A method for processing a large marine propeller

Abstract

The application discloses a processing method suitable for large marine propellers and relates to the technical field of precise composite processing of large marine propellers, and comprises the following steps: installing a propeller and a standard ball on a rotating table of a machine tool, installing a high-precision calibration needle on a main shaft, establishing a positioning datum of a workpiece by using a fitting algorithm, constructing a workpiece coordinate system and a conversion relationship between the workpiece coordinate system and a machine tool coordinate system; acquiring point cloud data of each part of the propeller, and converting the point cloud data to the workpiece coordinate system by using coordinate conversion; performing coarse and fine splicing on blade root point cloud, combining data of the rest parts, and forming workpiece coordinate system point cloud data of the propeller; calculating polishing amount by using a normal projection algorithm based on the data; generating a machining track containing blade thickness calibration according to the polishing amount, and milling and polishing the propeller according to the machining track. The method significantly improves positioning accuracy, effectively eliminates clamping errors, is high in polishing amount calculation precision and efficiency and strong in self-adaptability, and significantly improves machining efficiency and shortens a production cycle.

Description

Technical Field

[0001] This invention relates to the field of precision composite machining technology for large marine propellers, and particularly to a machining method suitable for large marine propellers. Background Technology

[0002] The manufacturing of complex components such as large propellers is a key technological aspect of shipbuilding and marine engineering, and its precision and efficiency directly affect the performance and quality of the final product. Currently, the manufacturing methods commonly used in the industry have many bottlenecks in terms of process flow, quality control, and production efficiency.

[0003] Currently, the mainstream manufacturing technologies for propellers are mainly divided into two modes:

[0004] The first type is a discrete, sequential, and remote manufacturing process, which is the most widely used traditional manufacturing process in the industry. Its steps are as follows: First, the blank undergoes initial inspection and rough milling; then, the semi-finished product is lifted off the machine tool and transported to an independent coordinate measuring machine or large laser tracker station for offline inspection, generating an inspection report; next, process engineers analyze the report, manually calculate the allowance deviation, and modify or regenerate the finishing program; subsequently, the workpiece is lifted back onto the machine tool for secondary clamping and alignment, and a new finishing program is executed; finally, the milled workpiece needs to be removed from the production line again, where a team of senior technicians spends several weeks manually grinding and polishing the entire surface to achieve the final surface roughness requirements. Essentially, this mode is a long-cycle, frequently interrupted, and highly dependent on human experience and physical strength discrete sequential process.

[0005] The second type is a limited closed-loop "inspection-milling" mode based on on-machine measurement, with some local improvements. This mode integrates a trigger-type probe or laser scanner into a high-end five-axis CNC machine tool to directly measure local or critical features of the workpiece on the machine tool after rough milling or before finish milling, and automatically updates the toolpath offset based on the measurement results before continuing machining. However, this mode has obvious limitations: its measurement range, accuracy, and efficiency are limited, and it completely lacks the capability for "grinding." The final critical surface finishing process still needs to be completed manually offline, failing to achieve a complete closed-loop process from machining to finishing.

[0006] The aforementioned existing technical solutions, whether in the traditional serial mode or the partially improved mode, all share the following common and prominent technical problems:

[0007] Repeated clamping leads to loss of datum and a break in the machining accuracy chain. Large workpieces are repeatedly hoisted and re-clamped in the "machining-inspection-remachining" cycle. Each operation introduces uncontrollable clamping and positioning errors, causing the machining datum to be lost during inter-process transfer. This "accuracy decay" effect causes errors to accumulate and amplify between different equipment datums, severely restricting the consistency of each blade profile, contour accuracy, and the improvement of overall manufacturing accuracy.

[0008] The production cycle is lengthy, resulting in low overall manufacturing efficiency. The majority (over 60%) of the total manufacturing cycle is spent on multiple lifting and transfers of workpieces between different processes, as well as waiting time in front of inspection stations. This leads to long periods of idle time for core, expensive equipment such as machine tools, resulting in extremely low overall equipment efficiency (OEE). Particularly noteworthy is the grinding and polishing process, which relies entirely on manual labor and has become a major bottleneck in the production flow. Its time-consuming nature and inability to be performed in parallel create a rigid constraint on shortening product delivery cycles.

[0009] The processes are fragmented, creating "data silos." There is a severe disconnect in information flow between processes: offline inspection data reports require manual interpretation and input into the CAM system; while the manual grinding process receives almost no quantifiable data feedback. This fragmentation prevents real-time, two-way data interaction between the milling and grinding processes, hindering collaborative optimization of process parameters at a global level. For example, the milling process cannot know the actual capabilities of the grinding process, potentially leading to unreasonable allowances; and the grinding process, lacking knowledge of the precise milled surface, can only rely on operator experience for trial-and-error operations.

[0010] Over-reliance on scarce manual skills leads to inconsistent product quality. The final finishing quality is highly dependent on the "feel," experience, physical ability, and even working condition of a few highly skilled workers—a classic case of "craftsman dependency." This results in significant differences in the work outcomes of different personnel, and even between different batches by the same person, leading to large fluctuations in product quality and making it difficult to achieve stable, standardized, high-quality output. Furthermore, the tacit experience and skills of craftsmen are difficult to effectively quantify, record, and digitally reproduce, facing the risk of being lost.

[0011] True adaptive machining and end-to-end quality control are impossible. Traditional manufacturing is essentially open-loop machining: the milling process rigidly executes preset programs, making it difficult to adapt to individual initial differences in the workpiece and deformation during machining; manual grinding is an entirely open-loop operation. The entire manufacturing process lacks a real-time perception and feedback mechanism for process conditions such as tool wear, workpiece deformation, and vibration. Problems often only become apparent during final inspection, leading to costly rework or scrap.

[0012] Overall manufacturing costs are high. The lengthy production cycle results in high capital tied up in work-in-process inventory, and the heavy reliance on highly skilled workers significantly increases labor costs. Furthermore, the hidden costs associated with repeated hoisting, such as safety risks, rework and scrap losses due to quality fluctuations, and damage to business reputation caused by delivery delays, cannot be ignored.

[0013] Therefore, using existing methods for propeller clamping and processing results in poor consistency of processing effects, leading to problems such as low positioning accuracy, poor processing efficiency, and unstable quality during propeller processing, making it difficult to develop towards a more intelligent and digital direction. Summary of the Invention

[0014] This invention provides a processing method suitable for large marine propellers to overcome the above-mentioned technical problems.

[0015] To achieve the above objectives, the technical solution of the present invention is as follows:

[0016] A processing method suitable for large marine propellers, comprising:

[0017] S1: Install the propeller and standard ball on the B-axis rotary table of the machine tool, and install the high-precision calibration needle on the A spindle. Use a fitting algorithm to determine the positioning reference of the propeller based on the positions of the standard ball and the high-precision calibration needle, and construct the workpiece coordinate system and the transformation relationship between the workpiece coordinate system and the machine tool coordinate system.

[0018] S2: Acquire point cloud data of the propeller in the camera coordinate system, convert the point cloud data in the camera coordinate system to point cloud data in the machine tool coordinate system based on the positioning reference, and determine the point cloud data of the propeller in the workpiece coordinate system based on the conversion relationship, including:

[0019] S21. Collect the initial point cloud data of the propeller, and obtain the point cloud data of each part of the blade tip, blade surface and blade root based on the initial point cloud data; convert the point cloud data of each part of the blade tip, blade surface and blade root into point cloud data in the machine tool coordinate system based on the positioning reference, and determine the point cloud data of each part of the blade tip, blade surface and blade root in the workpiece coordinate system based on the conversion relationship.

[0020] S22. Perform coarse stitching on the point cloud data of each part of the blade root in the workpiece coordinate system, and perform fine stitching on the coarse stitching result based on the ICP algorithm to obtain complete point cloud data of the blade root in the workpiece coordinate system.

[0021] S23. Collect point cloud data of the propeller hub in the camera coordinate system, convert the point cloud data in the camera coordinate system into point cloud data in the machine tool coordinate system based on the positioning reference, and determine the point cloud data in the workpiece coordinate system based on the conversion relationship.

[0022] S24. Combine the point cloud data of the blade tip, blade surface, hub, and complete blade root in the workpiece coordinate system to form the final point cloud data of the propeller in the workpiece coordinate system.

[0023] S3: Calculate the grinding amount based on the point cloud data of the propeller in the workpiece coordinate system according to the normal projection;

[0024] S4: Generate a machining trajectory that includes blade thickness calibration based on the grinding amount, and perform milling and polishing on the propeller according to the machining trajectory to complete the machining of the propeller.

[0025] Beneficial effects: This invention provides a processing method suitable for large marine propellers, which has the following advantages:

[0026] 1. This invention significantly improves positioning accuracy and eliminates clamping errors: By accurately calibrating the standard ball and calibration needle, and combining the fitting algorithm to determine the propeller positioning reference, the traditional positioning method that relies on manual experience is transformed into a digital and quantitative precise positioning method. The positioning accuracy can reach the micrometer level, fundamentally eliminating the systematic errors introduced by manual clamping and alignment, and laying the foundation for subsequent high-precision processing.

[0027] 2. The grinding amount calculation of the present invention is accurate, efficient and highly adaptable: Based on the precisely aligned workpiece coordinate system point cloud data, the grinding amount is calculated by the normal projection algorithm, which fully considers the geometric characteristics of the complex curved surface of the propeller, and the calculation result is more consistent with the actual physical removal amount.

[0028] 3. This invention significantly improves processing efficiency and shortens the production cycle: The automated positioning and measurement of this invention greatly reduces the time required for manual clamping and alignment. Intelligent trajectory planning based on precise allowance calculation avoids ineffective idle runs and repetitive processing, achieving optimal allocation of material removal. Simultaneously, it enables refined perception of the complex geometry of the blades during processing, taking blade thickness into account, improving processing accuracy and avoiding processing errors caused by curvature changes, ensuring the smoothness and consistency of the blade tip radius transition. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 A schematic diagram illustrating a processing method suitable for large marine propellers provided by the present invention;

[0031] Figure 2A schematic diagram of the propeller clamping and the machine tool;

[0032] Figure 3 This is a full-contour region image of a single blade after scanning in an embodiment of the present invention;

[0033] Figure 4 This is a partial contour region diagram of a single blade after scanning in an embodiment of the present invention;

[0034] Figure 5 This is a schematic diagram of the blade support processing in an embodiment of the present invention;

[0035] Figure 6 This is a schematic diagram of the propeller hub in an embodiment of the present invention. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0037] This embodiment provides a processing method suitable for large marine propellers, such as... Figure 1 As shown, it includes:

[0038] S1: Install the propeller and standard ball on the B-axis rotary table of the machine tool, and install the high-precision calibration needle on the A spindle. Use a fitting algorithm to determine the positioning reference of the propeller based on the positions of the standard ball and the high-precision calibration needle, and construct the workpiece coordinate system and the transformation relationship between the workpiece coordinate system and the machine tool coordinate system.

[0039] S2: Acquire point cloud data of the propeller in the camera coordinate system, convert the point cloud data in the camera coordinate system to point cloud data in the machine tool coordinate system based on the positioning reference, and determine the point cloud data of the propeller in the workpiece coordinate system based on the conversion relationship, including:

[0040] S21. Collect the initial point cloud data of the propeller, and obtain the point cloud data of each part of the blade tip, blade surface and blade root based on the initial point cloud data; convert the point cloud data of each part of the blade tip, blade surface and blade root into point cloud data in the machine tool coordinate system based on the positioning reference, and determine the point cloud data of each part of the blade tip, blade surface and blade root in the workpiece coordinate system based on the conversion relationship.

[0041] S22. Perform coarse stitching on the point cloud data of each part of the blade root in the workpiece coordinate system, and perform fine stitching on the coarse stitching result based on the ICP algorithm to obtain complete point cloud data of the blade root in the workpiece coordinate system.

[0042] S23. Collect point cloud data of the propeller hub in the camera coordinate system, convert the point cloud data in the camera coordinate system into point cloud data in the machine tool coordinate system based on the positioning reference, and determine the point cloud data in the workpiece coordinate system based on the conversion relationship.

[0043] S24. Combine the point cloud data of the blade tip, blade surface, hub, and complete blade root in the workpiece coordinate system to form the final point cloud data of the propeller in the workpiece coordinate system.

[0044] S3: Calculate the grinding amount based on the point cloud data of the propeller in the workpiece coordinate system according to the normal projection;

[0045] S4: Generate a machining trajectory that includes blade thickness calibration based on the grinding amount, and perform milling and polishing on the propeller according to the machining trajectory to complete the machining of the propeller.

[0046] In a specific embodiment, the propeller and a standard ball are mounted on the B-axis rotary table of the machine tool, and a high-precision calibration needle is mounted on the A spindle. A fitting algorithm is used to determine the propeller's positioning reference based on the positions of the standard ball and the high-precision calibration needle. The scheme for constructing the workpiece coordinate system and the transformation relationship between the workpiece coordinate system and the machine tool coordinate system is as follows:

[0047] I. Installing the workpiece and calibration device

[0048] The propeller is manually clamped onto the B-axis rotary table, and the linear laser sensor is mounted above the A-axis spindle. A standard sphere with a known radius is fixed on the B-axis rotary table, ensuring its position remains stable as the table rotates. A high-precision calibration pin, whose tip position has been calibrated using a machine tool probe, is installed on the A-axis spindle. The overall clamping is as follows: Figure 2 As shown;

[0049] II. Determining the propeller's positioning reference based on the positions of a standard sphere and a high-precision calibration needle using a fitting algorithm, including:

[0050] Set spindle A to the 0-degree position, and control the B-axis rotary table to rotate at the set angle. Perform the following steps at each angle:

[0051] Step 1: Obtain the point cloud data of the standard sphere in the camera coordinate system at various rotation angles:

[0052] The B-axis rotary table is controlled to rotate according to the set rotation angle. At each rotation angle, the standard sphere on the B-axis rotary table is scanned by the line scan laser sensor to obtain the point cloud data of the standard sphere in the camera coordinate system at each angle.

[0053] Step 2: Move spindle A until the tip of the high-precision calibration needle contacts the preset contact point on the standard ball, and obtain the machine coordinates of the contact point in the machine coordinate system, that is, the true coordinates of the contact point:

[0054] With the A-axis at 0 degrees, at each set rotation angle, move the A spindle so that the tip of the high-precision calibration needle contacts the preset contact point on the standard ball, and obtain the machine coordinates of the contact point in the machine coordinate system, that is, the true coordinates of the contact point.

[0055] In this scheme, the B-axis rotary table is controlled to rotate sequentially to 0°, 90°, 180°, and 270°. At each rotation angle, the A-axis drives the line-scanning laser sensor to scan the standard sphere at a constant speed to obtain the point cloud data of the standard sphere in the camera coordinate system.

[0056] Step 3: Add the actual machine coordinates of the contact point to the radius of the standard sphere to obtain the actual machine coordinates of the center of the standard sphere in the machine coordinate system at each rotation angle. The actual machine tool coordinates include the X, Y, Z, A-axis, and B-axis positions; k represents the rotation angle.

[0057] Place the A-axis at 0 degrees and at a set angle, move the A-spindle until the high-precision calibration pin touches the contact point of the standard ball, and record the machine tool coordinates at each angle.

[0058] In this design, spindle A is rotated 90 degrees.

[0059] The point cloud data of the standard sphere in the camera coordinate system is filtered and denoised to obtain standard point cloud data:

[0060] The standard sphere is fixed to a B-axis rotary table. The point cloud data filtering and denoising mainly removes the point cloud data of the fixed rod to ensure the coordinate accuracy of the subsequent point cloud fitting.

[0061] The standard point cloud data is fitted using the least squares method to determine the coordinates of the sphere center in the camera coordinate system, the rotation center OB of the B-axis rotary table, and the rotation center OA of the A-axis. The steps are as follows:

[0062] Step 1: Fit the standard point cloud data at each rotation angle using the least squares method to obtain the coordinates of the center of the standard sphere in the camera coordinate system at each rotation angle. ;

[0063] Step 2: Fit the coordinates of the sphere center using the least squares method to obtain the rotation center OB of the B-axis rotary table;

[0064] Step 3: Connect the machine tool coordinates acquired at the 0-degree position and the set angle position of the A-axis, and draw the perpendicular bisector of the line segment. The intersection of this perpendicular bisector and the rotation axis of the A-axis is the rotation center OA of the A-spindle.

[0065] The specific coordinates of OA are obtained through geometric principles and axis constraints. This is a conventional geometric solution, and those skilled in the art can know how to construct equations for calculation. Therefore, the specific solution process will not be explained in detail.

[0066] based on and Construct a hand-eye matrix, i.e., a positioning reference, to perform the transformation between the camera coordinate system and the machine tool coordinate system based on the hand-eye matrix. The steps are as follows:

[0067] Step 1: Perform SVD decomposition algorithm on... By decomposing the matrix, the optimal rotation matrix and relative translation can be obtained. ;

[0068] Step 2: Add the coordinates of OB and OA to the relative translation to obtain the complete translation vector;

[0069] Step 3: Combine the rotation matrix and translation vector to form the final 4×4 homogeneous transformation matrix, i.e., the hand-eye matrix, as follows:

[0070]

[0071] Where R represents the rotation matrix and t represents the relative translation;

[0072] Establish the workpiece coordinate system and the transformation relationship between the workpiece coordinate system and the machine tool coordinate system, including:

[0073] Establish the workpiece coordinate system with the rotation center OB of axis B as the origin of the workpiece coordinate system;

[0074] Step 4: Establish the transformation relationship between the workpiece coordinate system and the machine tool coordinate system as follows:

[0075]

[0076] in, Let the coordinates of the point be in the machine tool coordinate system. Let be the coordinates of the point in the workpiece coordinate system. The coordinates of the origin of the workpiece coordinate system in the machine tool coordinate system.

[0077] In a specific embodiment, the scheme for acquiring point cloud data of the propeller in the camera coordinate system, converting the point cloud data in the camera coordinate system into point cloud data in the machine tool coordinate system based on the positioning reference, and determining the point cloud data of the propeller in the workpiece coordinate system based on the conversion relationship is as follows:

[0078] S21. Acquire initial point cloud data of the propeller, and obtain point cloud data for each part of the blade tip, blade surface, and blade root based on the initial point cloud data; convert the point cloud data of each part of the blade tip, blade surface, and blade root into point cloud data in the machine tool coordinate system based on the positioning reference, and determine the point cloud data of each part of the blade tip, blade surface, and blade root in the workpiece coordinate system based on the conversion relationship, including:

[0079] S211: The number of blades is identified based on the initial point cloud data;

[0080] Step 1: Perform voxel mesh downsampling on the initial point cloud data:

[0081] The three-dimensional space of the initial point cloud data is divided into several uniform grids according to a preset voxel grid size, and the centroid of the point set formed by the points in the initial point cloud data in each grid is taken as the centroid of the current grid, thus obtaining the downsampled point cloud data.

[0082] Specifically, by downsampling, the amount of point cloud data can be significantly reduced while keeping the propeller geometry unchanged, thereby improving the speed of subsequent data processing.

[0083] An adaptive ROI extraction algorithm was used to extract effective point cloud regions that are only related to the blades from the downsampled point cloud data.

[0084] Specifically, the adaptive ROI extraction algorithm can remove interfering point clouds (such as reflection points from environmental debris) in the vertical direction of the camera and the point cloud data of the turntable itself, retaining only the effective point cloud area related to the blades, thus avoiding irrelevant data from interfering with subsequent segmentation and registration.

[0085] Leaf point cloud segmentation:

[0086] The region growing segmentation algorithm is used to divide the effective point cloud region into two independent leaf point cloud sets by dividing points with the same normal direction and curvature into the same sub-region.

[0087] Specifically, this embodiment uses the local geometric features (normal direction and curvature) of the point cloud as the judgment basis. Starting from the preset seed point, through iterative calculation, it gradually merges adjacent points with the same normal direction and curvature into the same region, and finally divides the effective point cloud into two independent leaf point cloud sets, realizing the separation of the leaf point cloud from other region point clouds.

[0088] An edge detection algorithm is used to extract the edge point cloud data of the two segmented blade point clouds. The edge point cloud data includes regions with surface discontinuities and sharp curvature changes in the point cloud, as well as the physical boundaries, edges, or surface intersections of the blades corresponding to these regions.

[0089] Point cloud registration calculation:

[0090] Coarse registration: Calculate the centroid coordinates of the point clouds at the edges of the two blades respectively, obtain the initial rigid body transformation matrix by aligning the centroids, and perform preliminary alignment of the point clouds at the edges of the two blades based on the initial rigid body transformation matrix to provide a good initial estimate for fine registration;

[0091] Fine registration: The ICP algorithm is used to minimize the distance between corresponding points on the edge point clouds of two blades, thereby iteratively optimizing the initial rigid body transformation matrix to obtain the optimal transformation matrix, and further aligning the edge point clouds of two blades based on the optimal transformation matrix.

[0092] Leaf count determination:

[0093] Extract the rotation matrix from the optimal transformation matrix, and convert the rotation matrix into rotation angle values ​​using a quaternion-to-Euler angle conversion algorithm;

[0094] The number of blades is determined based on the rotation angle value;

[0095] Specifically, if the rotation angle is around 90 degrees, the propeller is determined to be a four-bladed propeller; if the rotation angle is around 72 degrees, the propeller is determined to be a five-bladed propeller. The identification result is stored as a key parameter for subsequent image path planning and processing trajectory generation.

[0096] S212: Create a single-leaf photography template, obtain the optimal photography path based on the single-leaf photography template and the number of leaves, and collect the original 3D point cloud data of the leaves according to the optimal photography path; such as Figure 3 and Figure 4 As shown; the leaf blade includes the leaf tip, leaf surface, and leaf root area:

[0097] Create a single-leaf photo template:

[0098] The movement path when collecting propeller point cloud data is obtained, and a single blade photograph template is obtained based on the movement path. The photograph template includes the spatial movement trajectory, attitude and shooting parameters of each point.

[0099] In this solution, the moving path of the large field-of-view camera when collecting propeller point cloud data is obtained, and a single blade image template is obtained based on the moving path. The image template includes the spatial moving trajectory (coordinate sequence), attitude, and shooting parameters at each point.

[0100] Determine the reference blade:

[0101] Set the physical zero point of the turntable to 0 degrees in the world coordinate system, and calculate the clockwise angle difference between the direction angle of each blade and the physical zero point of the turntable.

[0102] Compare the magnitudes of all clockwise angle differences and select the blade with the smallest clockwise angle difference as the reference blade closest to the physical zero point of the turntable;

[0103] The image path for generating the reference blade:

[0104] The transformation matrix between the reference blade and the photographic template is calculated using coarse and fine registration methods, and the rotation angle between the reference blade and the photographic template is obtained by converting quaternions to Euler angles.

[0105] The angle between the reference blade and the image template is converted into a transformation matrix. The image template is then transformed based on the transformation matrix to obtain the image path of the reference blade.

[0106] Specifically, for the reverse side of the reference blade, keep the XY coordinates of the shooting point completely consistent, and rotate the Z axis by 180 degrees to obtain the shooting path of the reverse side of the reference blade, ensuring that both the front and back sides of the blade can be completely scanned.

[0107] In this embodiment, all transformation matrices contain only rotational transformations, and the translation component is set to zero. Since the blades rotate around the center of the turntable and are placed horizontally on the turntable, their vertical (Z-axis) position remains unchanged, ensuring the accuracy of path expansion.

[0108] Multi-leaf photography path expansion:

[0109] The Nth blade is obtained by rotating the photographing path of the reference blade according to the rotation angle. The image path is then used to obtain the original image path of all the blades. The calculation process for the rotation angle is as follows:

[0110] Rotation angle = rotation angle between the reference blade and the photo template + (N-1) × rotation angle value determined during the blade number recognition process; for example, the rotation angle between adjacent blades of a four-bladed propeller is 90 degrees, and that of a five-bladed propeller is 72 degrees.

[0111] Optimize the original image paths for all leaves:

[0112] The original photography paths of all leaves are integrated into a single point set. The problem of the order of visiting photography points is modeled as a traveling salesman problem, i.e., any two photography points are defined... and The cost of moving between them is the Euclidean distance between them:

[0113]

[0114] Starting from any point, the nearest neighbor greedy algorithm is used to select the next point with the closest Euclidean distance as the next destination, thus obtaining an initial feasible path;

[0115] The initial feasible path is iteratively improved by the 2-opt local optimization method: try to swap any two segments in the path, calculate the change in the total path length before and after the swap, and only perform the swap if the swap can reduce the total distance. By iteratively finding the optimal photo-taking path, the order of accessing the photo-taking point indexes that minimizes the sum of Euclidean distances, is the optimal photo-taking path.

[0116] Based on the optimal image acquisition path, the camera movement is controlled to acquire the original 3D point cloud data of each leaf, ensuring the efficiency and completeness of data acquisition.

[0117] In this scheme, the control line-scanning laser sensor performs multi-angle scanning of the leaf tip, leaf surface and leaf root according to the optimal imaging path, thereby collecting the original three-dimensional point cloud data of each part of each leaf.

[0118] Specifically, since the leaf root region is typically a deep cavity structure, ordinary three-axis scanning (moving only XYZ) is easily obstructed by the leaf, resulting in blind spots. In this embodiment, by utilizing the rotation of the A-axis and B-axis, the scanning head can flexibly change its orientation, thereby bypassing obstructions and directly illuminating the inner wall of the leaf root. Specifically, the A-axis (usually the main rotation axis) is responsible for a wide range of orientation adjustments. For example, the scanning head can be positioned to enter the deep cavity from above the leaf, or the pitch angle of the scanning head can be adjusted so that its beam is perpendicular to the inner wall of the leaf root. The B-axis (usually the auxiliary rotation axis) is responsible for fine-tuning the orientation of the scanning head, working in conjunction with the A-axis to achieve omnidirectional coverage of the scanning head within the deep cavity, ensuring no blind spots. In this embodiment, to ensure subsequent stitching accuracy, the point cloud data from two adjacent scans are set to have at least 30% overlap.

[0119] S213: Obtain the curvature of the leaf tip and leaf surface regions based on the original 3D point cloud data of the leaf tip and leaf surface. Optimize the original 3D point cloud data of the leaf tip and leaf surface using an adaptive downsampling strategy based on the curvature of these regions to obtain the final point cloud data of the leaf tip and leaf surface, including:

[0120] Calculate any point in the original 3D point cloud data for each leaf tip and leaf surface. The local curvature includes:

[0121] by Set the radius around the center. Select all points within this radius to form a neighborhood. N i ;

[0122] Computing the neighborhood N iThe center of gravity, the formula is:

[0123]

[0124] in, For the neighborhood N i The number of points within;

[0125] The formula for calculating the covariance matrix is:

[0126]

[0127] For covariance matrix Eigenvalue decomposition yields three eigenvalues. , and ,and ;

[0128] calculate p i The local curvature estimate is given by the formula:

[0129]

[0130] Curvature grading:

[0131] Preset curvature threshold and The point cloud is divided into three regions based on a preset curvature threshold:

[0132] Will The area is set as a high curvature region; such as the leaf edge, leaf root, and other areas with drastic curvature changes.

[0133] Will The region is set to a medium curvature region;

[0134] Will The region is set as a low curvature region; such as the flat surface region of a blade.

[0135] Point cloud adaptive downsampling:

[0136] Based on the curvature grading results, the original 3D point cloud data of the leaf tip and leaf surface were downsampled using different voxel sizes to obtain optimized point cloud data for the leaf tip and leaf surface, where:

[0137] Downsampling of high curvature regions was performed using small voxel sizes of 0.1 × 0.1 × 0.1 mm³ to ensure preservation of key geometric features;

[0138] Downsampling of medium curvature regions is performed using medium voxel sizes of 0.3×0.3×0.3 mm³ to balance accuracy and efficiency;

[0139] Downsampling of low curvature regions is performed using a large voxel size of 0.75×0.75×0.75 mm³ to reduce data redundancy;

[0140] Specifically, by adaptive downsampling of point clouds, the optimization effect is achieved with the accuracy loss of key feature regions not exceeding 0.05mm and the overall data volume reduced by 40%-60%.

[0141] S214: Convert the final point cloud data of the blade tip and blade surface and the original three-dimensional point cloud data of each part of the blade root into point cloud data in the machine tool coordinate system based on the positioning reference, and determine the point cloud data of each part of the blade tip, blade surface and blade root in the workpiece coordinate system based on the conversion relationship.

[0142] S22. Perform coarse stitching on the point cloud data of each part of the blade root in the workpiece coordinate system, and perform fine stitching on the coarse stitching result based on the ICP algorithm to obtain complete point cloud data of the blade root in the workpiece coordinate system.

[0143] In this scheme, the point cloud data of each part of the blade root in the workpiece coordinate system are coarsely stitched together directly. At this time, due to mechanical errors and cumulative errors, the point cloud may be misaligned or have ghosting. This embodiment uses the ICP algorithm to finely stitch the coarse stitching result (using the coarse stitching result as the excellent initial value of the ICP algorithm). Utilizing the 30% overlap area of ​​the point cloud data from two adjacent scans, the ICP algorithm automatically finds the best matching position between adjacent point clouds. Through iterative calculation, the error of the overlap area is minimized, thereby obtaining the complete three-dimensional point cloud data of the blade root.

[0144] S23. Acquire point cloud data of the propeller hub in the camera coordinate system, convert the point cloud data in the camera coordinate system into point cloud data in the machine tool coordinate system based on the positioning reference, and determine the point cloud data in the workpiece coordinate system based on the conversion relationship:

[0145] In this scheme, a linear laser sensor is controlled to scan the propeller hub, such as... Figure 6 As shown, point cloud data of the propeller hub in the camera coordinate system is collected. Based on the positioning datum, the point cloud data in the camera coordinate system is converted into point cloud data in the machine tool coordinate system. Based on the conversion relationship, the point cloud data in the workpiece coordinate system is determined.

[0146] The rotor hub is scanned from multiple angles using a line-scan laser sensor to obtain complete surface point cloud data.

[0147] S24. Combine the point cloud data of the blade tip, blade surface, hub, and complete blade root in the workpiece coordinate system to form the final point cloud data of the propeller in the workpiece coordinate system.

[0148] In this solution, all point cloud data undergo point cloud preprocessing and noise reduction:

[0149] A statistical outlier removal method is adopted to eliminate noise points; a voxel grid filter is used to uniformly downsample the point cloud, which reduces computational complexity while preserving geometric features.

[0150] Statistical outlier removal is a denoising algorithm based on statistical analysis of the local neighborhood of point clouds. The neighborhood distances of normal points follow a normal distribution, while the neighborhood distances of outliers deviate significantly from this distribution. The algorithm removes the detected outliers from the original point cloud. Voxel mesh downsampling divides the three-dimensional space into uniform grids (voxels) and replaces all points within each voxel with a representative point (usually the centroid), thereby significantly reducing the amount of point cloud data while preserving geometric features. Statistical outlier removal methods and voxel mesh filters are common techniques in point cloud processing. Those skilled in the art know how to use these two algorithms to process point cloud data, so they will not be elaborated upon further.

[0151] The preprocessed point cloud data is converted from the camera coordinate system to the machine tool coordinate system using a hand-eye matrix. Then, based on the transformation relationship between the machine tool coordinate system and the workpiece coordinate system, the point cloud data in the machine tool coordinate system is converted to the point cloud data in the workpiece coordinate system.

[0152] The present invention has the following advantages:

[0153] 1. By automatically identifying the number of blades using point cloud data, the processing equipment can accurately identify the position of each blade, avoiding missed or repeated processing, and greatly improving the accuracy and reliability of the processing. By identifying the curvature changes in different areas of the blade, a refined perception of the complex geometry of the blade tip is achieved, thereby optimizing data processing efficiency while ensuring processing accuracy, avoiding processing errors caused by curvature changes, and ensuring the smoothness and consistency of the blade tip rounded corner transition;

[0154] 2. A new paradigm for deep cavity inspection using machine tool spindles equipped with vision sensors:

[0155] Traditional detection methods are limited by robot precision or the field of view of external cameras, making it impossible to fully acquire three-dimensional data of the propeller's deep cavity region. This invention innovatively integrates a vision sensor into the machine tool spindle, directly reusing high-precision machining equipment as a high-precision measuring device. Through the inherent micron-level positioning accuracy of the machine tool, it achieves blind-spot-free, high-resolution scanning of narrow areas such as the blade root and tenon groove.

[0156] 3. High-precision direct stitching mechanism for point clouds based on absolute coordinates of machine tools:

[0157] Unlike traditional point cloud stitching methods that rely on feature matching (which are prone to failure in areas with weak texture), this method proposes a direct stitching framework based on machine tool absolute coordinates. By utilizing high-precision readings from the machine tool coordinate system, the spatial transformation between point clouds in each frame is directly calculated, fundamentally avoiding the problem of matching error accumulation. This achieves sub-millimeter-level stitching accuracy and is computationally efficient and robust.

[0158] 4. Point cloud stitching mechanism directly driven by machine tool coordinate system:

[0159] Abandoning traditional feature-matching-based point cloud stitching methods, this method proposes a direct stitching framework based on the absolute coordinate system of a machine tool. By utilizing high-precision encoder readings from each axis of the machine tool, the spatial transformation matrix of each frame of point cloud is directly calculated, fundamentally eliminating the accumulation of matching errors and achieving sub-millimeter-level stitching accuracy, which is especially suitable for weakly textured and highly reflective metal surfaces.

[0160] In a specific embodiment, the scheme for calculating the grinding amount based on the point cloud data of the propeller in the workpiece coordinate system using normal projection is as follows:

[0161] The grinding amount is calculated based on the point cloud data in the workpiece coordinate system using normal projection, including:

[0162] The theoretical model is processed into point cloud data: a CAD model of the propeller is constructed, and the CAD model is processed into point cloud data according to a preset sampling density, which is converted into theoretical point cloud data while preserving the complete surface features;

[0163] This solution uses the STEPCAFControl_Reader and XCAFDoc_ShapeTool functions of Opencascade to convert the CAD model into a point cloud, thereby obtaining the theoretical point cloud.

[0164] The ISS feature point extraction algorithm is used to extract the ISS key points of the actual point cloud, namely the point cloud data of the propeller in the workpiece coordinate system and the theoretical point cloud data.

[0165] Perform FPFH feature matching on the extracted ISS key points to establish initial corresponding point pairs;

[0166] The ISS feature point extraction algorithm identifies feature points with significant geometric changes by analyzing the distribution of eigenvalues ​​of the covariance matrix of each point's neighborhood. The algorithm assumes that the local surface at a feature point exhibits significant changes in all three principal directions, manifested by relatively large eigenvalues ​​in all three directions; points meeting these requirements will be extracted.

[0167] Fast Point Feature Histogram (FPFH) is a local feature descriptor used to describe the geometric features of a point's neighborhood in a 3D point cloud. For each query point, it is necessary to calculate its relationships with all neighboring points. When ISS and FPFH are used simultaneously, FPFH will only perform feature matching on the feature points obtained by ISS, greatly improving the registration speed and quality.

[0168] ISS and FPFH are both commonly used techniques in the field of technical expertise, so they will not be described in detail.

[0169] S32. Based on the SVD algorithm, perform best-fit registration on the initial corresponding point pairs to obtain the actual point cloud after registration and alignment, including:

[0170] Let the actual point cloud in the initial corresponding point pair be... The theoretical point cloud corresponds to the point as The centroids of the actual point cloud and the theoretical corresponding point cloud are calculated as follows:

[0171]

[0172]

[0173] The covariance matrix calculated based on the centroids of the two point clouds is as follows:

[0174]

[0175] The covariance matrix is ​​decomposed by SVD as follows:

[0176]

[0177] Obtain the optimal rotation matrix Translation vector for:

[0178]

[0179]

[0180] Transforming the optimal rotation matrix and translation vector onto the actual point cloud, we obtain the registered and aligned actual point cloud as follows:

[0181]

[0182] in, This represents the actual point cloud after registration and alignment.

[0183] S33. Calculate the amount of grinding for each point in the actual point cloud after registration and alignment based on normal projection;

[0184] Since 3D point cloud data consists of discrete point sets, geometric subtraction between surfaces cannot be performed directly. Therefore, this solution uses the corresponding point difference method based on model point normal search to calculate the grinding amount. The specific steps are as follows:

[0185] The normal vectors of the theoretical point cloud are calculated using principal component analysis, and the directions of the normal vectors are verified, including:

[0186] For any point in the theoretical point cloud Extract the local neighborhood point cloud of this point;

[0187] Calculate the covariance matrix of the local neighborhood point cloud;

[0188] Eigenvalue decomposition of the covariance matrix yields an orthogonal matrix. ;in The direction of the first principal component describing the distribution of the neighborhood point cloud, i.e. , Description and The orthogonal directions of the second principal components, i.e. , Description and , All orthogonal directions, i.e. ;

[0189] Perform direction verification on the normal vector:

[0190] For the upper surface of the propeller (pressure surface / suction surface): normal vector (Pointing outwards from the workpiece, in the direction of material surplus);

[0191] For the lower surface of the propeller (non-working surface): normal vector (Pointing to the inside of the workpiece);

[0192] Define the principal direction constraints and correct the normal vector, including:

[0193] Based on the geometric features of the current region of the propeller, determine the main direction vector of that region;

[0194] Based on the geometric features of the current region of the propeller (e.g., upper / lower surface), determine the principal direction vector of that region (e.g., a combination of PC1, PC2, and PC3, or specified according to process requirements), and set an angle threshold. After setting the angle threshold, calculate the angle between the current normal vector and the principal direction of the region. If the angle is less than or equal to the angle threshold, the normal vector direction conforms to the principal direction constraint of the region, and this normal vector is retained and used as the corrected normal vector. If the angle is greater than the set angle threshold, it indicates that the normal vector direction does not conform to the principal direction constraint of the region, and the corrected normal vector is... Project the vector onto the principal direction and normalize it to obtain the corrected normal vector;

[0195] Step 2: Using any point in the theoretical point cloud Starting from the normal direction, search for the actual corresponding point, including:

[0196] For a point on the upper surface of the propeller, along Corrected normal vector A ray is emitted in the positive direction. All points within a set threshold range near this ray are searched in the registered and aligned actual point cloud as candidate points. The relationship between all candidate points and the target point is calculated. Find the point closest to the theoretical point along the ray direction as the distance to the theoretical point. Corresponding actual measurement points ;

[0197] For a point on the lower surface of the propeller, along Corrected normal vector A ray is emitted in the negative direction. All points within a set threshold range near this ray are searched in the registered and aligned actual point cloud as candidate points. The relationship between all candidate points and the target point is calculated. Find the point closest to the theoretical point along the ray direction as the distance to the theoretical point. Corresponding actual measurement points ;

[0198] Step 3: Calculate the amount of polishing required at a single point, including:

[0199] For point pairs Calculate the amount of polishing for:

[0200]

[0201] in, express and The straight-line distance between them;

[0202] Step 4: Calculate the amount of polishing required for the entire point cloud, including:

[0203] Traverse all points in the theoretical point cloud, execute steps two and three, and generate a set of grinding amount values ​​that correspond one-to-one with each theoretical point. ;

[0204] Step 5: Combine the coordinates of each point in the theoretical point cloud with the remaining values. By associating these elements, a three-dimensional margin distribution mapping relationship covering the entire propeller surface is formed, yielding the grinding amount at each point.

[0205] Specifically, the theoretical model point cloud is a set of theoretically designed points that perfectly match the actual propeller's spatial pose, satisfying the "horizontal placement constraint" (the bottom surface of the propeller hub is horizontal, and the axis is parallel to the Z-axis). This serves as the core geometric data for subsequent path planning, coordinate system establishment, and thickness calculation. The machining allowance distribution set includes data from the theoretical point cloud. Each point in The one-to-one corresponding normal machining allowance value provides data support for the accuracy compensation of subsequent machining trajectories.

[0206] This solution has the following advantages:

[0207] 1. This solution effectively solves the technical challenge of not being able to directly perform surface subtraction operations on point cloud data. Existing technologies typically use simple Euclidean distance matching or allowance calculation methods based on horizontal projection when calculating grinding amounts, which do not fully consider the spatial geometric characteristics of complex surfaces. This results in insufficient accuracy in allowance calculations for workpieces with complex surfaces, such as propellers, and an inability to accurately reflect the actual material removal requirements.

[0208] 2. This scheme proposes a calculation method that starts from the registered theoretical model points and searches for corresponding points in the actual measured point cloud along the direction of their normal vector. This method realizes accurate distance measurement between the theoretical surface and the actual surface in the normal direction from a geometrical perspective, making the calculation results more consistent with the physical meaning of actual processing and greatly improving the accuracy of grinding amount calculation for complex surfaces.

[0209] 3. For the horizontal clamping method commonly used in automated machining of propeller-type workpieces, this solution explicitly proposes a horizontal placement constraint, optimizing the complex three-dimensional allowance search problem into a unidirectional search problem along the normal direction. Simultaneously, based on the engineering fact that the thickness of the blank material is always greater than the theoretical model, a unidirectional allowance constraint is proposed. This dual constraint not only conforms to the physical laws of actual machining scenarios but also eliminates the need for the algorithm to handle complex multiple solutions, significantly reducing computational redundancy. While ensuring the accuracy of the results, it significantly improves computational efficiency, making it more suitable for applications involving online inspection and real-time machining.

[0210] 4. A multi-level point cloud registration strategy integrating key point feature matching and global best fitting solves the problem of large initial deviations between actual point clouds and theoretical models in spatial pose:

[0211] Existing ICP (Iterative Closest Point)-based registration methods typically have high requirements for the initial position of the point cloud, and are prone to getting trapped in local optima when the initial deviation is large. This method uses ISS keypoint extraction combined with FPFH feature descriptors for initial corresponding point matching, providing high-quality initial transformation estimates for subsequent registration. Based on this, a best-fit algorithm based on SVD is applied for global optimization registration. This enables the system to quickly and accurately complete point cloud alignment from any initial pose, enhancing the method's adaptability to different clamping postures.

[0212] In a specific embodiment, the process of generating a machining trajectory that includes blade thickness calibration based on the grinding amount, and then milling and polishing the propeller according to the machining trajectory to complete the machining of the propeller is as follows:

[0213] S41. Obtain local processing direction parameters based on theoretical point clouds, including:

[0214] S411: Based on the radial (hub → blade tip) and circumferential (blade span) characteristics of the propeller, the global path direction is initialized to the circumferential direction and the scanning direction is initialized to the radial direction.

[0215] S412: Optimize the global path direction and scan direction using principal curvature analysis, ensuring the global path direction follows the local minimum curvature direction and the scan direction follows the local maximum curvature direction. This guarantees uniform coverage of the machining trajectory in areas of abrupt curvature changes, avoiding sparse or overly dense trajectories. The specific steps are as follows:

[0216] 1. Extracting the local point cloud neighborhood:

[0217] For each point to be optimized along the global path, a local point cloud neighborhood with a known radius is extracted around it. The neighborhood radius is typically set based on the point cloud density and the leaf feature size.

[0218] 2. Calculate the local surface normal vector and principal curvature of the local point cloud neighborhood, including:

[0219] Normal vector calculation: The covariance matrix of the neighborhood point cloud is calculated by principal component analysis (PCA), and the eigenvector corresponding to the smallest eigenvalue is the local normal vector;

[0220] Principal curvature calculation:

[0221] By fitting a quadratic surface (such as a plane, cylinder, or parabola) to the local point cloud neighborhood, the equation of the fitted surface is obtained.

[0222] The two principal curvatures of the point to be optimized are calculated based on the fitted surface equation. and ( ), respectively corresponding to the local maximum curvature direction and the direction of local minimum curvature ;

[0223] 3. Directional projection and alignment:

[0224] Will and By fusing the global path direction and the scan direction, we obtain the optimized scan direction and the optimized global path direction:

[0225] Project the global circumferential direction and the global radial direction onto the tangent planes respectively, calculate the dot product of the two tangent planes after projection, and use the two dot product results as the first direction similarity and the second direction similarity respectively;

[0226] The calculation of the dot product of projection and tangent plane is a commonly used calculation formula in data processing, so it will not be explained in detail. Those skilled in the art know how to calculate the dot product.

[0227] Determine if the absolute value of the first-direction similarity is greater than the set similarity threshold. If it is, it means that the current global path direction is consistent with the local minimum curvature direction, and the global path direction remains the circumferential direction. If not, it means that the curvature feature of the current region is significant, and the global path direction is corrected to the local minimum curvature direction. This yields the optimized global path direction;

[0228] Determine if the absolute value of the second-direction similarity is greater than the set similarity threshold. If yes, it means the current scanning direction is consistent with the local maximum curvature direction, and the scanning direction remains radial. If no, it means the curvature feature of the current region is significant, and the scanning direction is corrected to the local maximum curvature direction. ; The optimized scanning direction is obtained;

[0229] S42. Divide the area to be polished into equally spaced sections along the optimized global path direction. Sort and resample the point cloud slices of each section at equal intervals to generate a two-dimensional path point sequence, including:

[0230] S421: Along the optimized global path direction, the area to be polished is divided into equal-spaced sections according to the preset path spacing, forming a series of parallel cross-sectional planes;

[0231] S422: For each cross-sectional plane, calculate the distance between the projected values ​​of all points and the current cross-sectional reference value. Filter out all Points smaller than the cross-sectional tolerance range form the original cross-sectional point set;

[0232] In this scheme, for each cross-sectional plane, all points whose projection values ​​along the global path direction in the aligned theoretical model point cloud fall within the set cross-sectional tolerance range (such as ±0.5mm) are extracted to form the original cross-sectional point set;

[0233] Point cloud slicing based on directional vector projection is a basic operation in 3D data processing and a common technique used by those skilled in the art. Therefore, the specific calculation process will not be described in detail.

[0234] S423: Sort the original cross-sectional point set of each cross-sectional plane from smallest to largest along the scanning direction, and then resample the sorted original point set at equal intervals according to the preset point spacing to generate a two-dimensional path point sequence with uniform point density, which not only preserves the key shape features of the cross section, but also avoids trajectory jitter caused by uneven density of the original point cloud.

[0235] S43. Combine the two-dimensional path point sequence with the normal coordinates of the corresponding cross section to form a three-dimensional trajectory point, insert safe transition points, and perform trajectory smoothing to generate a three-dimensional smooth trajectory point sequence, including:

[0236] S431: Combine each two-dimensional path point sequence with the normal coordinate value of the corresponding cross section to synthesize a three-dimensional spatial trajectory point, ensuring the spatial accuracy of the trajectory point;

[0237] S432: Insert a safe transition point between the start and end points of two adjacent two-dimensional path point sequences, thereby integrating the three-dimensional path point sequences of all sections to obtain an integrated three-dimensional path point sequence; wherein, the safe transition point is formed by raising the midpoint of the line connecting the start and end points to a preset safe height along the corrected normal vector direction;

[0238] Specifically, inserting a safety transition point can prevent interference between the tool and the workpiece, ensuring safe and smooth movement between paths;

[0239] S433: The integrated 3D path point sequence is smoothed by moving average or spline curve. The processed 3D path point sequence is the initial processing trajectory. Averaging or smoothing can eliminate the small jitters introduced by discrete sampling, ensuring smooth movement of the robot during processing and improving the quality of the processed surface.

[0240] S44. Perform local theoretical thickness calculation and rigid region marking on the initial machining trajectory to obtain a machining trajectory that includes blade thickness calibration, and use this as the final machining trajectory, including:

[0241] Establishing XY index mapping: Traversing the theoretical point cloud For each point Its XY coordinates Use it as an index key to store the corresponding Z value. Store the data in the list of Z values ​​corresponding to the index key to form a mapping table that represents the mapping relationship between XY coordinates and multiple Z values, thereby achieving data structuring;

[0242] Extracting Z-value extrema: For each unique XY location, extract the maximum Z-value from its corresponding list of Z-values ​​as the upper surface Z-value of the XY location. The minimum Z value is taken as the Z value of the lower surface. (Based on the theoretical constraint of symmetrical distribution on the upper and lower surfaces of the blade);

[0243] Calculate the local theoretical thickness: Calculate the local theoretical thickness at each XY position. This thickness value directly reflects the rigidity of the blade in that region. The formula is:

[0244]

[0245] Rigid region identification and marking: preset rigidity threshold , , For each trajectory point in the three-dimensional path point sequence Corresponding to its corresponding local theoretical thickness Mark region attributes according to the following rules:

[0246] like If the region where the trajectory points are marked is a rigid region, then the region is encoded as C. k =1;

[0247] like If the region attribute of the marked trajectory point is a standard rigid region, and C is encoded... k =2;

[0248] like If the region where the trajectory points are marked is a transition region, then C is encoded. k =3;

[0249] like Then the region attribute of the marked trajectory points is a flexible region, and C is encoded. k =4;

[0250] Based on the above marking results, a machining trajectory including blade thickness calibration was obtained;

[0251] The propeller is milled and polished according to the machining trajectory to complete the propeller machining, including:

[0252] Receive a sequence of path points with region markings and a set of machining allowances, and adjust the machining parameters according to the rigid markings of the path points, wherein:

[0253] For the rigid region (C) k =1): The processing speed and feed rate can be appropriately increased to improve processing efficiency;

[0254] For the standard rigid region (C) k =2): Use conventional processing speed and force;

[0255] For the transition region (C) k =3): Reduce the feed rate, maintain a stable machining speed, and avoid vibration caused by sudden changes in rigidity;

[0256] For flexible regions (C) k =4): Significantly reduce processing speed and force, while simultaneously activating a localized stiffness enhancement device at the blade tip, such as Figure 5 As shown, this reduces vibration deformation in thin-walled regions;

[0257] At the same time, the processing depth of each path point is compensated and adjusted based on the processing allowance value to ensure that the actual removal amount is consistent with the theoretical calculation.

[0258] In this embodiment, based on the processing requirements, rough machining, semi-finishing, and finishing are performed. The specific steps are as follows:

[0259] Step 1: Set the target removal amounts for roughing, semi-finishing, and finishing of the propeller; the roughing and semi-finishing are both milling operations, and the finishing is grinding and polishing.

[0260] Roughing (milling): Remove most of the machining allowance (such as 70%-80% of the total allowance), use larger cutting parameters to quickly approach the theoretical shape, and at the same time reserve a uniform allowance for semi-finishing;

[0261] Semi-finishing (milling): Further optimize the workpiece shape, remove surface defects caused by roughing, and leave a small amount of finishing allowance (such as 0.1-0.3mm).

[0262] Precision machining (grinding): For critical parts such as blade tip radius, fine grinding process is used to ensure that the surface roughness and dimensional accuracy of the machined parts meet the design requirements;

[0263] Step 2: Perform milling machining and monitor the actual amount of material removed during milling after the operation is completed.

[0264] Perform roughing according to the preset processing trajectory and the target removal amount of roughing;

[0265] After rough machining is completed, the workpiece is scanned in three dimensions to calculate the actual amount of material removed.

[0266] Calculate the difference between the actual removal amount and the target removal amount of roughing. If the difference is greater than 0, it means that the processing is over-processed. Subtract the difference from the target removal amount of semi-finishing to obtain the new target removal amount of semi-finishing. If the difference is less than 0, it means that the processing is under-processed. Add the difference to the target removal amount of semi-finishing to obtain the new target removal amount of semi-finishing.

[0267] Perform semi-finishing according to the new target removal rate;

[0268] Step 3: Based on the deviation between the actual milling removal amount in the semi-finishing stage and the preset milling target amount, adjust the removal amount in the grinding and polishing stage:

[0269] Calculate the actual removal amount in the semi-finishing process and the new target removal amount in the semi-finishing process. If the difference is greater than 0, it indicates over-processing. Subtract the difference from the target removal amount in the finishing process to obtain the new target removal amount in the finishing process. If the difference is less than 0, it indicates under-processing. Add the difference to the target removal amount in the semi-finishing process to obtain the new target removal amount in the finishing process.

[0270] Perform finishing according to the new target removal amount, and polish the workpiece.

[0271] In this plan, if the allowance is ≤0.5mm, it is determined that the milling amount is too large, and the toolpath needs to be adjusted according to the operator's experience before finishing. If 0.5mm≤allowance≤2mm, it is determined to be a qualified area and the next finishing operation can be carried out. If the allowance is ≥2mm, it is determined that the milling amount is insufficient and milling needs to be performed again.

[0272] At the end of milling, ensure that the surface has a uniform polishing allowance of 0.5-2mm.

[0273] During the processing, the positioning reference position of the propeller is scanned by the scanning device according to the set scanning time. It is calculated whether the current positioning reference of the propeller is consistent with the initial positioning reference. If they are inconsistent, the offset parameter is calculated to reset the propeller position.

[0274] Based on the scan data, recalculate the current coordinates of the positioning reference point in the machine tool coordinate system. If it is inconsistent with the initial positioning reference, stop machining, subtract the coordinates of the current reference point from the coordinates of the reference point during positioning to obtain the offset, input the offset into the machine tool control system, adjust the angle of the B-axis rotary table to return the reference point coordinates to the initial position, and continue machining.

[0275] The processing methods in this plan specifically include:

[0276] Macro motion control (machine tool): Based on the three-dimensional machining trajectory, control the machine tool to quickly and roughly position the tool to a safe position near the blade tip, and provide a stable machining platform to ensure the basic accuracy of the tool movement;

[0277] Micro-motion control (FTS): Under conditions where the machine tool is almost stationary or feeds at low speed, the FTS (Flexible Transmission System) is controlled to execute the machining trajectory of fine parts such as blade tip fillets with extremely high frequency and precision, so as to achieve high-precision forming.

[0278] Specifically, this embodiment also includes closed-loop detection and data archiving:

[0279] Real-time inspection during processing: After each pre-set length (e.g., 5mm) of fillet processing is completed, the dimensional error of that area is immediately measured by scanning to evaluate the contour accuracy and surface roughness; if the inspection result is unqualified, the processing parameters are immediately adjusted based on the error data, and compensation and fine-tuning are performed in that local area to avoid error accumulation;

[0280] Post-processing full-area inspection: After processing, the propeller blades (including the blade tip radius) are continuously scanned in three dimensions to generate a final three-dimensional quality map, which is then compared with the theoretical model to verify whether the processing accuracy meets the requirements.

[0281] Data archiving: All data during the processing (including point cloud data, registration results, processing allowance, trajectory parameters, inspection reports, etc.) are automatically archived and associated with the digital twin model of the workpiece to achieve full lifecycle traceability;

[0282] Handling of non-conformities: If the overall inspection results do not meet the design requirements, return to the machining allowance calculation step, regenerate the machining allowance distribution and machining trajectory, and perform secondary machining until the inspection is qualified.

[0283] This invention first generates machining trajectory points covering the entire surface of the propeller. Then, based on the local thickness information (the distance between the two surfaces of the theoretical model at that point) at the location of each trajectory point, it automatically marks whether the point belongs to a rigid or flexible region and attaches the region label information to the machining trajectory, providing a basis for subsequently matching different machining parameters and methods. This method can adaptively generate machining strategies based on the propeller's own geometric characteristics without manual intervention in region division, significantly improving the adaptability of the machining process.

[0284] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A processing method suitable for large marine propellers, characterized in that, include: S1: Install the propeller and standard ball on the B-axis rotary table of the machine tool, and install the high-precision calibration needle on the A spindle. Use a fitting algorithm to determine the positioning reference of the propeller based on the positions of the standard ball and the high-precision calibration needle, and construct the workpiece coordinate system and the transformation relationship between the workpiece coordinate system and the machine tool coordinate system. S2: Acquire point cloud data of the propeller in the camera coordinate system, convert the point cloud data in the camera coordinate system to point cloud data in the machine tool coordinate system based on the positioning reference, and determine the point cloud data of the propeller in the workpiece coordinate system based on the conversion relationship, including: S21. Collect initial point cloud data of the propeller, and based on the initial point cloud data, obtain point cloud data for each part of the blade tip, blade surface, and blade root, including: S211: The number of blades is identified based on the initial point cloud data; S212: Create a single leaf photography template, obtain the optimal photography path based on the single leaf photography template and the number of leaves, and collect the original three-dimensional point cloud data of the leaf based on the optimal photography path; wherein the leaf includes the leaf tip, leaf surface and leaf root regions; S213: Obtain the curvature of the leaf tip and leaf surface regions based on the original three-dimensional point cloud data of the leaf tip and leaf surface regions. Based on the curvature of the leaf tip and leaf surface regions, adopt an adaptive downsampling strategy to optimize the original three-dimensional point cloud data of the leaf tip and leaf surface regions to obtain the final point cloud data of the leaf tip and leaf surface. S214: Convert the final point cloud data of the blade tip and blade surface and the original three-dimensional point cloud data of each part of the blade root into point cloud data in the machine tool coordinate system based on the positioning reference, and determine the point cloud data of each part of the blade tip, blade surface and blade root in the workpiece coordinate system based on the conversion relationship. Based on the positioning reference, the point cloud data of each part of the blade tip, blade surface and blade root are converted into point cloud data in the machine tool coordinate system, and the point cloud data of each part of the blade tip, blade surface and blade root in the workpiece coordinate system are determined based on the conversion relationship. S22. Perform coarse stitching on the point cloud data of each part of the blade root in the workpiece coordinate system, and perform fine stitching on the coarse stitching result based on the ICP algorithm to obtain complete point cloud data of the blade root in the workpiece coordinate system. S23. Collect point cloud data of the propeller hub in the camera coordinate system, convert the point cloud data in the camera coordinate system into point cloud data in the machine tool coordinate system based on the positioning reference, and determine the point cloud data in the workpiece coordinate system based on the conversion relationship. S24. Combine the point cloud data of the blade tip, blade surface, hub, and complete blade root in the workpiece coordinate system to form the final point cloud data of the propeller in the workpiece coordinate system. S3: Calculate the grinding amount based on the point cloud data of the propeller in the workpiece coordinate system according to the normal projection; S4: Generate a machining trajectory that includes blade thickness calibration based on the grinding amount, and perform milling and polishing on the propeller according to the machining trajectory to complete the machining of the propeller.

2. The processing method for large marine propellers according to claim 1, characterized in that, The number of blades was identified based on the initial point cloud data, including: Step 1: Perform voxel mesh downsampling on the initial point cloud data: The three-dimensional space of the initial point cloud data is divided into several uniform grids according to a preset voxel grid size, and the centroid of the point set formed by the points in the initial point cloud data in each grid is taken as the centroid of the current grid, thus obtaining the downsampled point cloud data. An adaptive ROI extraction algorithm was used to extract effective point cloud regions that are only related to the blades from the downsampled point cloud data. Leaf point cloud segmentation: The region growing segmentation algorithm is used to divide the effective point cloud region into two independent leaf point cloud sets by dividing points with the same normal direction and curvature into the same sub-region. An edge detection algorithm is used to extract the edge point cloud data of the two segmented blade point clouds. The edge point cloud data includes regions with surface discontinuities and sharp curvature changes in the point cloud, as well as the physical boundaries, edges, or surface intersections of the blades corresponding to these regions. Point cloud registration calculation: Coarse registration: Calculate the centroid coordinates of the point clouds at the edges of the two blades respectively, obtain the initial rigid body transformation matrix by aligning the centroids, and perform preliminary alignment of the point clouds at the edges of the two blades based on the initial rigid body transformation matrix to provide an initial estimate for fine registration; Fine registration: The ICP algorithm is used to minimize the distance between corresponding points on the edge point clouds of two blades, thereby iteratively optimizing the initial rigid body transformation matrix to obtain the optimal transformation matrix, and further aligning the edge point clouds of two blades based on the optimal transformation matrix. Leaf count determination: Extract the rotation matrix from the optimal transformation matrix, and convert the rotation matrix into rotation angle values ​​using a quaternion-to-Euler angle conversion algorithm; The number of blades is determined based on the rotation angle value.

3. The processing method for large marine propellers according to claim 2, characterized in that, A single-leaf photography template is created. Based on the template and the number of leaves, an optimal photography path is obtained. The original 3D point cloud data of the leaves is then acquired using this optimal path, including: Create a single-leaf photo template: The movement path when collecting propeller point cloud data is obtained, and a single blade image template is obtained based on the movement path. The image template includes the spatial movement trajectory, attitude, and shooting parameters of each point. Determine the reference blade: Set the physical zero point of the turntable to 0 degrees in the workpiece coordinate system, and calculate the clockwise angle difference between the direction angle of each blade and the physical zero point of the turntable. Compare the magnitudes of all clockwise angle differences and select the blade with the smallest clockwise angle difference as the reference blade closest to the physical zero point of the turntable; The image path for generating the reference blade: The transformation matrix between the reference blade and the photographic template is calculated using coarse and fine registration methods, and the rotation angle between the reference blade and the photographic template is obtained by converting quaternions to Euler angles. The angle between the reference blade and the image template is converted into a transformation matrix. The image template is then transformed based on the transformation matrix to obtain the image path of the reference blade. Multi-leaf photography path expansion: The Nth blade is obtained by rotating the photographic path of the reference blade according to the rotation angle. The image path is then used to obtain the original image path of all the blades. The calculation process for the rotation angle is as follows: Rotation angle = rotation angle between the reference blade and the photo template + (N-1) × rotation angle value determined during the blade number recognition process; Optimize the original image paths for all leaves: The original photography paths of all leaves are integrated into a single point set. The problem of the order of visiting photography points is modeled as a traveling salesman problem, i.e., any two photography points are defined... and The cost of moving between them is the Euclidean distance between them: Starting from any point, the nearest neighbor greedy algorithm is used to select the next point with the closest Euclidean distance as the next destination, thus obtaining an initial feasible path; The initial feasible path is iteratively improved by the 2-opt local optimization method: try to swap any two segments in the path, calculate the change in the total path length before and after the swap, and only perform the swap if the swap can reduce the total distance. By iteratively finding the optimal photo-taking path, the order of accessing the photo-taking point indexes that minimizes the sum of Euclidean distances, is the optimal photo-taking path. Based on the optimal shooting path, the camera is moved to acquire the original three-dimensional point cloud data of each leaf.

4. The processing method for large marine propellers according to claim 3, characterized in that, The curvature of the leaf tip and leaf surface regions is obtained from the original 3D point cloud data. Based on the curvature of the leaf tip and leaf surface regions, an adaptive downsampling strategy is used to optimize the original 3D point cloud data of the leaf tip and leaf surface, resulting in the final point cloud data of the leaf tip and leaf surface, including: Calculate any point in the original 3D point cloud data for each leaf tip and leaf surface. The local curvature includes: by Set the radius around the center. Select all points within this radius to form a neighborhood. N i ; Computing the neighborhood The center of gravity, the formula is: in, For the neighborhood N i The number of points within; The formula for calculating the covariance matrix is: For covariance matrix Eigenvalue decomposition yields three eigenvalues. , and ,and ; calculate p i The local curvature estimate is given by the formula: Curvature grading: Preset curvature threshold and The point cloud is divided into three regions based on a preset curvature threshold: Will The region is set as a high curvature region; Will The region is set to a medium curvature region; Will The region is set to a low curvature region; Point cloud adaptive downsampling: Based on the curvature grading results, the original three-dimensional point cloud data of the leaf tip and leaf surface were downsampled using different voxel sizes to obtain optimized point cloud data of the leaf tip and leaf surface.

5. A processing method for large marine propellers according to claim 1, characterized in that, The grinding amount is calculated based on the point cloud data of the propeller in the workpiece coordinate system using normal projection, including: The theoretical model is processed into point cloud data: a CAD model of the propeller is constructed, and the CAD model is processed into point cloud data according to a preset sampling density. The ISS feature point extraction algorithm is used to extract the ISS key points of the actual point cloud, namely the point cloud data of the propeller in the workpiece coordinate system and the theoretical point cloud data. Perform FPFH feature matching on the extracted ISS key points to establish initial corresponding point pairs; S32. Based on the SVD algorithm, perform best-fit registration on the initial corresponding point pairs to obtain the actual point cloud after registration and alignment; S33. Calculate the amount of grinding for each point in the actual point cloud after registration and alignment based on normal projection.

6. A processing method for large marine propellers according to claim 5, characterized in that, Based on the SVD algorithm, the initial corresponding point pairs are best-fitted and registered to obtain the actual point cloud after registration and alignment, including: Let the actual point cloud in the initial corresponding point pair be... The theoretical point cloud corresponds to the point as The centroids of the actual point cloud and the theoretical corresponding point cloud are calculated as follows: The covariance matrix calculated based on the centroids of the two point clouds is as follows: The covariance matrix is ​​decomposed by SVD as follows: Obtain the optimal rotation matrix Translation vector for: Transforming the optimal rotation matrix and translation vector onto the actual point cloud, we obtain the registered and aligned actual point cloud as follows: in, This represents the actual point cloud after registration and alignment.

7. A processing method for large marine propellers according to claim 5, characterized in that, The specific steps for calculating the amount of grinding required for each point in the actual point cloud after registration and alignment are as follows: The normal vectors of the theoretical point cloud are calculated using principal component analysis, and the directions of the normal vectors are verified, including: For any point in the theoretical point cloud Extract the local neighborhood point cloud of this point; Calculate the covariance matrix of the local neighborhood point cloud; Eigenvalue decomposition of the covariance matrix yields an orthogonal matrix. ;in The direction of the first principal component describing the distribution of the neighborhood point cloud, i.e. , Description and The orthogonal directions of the second principal components, i.e. , Description and , All orthogonal directions, i.e. ; Perform direction verification on the normal vector: For the upper surface of the propeller, the normal vector ; For the lower surface of the propeller, the normal vector ; Define the principal direction constraints and correct the normal vector, including: Based on the geometric features of the current region of the propeller, determine the main direction vector of that region; Set an angle threshold, calculate the angle between the current normal vector and the main direction of the region. If the angle is less than or equal to the angle threshold, it means the normal vector direction conforms to the main direction constraint of the region; retain the normal vector and use it as the corrected normal vector. If the angle is greater than the set angle threshold, it means the normal vector direction does not conform to the main direction constraint of the region; then... Project the vector onto the principal direction and normalize it to obtain the corrected normal vector; Step 2: Using any point in the theoretical point cloud Starting from the normal direction, search for the actual corresponding point, including: For a point on the upper surface of the propeller, along Corrected normal vector A ray is emitted in the positive direction. All points within a set threshold range near this ray are searched in the registered and aligned actual point cloud as candidate points. The relationship between all candidate points and the target point is calculated. Find the point closest to the theoretical point along the ray direction as the distance to the theoretical point. Corresponding actual measurement points ; For a point on the lower surface of the propeller, along Corrected normal vector A ray is emitted in the negative direction. All points within a set threshold range near this ray are searched in the registered and aligned actual point cloud as candidate points. The relationship between all candidate points and the target point is calculated. Find the point closest to the theoretical point along the ray direction as the distance to the theoretical point. Corresponding actual measurement points ; Step 3: Calculate the amount of polishing required at a single point, including: For point pairs Calculate the amount of polishing for: in, express and The straight-line distance between them; Step 4: Calculate the amount of polishing required for the entire point cloud, including: Traverse all points in the theoretical point cloud, execute steps two and three, and generate a set of grinding amount values ​​that correspond one-to-one with each theoretical point. ; Step 5: Combine the coordinates of each point in the theoretical point cloud with the remaining values. By associating these elements, a three-dimensional margin distribution mapping relationship covering the entire propeller surface is formed, yielding the grinding amount at each point.

8. A processing method for large marine propellers according to claim 1, characterized in that, Based on the grinding amount, a machining trajectory including blade thickness calibration is generated, including: S41. Obtain local processing direction parameters based on theoretical point clouds, including: S411: Based on the radial and circumferential features of the propeller, the global path direction is initialized to the circumferential direction and the scanning direction is initialized to the radial direction; S412: Use principal curvature analysis to optimize the global path direction and scan direction, so that the global path direction is along the local minimum curvature direction and the scan direction is along the local maximum curvature direction; S42. Divide the area to be polished into equally spaced sections along the optimized global path direction. Sort and resample the point cloud slices of each section at equal intervals to generate a two-dimensional path point sequence, including: S421: Along the optimized global path direction, the area to be polished is divided into equal-spaced sections according to the preset path spacing, forming a series of parallel cross-sectional planes; S422: For each cross-sectional plane, calculate the distance between the projected values ​​of all points and the current cross-sectional reference value. Filter out all Points smaller than the cross-sectional tolerance range form the original cross-sectional point set; S423: Sort the original cross-sectional point set of each cross-sectional plane from smallest to largest along the optimized scanning direction, and then resample the sorted original cross-sectional point set at equal intervals according to the preset point spacing to generate a two-dimensional path point sequence; S43. Combine the two-dimensional path point sequence with the normal coordinates of the corresponding cross section to form a three-dimensional trajectory point, insert safe transition points, and perform trajectory smoothing to generate a three-dimensional smooth trajectory point sequence, including: S431: Combine each two-dimensional path point sequence with the normal coordinate value of the corresponding cross section to synthesize a three-dimensional spatial trajectory point; S432: Raise the midpoint of the line connecting the start and end points of two adjacent two-dimensional path point sequences along the normal direction to a preset safe height to form a safe transition point; S433: Perform moving average or spline curve smoothing on the integrated 3D path point sequence to obtain the initial processing trajectory; S44. Perform local theoretical thickness calculation and rigid region marking on the initial machining trajectory to obtain a machining trajectory that includes blade thickness calibration, and use this as the final machining trajectory, including: Establishing XY index mapping: Traversing the theoretical point cloud For each point Its XY coordinates Use it as an index key to store the corresponding Z value. Store the Z-values ​​corresponding to the index key in the list to form a mapping table that represents the mapping relationship between XY coordinates and multiple Z-values; Extracting Z-value extrema: For each unique XY location, extract the maximum Z-value from its corresponding list of Z-values ​​as the upper surface Z-value of the XY location. The minimum Z value is taken as the Z value of the lower surface. ; Calculate the local theoretical thickness: Calculate the local theoretical thickness at each XY position using the following formula: Rigid region identification and marking: preset rigidity threshold , , For each trajectory point in the three-dimensional path point sequence Corresponding to its corresponding local theoretical thickness Mark region attributes according to the following rules: like If the region where the trajectory points are marked is a rigid region, then the region is encoded as C. k =1; like If the region attribute of the marked trajectory point is a standard rigid region, and C is encoded... k =2; like If the region where the trajectory points are marked is a transition region, then C is encoded. k =3; like Then the region attribute of the marked trajectory points is a flexible region, and C is encoded. k =4; Based on the above marking results, a machining trajectory including blade thickness calibration was obtained.

9. A processing method for large marine propellers according to claim 1, characterized in that, The propeller is milled and polished according to the machining trajectory to complete the propeller machining, including: Step 1: Set the target removal amounts for roughing, semi-finishing, and finishing of the propeller; the roughing and semi-finishing are both milling operations, and the finishing is grinding and polishing. Step 2: Perform milling machining and monitor the actual amount of material removed during milling after the operation is completed. Perform roughing according to the preset processing trajectory and the target removal amount of roughing; After rough machining is completed, the workpiece is scanned in three dimensions to calculate the actual amount of material removed. Calculate the difference between the actual removal amount and the target removal amount of roughing. If the difference is greater than 0, it means that the processing is over-processed. Subtract the difference from the target removal amount of semi-finishing to obtain the new target removal amount of semi-finishing. If the difference is less than 0, it means that the processing is under-processed. Add the difference to the target removal amount of semi-finishing to obtain the new target removal amount of semi-finishing. Perform semi-finishing according to the new target removal rate; Step 3: Based on the deviation between the actual milling removal amount in the semi-finishing stage and the preset milling target amount, adjust the removal amount in the grinding and polishing stage: Calculate the actual removal amount in the semi-finishing process and the new target removal amount in the semi-finishing process. If the difference is greater than 0, it indicates over-processing. Subtract the difference from the target removal amount in the finishing process to obtain the new target removal amount in the finishing process. If the difference is less than 0, it indicates under-processing. Add the difference to the target removal amount in the finishing process to obtain the new target removal amount in the finishing process. Perform finishing according to the new target removal amount, and polish the workpiece.

10. A processing method for large marine propellers according to claim 8, characterized in that, Principal curvature analysis is used to optimize the global path direction and scan direction, ensuring that the global path direction follows the local minimum curvature direction and the scan direction follows the local maximum curvature direction. The specific steps are as follows: Step 1: For each point to be optimized along the global path, extract a local point cloud neighborhood with a known radius around it; Step 2: Calculate the local surface normal vector based on the local point cloud neighborhood, including: The covariance matrix of the local point cloud neighborhood is calculated by principal component analysis, and the eigenvector corresponding to the smallest eigenvalue is the local normal vector. Step 3: Perform quadratic surface fitting on the local point cloud neighborhood, and calculate the two principal curvatures of the point from the fitted surface equation. and , , respectively corresponding to the direction of maximum curvature and the direction of minimum curvature ; Step 4: and By fusing the global path direction and the scan direction, we obtain the optimized scan direction and the optimized global path direction: Project the global circumferential direction and the global radial direction onto the tangent planes respectively, calculate the dot product of the two tangent planes after projection, and use the two dot product results as the first direction similarity and the second direction similarity respectively; Determine if the absolute value of the first-direction similarity is greater than the set similarity threshold. If it is, it means that the current global path direction is consistent with the local minimum curvature direction, and the global path direction remains the circumferential direction. If not, it means that the curvature feature of the current region is significant, and the global path direction is corrected to the local minimum curvature direction. This yields the optimized global path direction; Determine if the absolute value of the second-direction similarity is greater than the set similarity threshold. If yes, it means the current scanning direction is consistent with the local maximum curvature direction, and the scanning direction remains radial. If no, it means the curvature feature of the current region is significant, and the scanning direction is corrected to the local maximum curvature direction. The optimized scanning direction is obtained.

11. A processing method for large marine propellers according to claim 1, characterized in that, The propeller is milled and polished according to the machining trajectory. During the machining process, the positioning reference position of the propeller is scanned by a scanning device at a set scanning time. The current positioning reference of the propeller is calculated to determine if it matches the initial positioning reference. If they do not match, an offset parameter is calculated to reset the propeller position. Based on the scan data, the current coordinates of the positioning reference point in the machine tool coordinate system are recalculated. If they are inconsistent with the initial positioning reference, machining is stopped. The current reference point coordinates are subtracted from the reference point coordinates at the time of positioning to obtain the offset. The offset is input into the machine tool control system, and the angle of the B-axis rotary table is adjusted to return the reference point coordinates to the initial position, and machining continues.

12. A processing method for large marine propellers according to claim 1, characterized in that, The propeller positioning reference is determined based on the positions of a standard sphere and a high-precision calibration needle using a fitting algorithm, including: Set spindle A to the 0-degree position, and control the B-axis rotary table to rotate at the set angle. Perform the following steps at each angle: Step 1: Obtain point cloud data of the standard sphere in the camera coordinate system at various rotation angles; Step 2: Move spindle A so that the tip of the high-precision calibration needle contacts the preset contact point on the standard ball, and obtain the machine coordinates of the contact point in the machine coordinate system, that is, the true coordinates of the contact point; Step 3: Add the actual machine coordinates of the contact point to the radius of the standard sphere to obtain the actual machine coordinates of the center of the standard sphere in the machine coordinate system at each rotation angle. The machine tool coordinate system includes the X, Y, Z, A-axis, and B-axis positions; k represents the rotation angle. Place the A-axis at 0 degrees and at a set angle, move the A-spindle until the high-precision calibration pin touches the contact point of the standard ball, and record the machine tool coordinates at each angle. The point cloud data of the standard sphere in the camera coordinate system is filtered and denoised to obtain standard point cloud data; The standard point cloud data is fitted using the least squares method to determine the coordinates of the sphere center in the camera coordinate system, the rotation center OB of the B-axis rotary table, and the rotation center OA of the A-axis. The steps are as follows: Step 1: Fit the standard point cloud data at each rotation angle using the least squares method to obtain the coordinates of the center of the standard sphere in the camera coordinate system at each rotation angle. ; Step 2: Fit the coordinates of the sphere center using the least squares method to obtain the rotation center OB of the B-axis rotary table; Step 3: Connect the machine tool coordinates acquired at the 0-degree position and the set angle position of the A-axis, draw the perpendicular bisector of the line segment, and the intersection of the perpendicular bisector with the rotation axis of the A-axis is the rotation center OA of the A-spindle; based on and Construct a hand-eye matrix, i.e., a positioning reference, to perform the transformation between the camera coordinate system and the machine tool coordinate system based on the hand-eye matrix. The steps are as follows: Step 1: Perform SVD decomposition algorithm on... By decomposing the matrix, the optimal rotation matrix and relative translation can be obtained. ; Step 2: Add the coordinates of OB and OA to the relative translation to obtain the complete translation vector; Step 3: Combine the rotation matrix and translation vector to form the final 4×4 homogeneous transformation matrix, i.e., the hand-eye matrix, as follows: Where R represents the rotation matrix and t represents the relative translation.

13. A processing method for large marine propellers according to claim 1, characterized in that, Establish the workpiece coordinate system and the transformation relationship between the workpiece coordinate system and the machine tool coordinate system, including: Establish the workpiece coordinate system with the rotation center OB of axis B as the origin of the workpiece coordinate system; The transformation relationship between the workpiece coordinate system and the machine tool coordinate system is established as follows: in, Let the coordinates of the point be in the machine tool coordinate system. Let be the coordinates of the point in the workpiece coordinate system. The coordinates of the origin of the workpiece coordinate system in the machine tool coordinate system.

Images