normalization processing is carried out to obtain a corrected normal vector; S72, in dots As a starting point, along the corrected normal vector Searching the actual corresponding points: for the point of the upper surface of the blade root, the edge Corrected normal vector In the actual point cloud, searching all points in the set threshold range near the ray as candidate points, and calculating all candidate points and Finding the closest point to the theoretical point in the ray direction as the distance to the theoretical point Corresponding actual measurement points ; For the point of the lower surface of the blade root, the edge Corrected normal vector thereof Searching for all points within a set threshold range around the ray in the actual point cloud as candidate points, and calculating all candidate points and Finding the closest point to the theoretical point in the ray direction as the distance to the theoretical point Corresponding actual measurement points ; S73, calculating single-point machining allowance: For point pairs Calculating the machining allowance The method comprises the following steps: ; Wherein, the Representation of And A linear distance therebetween; s74, calculating the machining allowance of the full point cloud: traversing all points in the theoretical point cloud model, executing S72 and S73, and generating a machining allowance value set corresponding to the theoretical points one by one ; S75, constructing a machining allowance distribution map: each theoretical point Coordinates and remaining magnitude of (2) And (3) correlating to form a three-dimensional allowance distribution mapping relation of the blade root, and further obtaining the machining allowance of each point.
  2. 2. The method for processing the blade root of the large-sized marine propeller according to claim 1, wherein in S1, determining the positioning reference of the propeller based on the positions of the standard ball and the high-precision calibration needle by a fitting algorithm comprises: The spindle A is placed at a position of 0 degrees, the spindle B is controlled to rotate according to a set angle, and the following steps are executed under each angle: s11, acquiring point cloud data of a standard ball under each rotation angle under a camera coordinate system; s12, moving the spindle A to enable the needle tip of the high-precision calibration needle to contact a preset contact point on the standard ball, and acquiring machine tool coordinates of the contact point under a machine tool coordinate system, namely real coordinates of a contact point; s13, adding the real machine tool coordinates of the contact point and the radius of the standard ball to obtain the real machine tool coordinates of the sphere center of the standard ball under the machine tool coordinate system under each rotation angle The machine tool coordinate comprises X, Y, Z, A axis and B axis positions, wherein k represents a rotation angle; The A shaft is respectively arranged at a position of 0 degree and a set angle, the A main shaft is moved to enable the high-precision calibration needle to touch the touch point of the standard ball, and machine tool coordinates under all angles are recorded; filtering and denoising the point cloud data of the standard ball under the camera coordinate system to obtain standard point cloud data; Fitting the standard point cloud data based on a least square method to determine the spherical center coordinates under a camera coordinate system, the rotation center OB of the B-axis rotating table and the rotation center OA of the A-axis rotating table, wherein the method comprises the following steps: fitting the standard point cloud data under each rotation angle by a least square method to obtain the coordinates of the sphere center of the standard sphere under each rotation angle under a camera coordinate system ; Fitting the spherical center coordinates by a least square method to obtain a rotation center OB of the B-axis rotating table; Connecting machine tool coordinates acquired by the position of the A shaft at 0 degree with the position of the set angle, and making a line segment perpendicular bisector, wherein the intersection point of the line segment perpendicular bisector and the rotation axis of the A shaft is the rotation center OA of the A main shaft; Based on And The method comprises the following steps of constructing a hand-eye matrix, namely a positioning reference, so as to convert a camera coordinate system and a machine tool coordinate system according to the hand-eye matrix: By SVD decomposition algorithm Decomposing to obtain optimal rotation matrix and relative translation ; Adding the coordinates of OB and OA with the relative translation to obtain a translation vector; Combining the rotation matrix and the translation vector forms the final 4×4 homogeneous transformation matrix, i.e., the hand-eye matrix, as follows: ; Where R represents a rotation matrix and t represents a relative translation.
  3. 3. The method for machining a blade root of a large marine propeller according to claim 2, wherein constructing the workpiece coordinate system and the conversion relation between the workpiece coordinate system and the machine tool coordinate system comprises: Establishing a workpiece coordinate system by taking a B-axis rotation center OB as an origin of the workpiece coordinate system; the conversion relation between the coordinate system of the workpiece and the coordinate system of the machine tool is constructed as follows: ; Wherein, the Is the coordinates of a point in the machine coordinate system, As the coordinates of the point in the object coordinate system, Is the coordinate of the origin of the object coordinate system in the machine coordinate system.
  4. 4. The method for processing a propeller blade root for a large ship according to claim 3, wherein in S2, a large-field camera is used to collect propeller point cloud data, and the number of blades is identified based on the propeller point cloud data, and the specific steps include: S21, performing voxel grid downsampling processing on the propeller point cloud data: Dividing the three-dimensional space of the propeller point cloud data into a plurality of uniform grids according to the size of a preset voxel grid, taking the gravity center of a point set formed by points in the propeller point cloud data in each grid as the gravity center point of the current grid, and replacing all the points in each grid with the gravity center point to obtain down-sampled point cloud data; S22, extracting an effective point cloud region only related to the blade from the down-sampled point cloud data by adopting a self-adaptive ROI extraction algorithm; S23, blade point cloud segmentation: dividing points with the same normal direction and curvature in an effective point cloud area into the same subarea by adopting an area growth segmentation algorithm, so as to divide the effective point cloud area into two independent blade point cloud sets; S24, carrying out blade edge point cloud extraction processing on the two segmented blade point clouds by adopting an edge detection algorithm to obtain edge point cloud data of the two blades, wherein the edge point cloud data comprise areas with discontinuous surface and sharp curvature change in the point clouds and physical boundaries, edges or surface intersecting lines of the corresponding blades in the areas; s25, point cloud registration calculation: coarse registration, namely respectively calculating barycentric coordinates of two blade edge point clouds, obtaining an initial rigid body transformation matrix in a barycentric alignment mode, and carrying out preliminary alignment on the two blade edge point clouds based on the initial rigid body transformation matrix; fine registration, namely minimizing the distance between corresponding points of two blade edge point clouds by adopting an ICP algorithm, so as to carry out iterative optimization on the initial rigid body transformation matrix to obtain an optimal transformation matrix, and carrying out further accurate alignment on the two blade edge point clouds based on the optimal transformation matrix; s26, judging the number of blades: Extracting a rotation matrix in the optimal transformation matrix, and converting the rotation matrix into a rotation angle value through a conversion algorithm of quaternion conversion Euler angles; and judging the number of the blades according to the rotation angle value.
  5. 5. The method for processing the blade root of the large-sized marine propeller according to claim 4, wherein in S3, the specific step of manufacturing a single-blade photographing template and obtaining an optimal photographing path according to the single-blade photographing template and the number of blades comprises: s31, manufacturing a single-blade photographing template: Acquiring a moving path when the large-view camera collects propeller point cloud data, and acquiring a photographing template of a single blade based on the moving path, wherein the photographing template comprises a space moving track, a gesture and each point position photographing parameter for photographing; S32, determining a reference blade: setting the angle of the physical zero point direction of the turntable in a workpiece coordinate system to be 0 degree, and respectively calculating the clockwise angle difference between the direction angle of each blade and the physical zero point of the turntable; Comparing the magnitudes of all the clockwise angle differences, and selecting the blade with the smallest clockwise angle difference as the reference blade closest to the physical zero point of the turntable; s33, generating a photographing path of the reference blade: Calculating a transformation matrix between the reference blade and the photographing template by adopting a coarse registration method and a fine registration method, and obtaining a corner between the reference blade and the photographing template by converting a quaternion into an Euler angle; Converting the rotation angle between the reference blade and the photographing template into a conversion matrix, and converting the photographing template based on the conversion matrix to obtain a photographing path of the reference blade; S34, multi-blade photographing path expansion: Performing rotation transformation on the photographing path of the reference blade according to the rotation angle to obtain the photographing path of the Nth blade, And further obtaining the original photographing paths of all the blades, wherein the calculation process of the rotation angle is as follows: Rotation angle = rotation angle between reference blade and photographing template + (N-1) x rotation angle value determined during blade number identification; s35, optimizing the original photographing paths of all the blades: Integrating the photographing points in the original photographing paths of all the blades into a point set, and modeling the access sequence problem of the photographing points as a travel business problem, namely defining the movement cost between any two photographing points S i and S j as the Euclidean distance between the two: d ij =||S i -S j ||; selecting the next photographing point with the nearest Euclidean distance as the next destination by adopting a nearest neighbor greedy algorithm by taking any photographing point as a starting point, thereby obtaining an initial photographing path; The initial photographing path is iteratively improved through a 2-opt local optimization method, namely any two sections of connection in the exchanging path are tried, the change of the total path length before and after exchanging is calculated, the exchanging is executed only when the total distance can be reduced by exchanging, and the photographing point index access sequence which enables the sum of Euclidean distances to be shortest, namely the optimal photographing path, is found through repeated iteration; and controlling the high-resolution vision sensor to perform multi-angle scanning on the blade root according to the optimal photographing path, so as to acquire the original three-dimensional point cloud data of each part of the blade root.
  6. 6. The method for processing a blade root of a large-sized marine propeller according to claim 5, wherein in S5, the specific step of stitching point cloud data under a workpiece coordinate system of each part of the blade root to obtain complete three-dimensional point cloud data of the blade root comprises: s51, directly performing rough stitching on point cloud data of each part of the blade root under the workpiece coordinate system according to the workpiece coordinate system; and S52, performing fine splicing on the coarse splicing result based on an ICP algorithm to obtain complete three-dimensional point cloud data of the blade root.
  7. 7. The method for processing a blade root of a large-sized marine propeller according to claim 6, wherein in S6, the specific step of obtaining a theoretical model of the propeller and registering the complete three-dimensional point cloud data of the blade root with the theoretical model of the propeller comprises: s61, performing point clouding treatment on the propeller theoretical model, namely constructing a CAD model of a blade root, and performing point clouding treatment on the CAD model according to a preset sampling density so as to convert the CAD model into theoretical point cloud data; S62, respectively extracting an actual point cloud, namely ISS key points of a complete three-dimensional point cloud data propeller of a blade root and theoretical point cloud data by using an ISS characteristic point extraction algorithm; Performing FPFH feature matching on the extracted ISS key points, and establishing initial corresponding point pairs; S63, carrying out best fit registration on the initial corresponding point pair based on SVD algorithm, thereby realizing registration of actual acquisition point cloud and theoretical point cloud data, and comprising the following steps: Setting the actual point cloud in the initial corresponding point pair as The theoretical point cloud corresponding point is The mass centers of the actual point cloud and the theoretical corresponding point cloud are calculated as follows: ; ; The covariance matrix is calculated based on the centroids of the two point clouds as follows: ; SVD decomposition is carried out on the covariance matrix to obtain: ; Obtaining an optimal rotation matrix Translation vector The method comprises the following steps: ; ; transforming the optimal rotation matrix and the translation vector into an actual acquisition point cloud for registration to obtain: 。
  8. 8. The method for machining a blade root of a large marine propeller according to claim 7, wherein the specific step of generating a blade root machining trajectory based on the machining allowance of the blade root comprises: S101, acquiring local machining direction parameters based on a theoretical model point cloud, wherein the method comprises the following steps: S1011, initializing a global path direction into a circumferential direction and initializing a scanning direction into a radial direction based on radial and circumferential characteristics of the propeller; S1012, optimizing the global path direction and the scanning direction by using a main curvature analysis method, so that the global path direction is along the local minimum curvature direction, and the scanning direction is along the local maximum curvature direction, and specifically comprising the following steps: Extracting a local point cloud neighborhood: extracting a local point cloud neighborhood around each point to be optimized in the global path direction, wherein the radius of the local point cloud neighborhood is set according to the point cloud density and the characteristic size of the blade; Calculating a local curved surface normal vector and a principal curvature of the local point cloud neighborhood, wherein: The covariance matrix of the local neighborhood point cloud is calculated through principal component analysis, and the feature vector corresponding to the minimum feature value is the local surface normal vector; Calculating the principal curvature: performing quadric surface fitting on the local point cloud neighborhood to obtain a fitted surface equation; Calculating two principal curvatures of point to be optimized based on fitting curved surface equation And And (2) and Respectively correspond to the maximum curvature direction And a minimum curvature direction ; Directional projection and alignment: Will be And Fusing with a global path direction and a scan direction, comprising: projecting the global circumferential direction and the global radial direction to the tangential planes respectively, calculating dot products of the two tangential planes after projection, and taking the obtained two dot product results as a first direction similarity and a second direction similarity respectively; Judging whether the absolute value of the similarity in the first direction is larger than a set similarity threshold value, if so, indicating that the current circumferential direction is consistent with the minimum curvature direction, and keeping the global path direction as the circumferential direction, otherwise, indicating that the curvature characteristic of the current area is obvious, and correcting the global path direction as the minimum curvature direction Judging whether the absolute value of the similarity in the second direction is larger than a set similarity threshold value, if so, indicating that the current radial direction is consistent with the maximum curvature direction, and keeping the scanning direction as the radial direction, otherwise, indicating that the curvature characteristic of the current area is obvious, and correcting the scanning direction as the maximum curvature direction ; S102, carrying out equidistant section division on the machining allowance of the blade root along the optimized global path direction, sequencing and equidistant resampling on the point cloud slices of each section to generate a two-dimensional path point sequence, wherein the method comprises the following steps of: Performing equidistant section division on the machining allowance of the blade root according to preset path intervals along the optimized global path direction to form a series of parallel section planes; For each section plane, calculating the distance between the projection values of all points and the current section reference value Screening out all Points smaller than the cross-section tolerance range form an original cross-section point set; Sequencing the original cross-section point sets of each cross-section plane from small to large along the optimized scanning direction, and then carrying out equidistant resampling on the sequenced original cross-section point sets according to preset point spacing to generate a two-dimensional path point sequence; s103, synthesizing the two-dimensional path point sequence and the normal coordinates of the corresponding section into three-dimensional track points, inserting safety transition points and performing track smoothing processing to generate a three-dimensional smooth track point sequence, wherein the three-dimensional smooth track point sequence comprises the following steps: Combining each two-dimensional path point sequence with the normal coordinate value of the corresponding section to synthesize a three-dimensional space track point; Inserting a safety transition point between the starting point and the end point of two adjacent two-dimensional path point sequences, so as to integrate the three-dimensional path point sequences of all sections to obtain an integrated three-dimensional path point sequence, wherein the safety transition point is formed by lifting the connecting line midpoint of the starting point and the end point to a preset safety height along the corrected normal vector direction; and carrying out moving average or spline curve smoothing treatment on the integrated three-dimensional path point sequence to obtain a final blade root processing track.

Description

Machining method suitable for large-scale marine propeller blade root Technical Field The invention relates to the technical field of precise manufacturing and special machining processes of large complex curved surface parts, in particular to a machining method suitable for a blade root of a large marine propeller. Background The blade root area of the large-scale propeller is positioned in the overlapping and shielding area of the adjacent blades, the physical space is narrow, the sight is blocked, the detection accessibility is seriously insufficient, and systematic blind areas exist in the geometric and surface quality data. Fixed equipment (e.g., three-dimensional machines, laser scanners) and conventional manual or simple tools cannot effectively access the back of the blade root and complex surfaces, resulting in an inability to obtain the full three-dimensional geometry of the area. The existing method mainly relies on manual visual observation, touch perception or using a simple measuring tool to perform local, contact type qualitative or rough quantitative evaluation, and has the following defects: (1) The detection result is greatly influenced by personnel experience and state, has poor repeatability and cannot meet the requirements of high-precision processing and detection; (2) The data is difficult to quantify and normalize, and the subsequent analysis is difficult. In order to obtain relatively complete blade root data, it is sometimes even necessary to employ extreme methods of "piecewise disassembly", or to spend a lot of time manually trying to locate and locally measure. Complete inspection of a large propeller takes up to 1-2 weeks, with the overlap area (with blade root) taking most of the time; (3) Incomplete and non-digital (mostly paper records or qualitative descriptions) detection data of the blade root area cannot be accurately compared with a CAD design model, accurate processing error feedback cannot be established, and complete data which can be used for performance analysis, life prediction and maintenance decision cannot be formed; in conclusion, the insufficient accessibility of the blade root leads to data loss, the lag of the detection method leads to low data quality, the low efficiency leads to difficult frequent and systematic acquisition of data, the three lead to low digitization level of blade root processing and management and control, and the realization of intelligent manufacturing and predictive maintenance is hindered. Disclosure of Invention The invention provides a processing method suitable for a blade root of a large-sized marine propeller so as to overcome the technical problems. In order to achieve the above object, the technical scheme of the present invention is as follows: A processing method suitable for a blade root of a large-sized marine propeller comprises the following specific steps: s1, installing a screw propeller and a standard ball on a rotating table of a B axis of a machine tool, installing a high-precision calibration needle on a main shaft A, determining a positioning reference of the screw propeller based on the positions of the standard ball and the high-precision calibration needle through a fitting algorithm, and constructing a workpiece coordinate system and a conversion relation between the workpiece coordinate system and the machine tool coordinate system; s2, acquiring propeller point cloud data, and identifying the number of blades based on the propeller point cloud data; S3, manufacturing a single-blade photographing template, and obtaining an optimal photographing path according to the single-blade photographing template and the number of blades; integrating high-resolution visual sensors on a main shaft A and a main shaft B of a machine tool, and enabling the high-resolution visual sensors to acquire original three-dimensional point cloud data of each part of a blade root according to the optimal photographing path; S4, converting original three-dimensional point cloud data of each part of the blade root into point cloud data under a machine tool coordinate system based on the positioning reference, and determining the point cloud data of each part of the blade root under a workpiece coordinate system based on the conversion relation; S5, splicing point cloud data of all parts of the blade root under a workpiece coordinate system, so that complete three-dimensional point cloud data of the blade root are obtained; S6, acquiring a propeller theoretical model, and registering the complete three-dimensional point cloud data of the blade root with the propeller theoretical model; and S7, calculating the machining allowance of the blade root based on the registration result, generating a machining track of the blade root based on the machining allowance of the blade root, and finishing machining of the blade root area according to the machining track. Further, in S1, determining, by a fitting algorithm, a positioning reference of th