Turbine blade laser processing online correction method, system, medium and equipment

Abstract

The invention discloses a turbine blade laser processing online correction method and system based on three-dimensional vision, a medium and equipment, in the method, binocular vision structured light is adopted to measure the contour of a to-be-processed surface at a scanning position, point cloud registration is utilized to map processing hole sites in a model coordinate system into measurement point cloud in a vision coordinate system, and a to-be-processed point site sequence is generated; five-axis machine tool machining coordinates are generated, and mark point machining is conducted on the pre-machining point positions on the surface of the workpiece; a mark point of a preprocessing point location is reconstructed by adopting a method based on a minimum epipolar constraint error, the mark point is merged with a processed workpiece surface point cloud, and when the workpiece surface contours before and after processing are measured to be consistent, the coordinate deviation of the preprocessing point location and the processed black mark point is directly evaluated; and according to the sampling points, fitting deviation functions of the processed black mark points and theoretical coordinates, and according to the deviation functions, predicting positioning errors of all the to-be-processed point positions and directly correcting the positioning errors into formal processing.

Description

Technical Field

[0001] This invention relates to the field of laser processing technology for aero-engine blades, and in particular to an online correction method, system, medium, and equipment for laser processing of turbine blades based on three-dimensional vision. Background Technology

[0002] Turbine blades are core components of aero-engines. Advanced engine blades utilize film cooling (WDC) as a key structure to enhance their performance. The machining and positioning of WDC requires high precision, directly impacting engine performance and safety. Traditional machining methods suffer from low efficiency and large heat-affected zones, making it difficult to meet the micron-level precision requirements of complex components. Laser processing, as an advanced high-performance manufacturing method, provides a new approach for machining critical aero-engine components.

[0003] Existing advanced machining guidance methods employ coordinate measuring machines (CMMs) for sampling point positioning to determine the workpiece's position and orientation. The accuracy of sampling point positioning using a contact probe of the CMM is directly affected by the actual surface contour error of the workpiece. When the surface contour has errors and the number of sampling points is small, the workpiece orientation calculation may have significant deviations, leading to positioning errors in machining. Increasing the number of surface contour samples can improve the accuracy of orientation calculation, but it will reduce the machining positioning speed to some extent. Furthermore, for parts with unknown models or large actual surface contour errors, the deviation of contact measurement point correction will be more significant in areas of abrupt curvature change (such as the leading edge of a blade). In addition, after using contact measurement to guide machining positioning, the assessment of positioning errors often relies solely on the surface contour, failing to identify and measure smaller features, thus presenting significant limitations.

[0004] The information disclosed in the background section is only for enhancing the understanding of the background of this invention, and therefore may contain information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] To address the shortcomings, this invention provides a method, system, medium, and equipment for online correction of laser processing of turbine blades based on three-dimensional vision. The complete contour of the part to be processed is obtained through three-dimensional vision, the workpiece posture is accurately calculated based on the complete surface information, the processing positioning deviation is evaluated online and the error is corrected. The consistency deviation between surface measurement and processing positioning is ensured within the vision system, thereby achieving high-precision laser processing of important structural features of aero-engine turbine blades.

[0006] A method for online correction of turbine blade laser processing based on 3D vision includes:

[0007] Step S1: Construct a five-axis kinematic model based on the geometry of the five-axis machine tool, and define the reference coordinate system, the vision coordinate system, and the model coordinate system;

[0008] Step S2: Fix a standard ball on the motion platform of the five-axis machine tool, obtain the five-axis motion trajectory by visual measurement of the standard ball, define the center of the rotation axis as the origin of the reference coordinate system, calculate the rigid transformation matrix [R|T]v_r between the reference coordinate system and the visual coordinate system, and obtain the rigid transformation matrix [R|T]r_l between the laser processing focus position and the origin of the reference coordinate system using visual measurement;

[0009] Step S3: Use binocular vision structured light to measure the contour of the surface to be processed at the scanning position, and use point cloud registration to map the processing hole positions in the model coordinate system to the measured point cloud in the vision coordinate system to generate the processing point position sequence {PL}, and sample from it as pre-processing point positions {pl}.

[0010] Step S4: Generate five-axis machine tool machining coordinates based on the pre-machining point {pl}, rigid transformation matrix [R|T]v_r and rigid transformation matrix [R|T]r_l, and perform marker machining at the pre-machining point on the workpiece surface;

[0011] Step S5: The surface contour of the processed workpiece is reconstructed again at the scanning position using binocular vision structured light, and the marker points of the pre-processed points are reconstructed at the same time, so that the marker points are merged with the point cloud of the surface of the processed workpiece; and the surface contour of the workpiece before and after processing is searched and evaluated. If the deviation of the surface contour of the workpiece before and after processing is less than the threshold th1, the visual guidance is effective and deviation correction is performed; otherwise, return to step S2 for recalibration.

[0012] Step S6: When the workpiece surface contour before and after machining is consistent, directly evaluate the coordinate deviation between the pre-machining point and the marker point after machining, and fit the deviation functions ΔX(x,y,z), ΔY(x,y,z), and ΔZ(x,y,z) between the marker point after machining and the theoretical coordinates according to the sampling points. Based on the deviation functions, predict the positioning error of all points to be machined and directly correct it in the formal machining.

[0013] In the aforementioned online correction method for laser processing of turbine blades based on three-dimensional vision, step S1 includes:

[0014] Step S11: The three-dimensional measurement uses a structured light measurement system in stereo vision;

[0015] Step S12: Construct a five-axis kinematic model (X, Y, Z, C1, C2) based on the geometry of the five-axis machine tool, and define the intersection of axis C1 and axis C2 as the origin of the reference coordinate system. The direction of the reference coordinate system is consistent with the axis (X, Y, Z) of the five-axis machine tool.

[0016] Step S13: Define the visual coordinate system and measurement position. The vision module is fixed to the Z-axis of the five-axis machine tool. The center of the optimal depth range of the vision measurement system relative to the reference coordinate system in the machine tool is taken as the measurement position. The visual coordinate system is the camera coordinate system after binocular vision calibration.

[0017] Step S14: Using the design coordinate system of the workpiece model as the model coordinate system, register the workpiece CAD model with the measured point cloud of the workpiece, and map the machining points {PL}d in the model to the measured point cloud using the nearest neighbor search method to generate the sequence of points {PL} to be machined.

[0018] {PL}=[R|T]*{PL}d, where [R|T] is the coordinate transformation matrix between the workpiece CAD model and the measured point cloud of the workpiece.

[0019] In the aforementioned online correction method for laser processing of turbine blades based on three-dimensional vision, step S2 includes:

[0020] Step S21: The standard ball is fixed on the motion axis. During the motion of the standard ball with the rotation axis, the structured light measurement system synchronously measures the pre-given trajectory of the standard ball and fits the coordinates of the ball's center. By fitting the spatial center (x, y, z) and the axis direction (n1, n2, n3), two sets of spatial straight line expressions corresponding to the rotation axes C1 and C2 are constructed. Then, the least squares solution is used to solve the intersection point of the spatial straight lines as the origin of the reference coordinate system, and the direction of the reference coordinate system (X, Y, Z) is defined to be consistent with the machine tool motion.

[0021] In the formula, Indicates the first Group 1 Given a set of spatial plane equation parameters, two intersecting planes represent a spatial straight line. Combined with the given spatial center of the fitted circle, (x,y,z) are the coordinates of the intersection point of the two spatial straight lines.

[0022] Step S22: Move the machine tool and visually measure the position of the center of the standard sphere. Then, in the visual coordinate system, fit the direction of the displacement axis by least squares through the trajectory of the sphere center. Solve the transformation matrix [R|T]v_r between the visual coordinate system and the reference coordinate system through the three-dimensional vector relationship.

[0023]

[0024] In the formula, The directions of the XYZ axes in the reference coordinate system, The vector fitted to the XYZ axes in the visual coordinate system. These are the spatial coordinate components of a point in the target coordinate system after rigid body transformation.

[0025] Step S23: By marking a silicon wafer of fixed thickness at the laser focal point, and using 3D vision scanning to obtain the position of the focal point in the vision system, the rigid transformation matrix [R|T]v_r between the reference coordinate system and the vision coordinate system is then used to solve for the rigid transformation matrix [R|T]r_l between the origin of the reference coordinate system and the laser processing focal point. The laser processing axis is consistent with the Z-axis direction of the machine tool, and there is only a displacement vector between the laser focal point and the reference coordinate system.

[0026]

[0027] In the formula, This represents the displacement relationship between the laser focus and the origin of the reference coordinate system.

[0028] In the aforementioned online correction method for laser processing of turbine blades based on three-dimensional vision, step 3 includes:

[0029] Step S31: Use binocular vision structured light to measure the contour of the surface to be processed at the scanning position;

[0030] Step S32: Use point cloud registration to register the theoretical model with the measured point cloud, and use the registered coordinate relationship to transform the machining coordinates in the CAD model from the model coordinate system to the visual coordinate system, so that the position to be processed is mapped to the measured point cloud;

[0031] Step S33: Construct a sequence of points to be processed {PL} based on the generated points to be processed, and sample points from it as pre-processed points {pl}.

[0032] In the aforementioned online correction method for laser processing of turbine blades based on three-dimensional vision, step S4 includes:

[0033] Step S41: Based on the pre-processed point {pl} in the vision measurement system, calculate the coordinates {pl}r in the reference coordinate system using the transformation matrix [R|T]v_r between the reference coordinate system and the vision coordinate system;

[0034] ,

[0035] Step S42: Based on the transformation matrix [R|T]r_l between the laser focus position and the origin of the reference coordinate system, transform the machining point to the laser focus position, and calculate and generate the machine tool machining coordinates {pl}l;

[0036] ,

[0037] Step S43: After calculating the motion coordinates of the pre-processing points using the computer, mark points are processed on the workpiece surface using a laser.

[0038] In the aforementioned online correction method for laser processing of turbine blades based on three-dimensional vision, step S5 includes:

[0039] Step S51: After processing black marker points on the workpiece surface using a laser, the workpiece surface contour is reconstructed again using binocular vision structured light at the scanning position, and the black marker points are simultaneously reconstructed in three dimensions. At this time, the marker points and the workpiece surface contour point cloud are in the same coordinate system. The three-dimensional reconstruction of the marker points includes:

[0040] Step S511: Detect the marker points and obtain the coordinates of the circle center in the image by fitting the circle center;

[0041] Step S512: Perform stereo matching of marker points in the image based on the minimum error of epipolar constraints. The marker points in the stereo view satisfy the following: In the formula and The coordinates of the detection center for the marker points in the right and left views of the stereo vision system. To describe the epipolar constraint relationship between corresponding points in the left and right views of a stereo vision system, the left and right view marker point pairs satisfy:

[0042] , This is the epipolar constraint error threshold.

[0043] Perform 3D reconstruction on the matching point pairs of the left and right views:

[0044] ,

[0045] In the formula, Let be the three-dimensional coordinates to be solved. These are the pixel coordinates of the left camera and the pixel coordinates of the right camera, respectively. These are the intrinsic and extrinsic parameter matrices for the left and right cameras, respectively.

[0046] Step S513: Reconstruct the marker points based on least squares, with the marker points and the workpiece surface contour point cloud in the same coordinate system;

[0047] Step S52: By using nearest neighbor search and evaluating the nearest neighbor distance between the processed point cloud and the point cloud before processing, the repeatability accuracy of the system is evaluated to ensure the consistency of the machine tool vision system.

[0048] Step S53: When the repeated measurement error of the workpiece surface contour before and after processing is less than the threshold th1, the visual guidance is effective; otherwise, it is necessary to return to step S2 and recalibrate the system.

[0049] In the aforementioned online correction method for laser processing of turbine blades based on three-dimensional vision, step S6 includes:

[0050] Step S61: When the workpiece surface contour before and after processing is consistent, directly calculate the coordinate deviation between the pre-processing point and the processing black mark point. When it is less than the threshold th2, proceed directly to formal processing.

[0051] Step S62: When the coordinate deviation between the pre-processing point and the processing marker point is greater than the threshold th2, fit the deviation function Δ(x,y,z) between the processed marker point and the theoretical coordinate in the XYZ directions according to the sampling points.

[0052]

[0053] Step S63: Based on the fitted deviation function, predict the positioning error of all points to be processed, and directly correct the formal processing position based on the error amount: Actual processing coordinates = theoretical processing coordinates - Δ(x,y,z).

[0054] A system for performing the method includes:

[0055] Five-axis linkage CNC machine tool;

[0056] A vision system, which is fixed to the Z-axis of a five-axis CNC machine tool;

[0057] The laser processing head has its focal point in the same field of view as the vision system.

[0058] The control unit is configured to perform fully automated operations, including point cloud acquisition, coordinate system calibration, machining point mapping, pre-machining control, marker point reconstruction, deviation fitting, and online correction of machining coordinates.

[0059] A computer storage medium including computer instructions that, when run on a computer, cause the computer to perform the method.

[0060] An electronic device, the electronic device comprising:

[0061] Memory, processor, and computer programs stored in memory and executable on the processor, wherein,

[0062] The processor implements the method when executing the program.

[0063] Compared with existing technologies, this invention has the following advantages: By pre-processing points and performing online measurement, evaluating the consistency between positioning and guided processing, and correcting the processing position online, this invention provides an open-loop implementation scheme for laser equipment positioning-processing-detection-correction that is highly accurate, flexible, practical, and automated. Within the accuracy range of visual measurement, it ensures the accuracy and reliability of vision-guided laser processing, effectively reducing the scrap rate caused by deviations during processing. Addressing the problem of deviations in vision-guided processing, this invention proposes a method using pre-processing points and online deviation fitting and correction; based on the principles of epipolar constraint minimum error and least squares 3D reconstruction, it performs accurate stereo matching and high-precision 3D reconstruction of marker point pairs; it uses the actual processing position deviation of the pre-processing points to fit a processing deviation function, correcting the deviation of the formal processing position of the processing sequence, reducing the impact of system errors on guided processing; and online detection of the pre-processing points ensures the accuracy and reliability of processing. To address the problem of solving the vision-machine tool coordinate transformation matrix, a high-precision, automated linear transformation method using a standard sphere is proposed. This method involves measuring the point cloud of the standard sphere using a structured light system and fitting the coordinates of the sphere's center to fit the direction of the machine tool's motion axes. By using point cloud data fitting, the impact of noise data during trajectory tracking is reduced, achieving high-precision acquisition of the machine tool's motion direction. Using a pre-defined standard sphere motion trajectory, the vision-machine tool coordinate transformation matrix can be automatically solved without manual intervention. Attached Figure Description

[0064] Various other advantages and benefits of the present invention will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiments below. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. It is obvious that the drawings described below are merely some embodiments of the invention, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort. Furthermore, the same reference numerals denote the same parts throughout the drawings.

[0065] In the attached diagram:

[0066] Figure 1 This is a flowchart of the present invention;

[0067] Figure 2 This is a schematic diagram of the equipment of the present invention;

[0068] Figure 3 This is a schematic diagram of the shaft fitting principle of the present invention;

[0069] Figure 4 This is a schematic diagram illustrating the definition of the origin of the reference coordinate system in this invention;

[0070] Figure 5This is a schematic diagram of the displacement axis fitting principle of the present invention;

[0071] Figure 6 This is a schematic diagram of the laser focus-visual coordinate relationship of the present invention;

[0072] Figure 7 This is a schematic diagram of the pre-processing point sampling of the present invention;

[0073] Figure 8 This is a schematic diagram of the principle of error reconstruction based on minimum epipolar constraint of the present invention;

[0074] Figure 9 This is a diagram illustrating the reconstruction effect of the pre-processed points according to the present invention;

[0075] Figure 10 This is a schematic diagram of the pre-processing accuracy evaluation of the present invention.

[0076] The present invention will be further explained below with reference to the accompanying drawings and embodiments. Detailed Implementation

[0077] Specific embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While specific embodiments of the invention are shown in the drawings, it should be understood that the invention may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this invention will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

[0078] It should be noted that certain terms are used in the specification and claims to refer to specific components. Those skilled in the art will understand that different terms may be used to refer to the same component. This specification and claims do not distinguish components based on differences in terminology, but rather on differences in function. The terms "comprising" or "including" used throughout the specification and claims are open-ended and should be interpreted as "comprising but not limited to." The following descriptions are preferred embodiments for carrying out the invention; however, these descriptions are for the purpose of understanding the general principles of the specification and are not intended to limit the scope of the invention. The scope of protection of this invention is determined by the appended claims.

[0079] To facilitate understanding of the embodiments of the present invention, further explanations and descriptions will be provided below with reference to the accompanying drawings and specific embodiments. The accompanying drawings do not constitute a limitation on the embodiments of the present invention.

[0080] like Figures 1 to 10 As shown, the online correction method for laser processing of turbine blades based on 3D vision includes the following steps:

[0081] Step S1: Construct a five-axis kinematic model based on the geometry of the five-axis machine tool, and define the reference coordinate system, the vision coordinate system, and the model coordinate system;

[0082] Step S2: Fix a standard ball on the motion platform of the five-axis machine tool, obtain the five-axis motion trajectory by visual measurement of the standard ball, define the center of the rotation axis as the origin of the reference coordinate system, calculate the rigid transformation matrix [R|T]v_r between the reference coordinate system and the visual coordinate system, and obtain the rigid transformation matrix [R|T]r_l between the laser processing focus position and the origin of the reference coordinate system using visual measurement;

[0083] Step S3: Use binocular vision structured light to measure the contour of the surface to be processed at the scanning position, and use point cloud registration to map the processing hole positions in the model coordinate system to the measured point cloud in the vision coordinate system to generate the processing point position sequence {PL}, and sample from it as pre-processing point positions {pl}.

[0084] Step S4: Generate five-axis machine tool machining coordinates based on the pre-machining point {pl}, rigid transformation matrix [R|T]v_r and rigid transformation matrix [R|T]r_l, and perform marker machining at the pre-machining point on the workpiece surface;

[0085] Step S5: The surface contour of the processed workpiece is reconstructed again at the scanning position using binocular vision structured light. At the same time, the marker points of the pre-processed points are reconstructed using the minimum epipolar constraint error method, so that the marker points are merged with the point cloud of the processed workpiece surface. The nearest neighbor search of the point cloud is used to evaluate the surface contour of the workpiece before and after processing. If the deviation of the surface contour of the workpiece before and after processing is less than the threshold th1, the visual guidance is effective and deviation correction is performed. Otherwise, return to step S2 to recalibrate the system.

[0086] Step S6: When the workpiece surface contour before and after machining is consistent, directly evaluate the coordinate deviation between the pre-machining point and the black marker point after machining, and fit the deviation functions ΔX(x,y,z), ΔY(x,y,z), and ΔZ(x,y,z) between the black marker point after machining and the theoretical coordinates according to the sampling points. Based on the deviation functions, predict the positioning error of all points to be machined and directly correct it in the formal machining.

[0087] In a preferred embodiment of the online correction method for laser processing of turbine blades based on three-dimensional vision, step S1 includes:

[0088] Step S11: The three-dimensional measurement uses a structured light measurement system in stereo vision;

[0089] Step S12: Construct a five-axis kinematic model (X, Y, Z, C1, C2) based on the geometry of the five-axis machine tool, and define the intersection of axis C1 and axis C2 as the origin of the reference coordinate system. The direction of the reference coordinate system is consistent with the axis (X, Y, Z) of the five-axis machine tool.

[0090] Step S13: Define the visual coordinate system and measurement position. The vision module is fixed to the Z-axis of the five-axis machine tool. The center of the optimal depth range of the vision measurement system relative to the reference coordinate system in the machine tool is taken as the measurement position. The visual coordinate system is the camera coordinate system after binocular vision calibration.

[0091] Step S14: Using the design coordinate system of the workpiece model as the model coordinate system, register the workpiece CAD model with the measured point cloud of the workpiece, and map the machining points {PL}d in the model to the measured point cloud using the nearest neighbor search method to generate the sequence of points {PL} to be machined.

[0092] {PL}=[R|T]*{PL}d, where [R|T] is the coordinate transformation matrix between the workpiece CAD model and the measured point cloud of the workpiece.

[0093] In a preferred embodiment of the online correction method for laser processing of turbine blades based on three-dimensional vision, step S2 includes:

[0094] Step S21: The standard ball is fixed on the motion axis. During the motion of the standard ball with the rotation axis, the structured light measurement system synchronously measures the pre-given trajectory of the standard ball and fits the coordinates of the ball's center. By fitting the spatial center (x, y, z) and the axis direction (n1, n2, n3), two sets of spatial straight line expressions corresponding to the rotation axes C1 and C2 are constructed. Then, the least squares solution is used to solve the intersection point of the spatial straight lines as the origin of the reference coordinate system, and the direction of the reference coordinate system (X, Y, Z) is defined to be consistent with the machine tool motion.

[0095] In the formula, Indicates the first Group 1 Given a set of spatial plane equation parameters, two intersecting planes represent a spatial straight line. Combined with the given spatial center of the fitted circle, (x,y,z) are the coordinates of the intersection point of the two spatial straight lines.

[0096] Step S22: Move the machine tool and visually measure the position of the center of the standard sphere. Then, in the visual coordinate system, fit the direction of the displacement axis by least squares through the trajectory of the sphere center. Solve the transformation matrix [R|T]v_r between the visual coordinate system and the reference coordinate system through the three-dimensional vector relationship.

[0097]

[0098] In the formula, The directions of the XYZ axes in the reference coordinate system, The vector fitted to the XYZ axes in the visual coordinate system. These are the spatial coordinate components of a point in the target coordinate system after rigid body transformation.

[0099] Step S23: By marking a silicon wafer of fixed thickness at the laser focal point, and using 3D vision scanning to obtain the position of the focal point in the vision system, the rigid transformation matrix [R|T]v_r between the reference coordinate system and the vision coordinate system is then used to solve for the rigid transformation matrix [R|T]r_l between the origin of the reference coordinate system and the laser processing focal point. The laser processing axis is consistent with the Z-axis direction of the machine tool, and there is only a displacement vector between the laser focal point and the reference coordinate system.

[0100]

[0101] In the formula, This represents the displacement relationship between the laser focus and the origin of the reference coordinate system.

[0102] In a preferred embodiment of the online correction method for laser processing of turbine blades based on three-dimensional vision, step 3 includes:

[0103] Step S31: Use binocular vision structured light to measure the contour of the surface to be processed at the scanning position;

[0104] Step S32: Use point cloud registration to register the theoretical model with the measured point cloud, and use the registered coordinate relationship to transform the machining coordinates in the CAD model from the model coordinate system to the visual coordinate system, so that the position to be processed is mapped to the measured point cloud;

[0105] Step S33: Construct a sequence of points to be processed {PL} based on the generated points to be processed, and sample points from it as pre-processed points {pl}.

[0106] In a preferred embodiment of the online correction method for laser processing of turbine blades based on three-dimensional vision, step S4 includes:

[0107] Step S41: Based on the pre-processed point {pl} in the vision measurement system, calculate the coordinates {pl}r in the reference coordinate system using the transformation matrix [R|T]v_r between the reference coordinate system and the vision coordinate system;

[0108] ,

[0109] Step S42: Based on the transformation matrix [R|T]r_l between the laser focus position and the origin of the reference coordinate system, transform the machining point to the laser focus position, and calculate and generate the machine tool machining coordinates {pl}l;

[0110] ,

[0111] Step S43: After calculating the motion coordinates of the pre-processing points using the computer, mark points are processed on the workpiece surface using a laser.

[0112] In a preferred embodiment of the online correction method for laser processing of turbine blades based on three-dimensional vision, step S5 includes:

[0113] Step S51: After processing black marker points on the workpiece surface using a laser, the workpiece surface contour is reconstructed again using binocular vision structured light at the scanning position, and the black marker points are simultaneously reconstructed in three dimensions. At this time, the marker points and the workpiece surface contour point cloud are in the same coordinate system. The three-dimensional reconstruction of the marker points includes:

[0114] Step S511: Detect the marker points and obtain the coordinates of the circle center in the image by fitting the circle center;

[0115] Step S512: Perform stereo matching of marker points in the image based on the minimum error of epipolar constraints. The marker points in the stereo view satisfy the following: In the formula and The coordinates of the detection center for the marker points in the right and left views of the stereo vision system. To describe the epipolar constraint relationship between corresponding points in the left and right views of a stereo vision system, the left and right view marker point pairs satisfy:

[0116] , This is the epipolar constraint error threshold.

[0117] Perform 3D reconstruction on the matching point pairs of the left and right views:

[0118] ,

[0119] In the formula, Let be the three-dimensional coordinates to be solved. These are the pixel coordinates of the left camera and the pixel coordinates of the right camera, respectively. These are the intrinsic and extrinsic parameter matrices for the left and right cameras, respectively.

[0120] Step S513: Reconstruct the marker points based on least squares, with the marker points and the workpiece surface contour point cloud in the same coordinate system;

[0121] Step S52: By using nearest neighbor search and evaluating the nearest neighbor distance between the processed point cloud and the point cloud before processing, the repeatability accuracy of the system is evaluated to ensure the consistency of the machine tool vision system.

[0122] Step S53: When the repeated measurement error of the workpiece surface contour before and after processing is less than the threshold th1, the visual guidance is effective; otherwise, it is necessary to return to step S2 and recalibrate the system.

[0123] In a preferred embodiment of the online correction method for laser processing of turbine blades based on three-dimensional vision, step S6 includes:

[0124] Step S61: When the workpiece surface contour before and after processing is consistent, directly calculate the coordinate deviation between the pre-processing point and the processing black mark point. When it is less than the threshold th2, proceed directly to formal processing.

[0125] Step S62: When the coordinate deviation between the pre-processing point and the processing marker point is greater than the threshold th2, fit the deviation function Δ(x,y,z) between the processed marker point and the theoretical coordinate in the XYZ directions according to the sampling points.

[0126]

[0127] Step S63: Based on the fitted deviation function, predict the positioning error of all points to be processed, and directly correct the formal processing position based on the error amount: Actual processing coordinates = theoretical processing coordinates - Δ(x,y,z).

[0128] In a preferred embodiment of the online correction method for laser processing of turbine blades based on three-dimensional vision, the three-dimensional vision system is a binocular structured light measurement system, which is installed on the Z-axis of a five-axis machine tool. Its measurement position is located at the depth center of the three-dimensional vision system and can completely cover the area to be processed on the blade. The marker points formed by pre-processing are black micro-holes or pits generated by laser ablation, which have high grayscale contrast and are convenient for the vision system to accurately identify and locate the center at the sub-pixel level.

[0129] A system for performing the method includes:

[0130] Five-axis linkage CNC machine tool;

[0131] A vision system, which is fixed to the Z-axis of a five-axis CNC machine tool;

[0132] The laser processing head has its focal point in the same field of view as the vision system.

[0133] The control unit is configured to perform fully automated operations, including point cloud acquisition, coordinate system calibration, machining point mapping, pre-machining control, marker point reconstruction, deviation fitting, and online correction of machining coordinates.

[0134] A computer storage medium including computer instructions that, when run on a computer, cause the computer to perform the method.

[0135] An electronic device, the electronic device comprising:

[0136] Memory, processor, and computer programs stored in memory and executable on the processor, wherein,

[0137] The processor implements the method when executing the program.

[0138] In one embodiment, step S5, the three-dimensional reconstruction of the marker points includes:

[0139] Detect the marker region in the left and right camera images and fit its center pixel coordinates;

[0140] Based on the epipolar geometric constraints, search in the right figure for the point that satisfies the minimum epipolar error with the marker point in the left figure;

[0141] For successfully matched point pairs, triangulation reconstruction is performed using the intrinsic and extrinsic parameter matrices of the binocular camera to obtain the three-dimensional coordinates of the marker points in the visual coordinate system.

[0142] In step S6, if the average deviation between the pre-processed marker point and the theoretical position is less than the second threshold, the error correction is skipped and the formal processing is performed directly; otherwise, based on the deviation data of all sampling points, a continuous spatial deviation field function is constructed using polynomial fitting or radial basis function interpolation.

[0143] The three-dimensional vision system is a binocular structured light measurement system, installed on the Z-axis of the machine tool. Its measurement position is located at the depth center of the vision system and can completely cover the area of ​​the blade to be processed.

[0144] In step S5, the Hausdorff distance or root mean square deviation of the point cloud before and after processing is calculated by nearest neighbor search. If the deviation is greater than the preset threshold, the system calibration is determined to be invalid, and the process returns to step S2 to recalibrate the vision-machine coordinate system.

[0145] In one embodiment, the method includes,

[0146] First, a binocular structured light measurement system is used to acquire blade surface data online, and the theoretical processing points are mapped to the surface to be processed.

[0147] Subsequently, a coding calibration board / standard ball is used to automatically calculate the coordinate system transformation relationship between the five-axis machining equipment and the vision system;

[0148] Then, sampling is performed on the theoretical processing point sequence, and the sampled points are pre-processed to produce clearly contrasting feature points.

[0149] Next, the surface contour and feature point coordinates of the blade after pre-processing are measured online using three-dimensional stereo vision, and then converted to the coordinate system of the point cloud measured before processing to further calculate the deviation from the theoretical processing coordinates.

[0150] Finally, based on the deviation between the measured values ​​and theoretical values ​​of the pre-processed feature points, a processing error correction function is constructed to achieve online positioning error correction for vision-guided laser processing.

[0151] In one embodiment, the online correction method for laser processing guidance and positioning errors of aero-engine blades based on 3D vision includes,

[0152] Step S1: First, construct a five-axis kinematic model based on the known geometry of the five-axis machine tool, and define the reference coordinate system, the vision coordinate system, and the model coordinate system. The vision measurement system can simultaneously capture the motion trajectories of the five motion axes of the machine tool within a certain range.

[0153] Step S2: Fix a standard ball on the motion platform of the machine tool, obtain the five-axis motion trajectory by visual measurement of the standard ball, define the center of the rotation axis as the origin of the reference coordinate system, calculate the rigid transformation matrix [R|T]v_r between the reference coordinate system and the visual coordinate system, and obtain the rigid transformation matrix [R|T]r_l between the laser processing focal position and the origin of the reference coordinate system by visual measurement.

[0154] Step S3: Use binocular vision structured light to measure the contour of the surface to be processed at the scanning position, and use point cloud registration to map the processing hole positions in the model coordinate system to the measured point cloud in the vision coordinate system to generate the processing point position sequence {PL}, and sample from it as pre-processing point positions {pl}.

[0155] Step S4: Based on the pre-processing points in the vision measurement system, and the transformation matrix [R|T]v_r between the reference coordinate system and the vision coordinate system, and the rigid transformation matrix [R|T]r_l between the laser processing focus position and the origin of the reference coordinate system, generate the machine tool processing coordinates, and perform mark point processing on the pre-processing points on the workpiece surface.

[0156] Step S5: The surface contour of the processed workpiece is reconstructed again at the scanning position using binocular vision structured light. At the same time, the marker points of the pre-processing points are reconstructed with high precision using the method based on minimum epipolar constraint error, so that the marker points are merged with the point cloud of the processed workpiece surface. The nearest neighbor search of the point cloud is used to evaluate the surface contour of the workpiece before and after processing. If the deviation of the surface contour of the workpiece before and after processing is small, the visual guidance is effective and deviation correction is performed. Otherwise, return to step S2 to recalibrate the system.

[0157] Step S6: When the workpiece surface contour before and after machining is consistent, directly evaluate the coordinate deviation between the pre-machining point and the black marker point after machining, and fit the deviation functions ΔX(x,y,z), ΔY(x,y,z), and ΔZ(x,y,z) between the black marker point after machining and the theoretical coordinates according to the sampling points. Based on the deviation functions, predict the positioning error of all points to be machined and directly correct it in the formal machining.

[0158] like Figure 2 As shown, step S1 includes:

[0159] Step S11: A laser processing guidance equipment for aero-engine blades based on three-dimensional vision mainly includes three parts: three-dimensional measurement, laser processing, and machine tool body. The three-dimensional measurement adopts a structured light measurement system in stereo vision.

[0160] Step S12: Based on the known machine tool geometry, construct a five-axis kinematic model (X, Y, Z, C1, C2), and define the intersection of axes C1 and C2 as the origin of the reference coordinate system. The direction of the reference coordinate system is consistent with the machine tool's (X, Y, Z) axes.

[0161] Step S13: Define the visual coordinate system and measurement position. The vision module is fixed to the Z-axis of the machine tool. The center of the optimal depth range of the vision measurement system relative to the reference coordinate system in the machine tool is taken as the measurement position. The visual coordinate system is the camera coordinate system after binocular vision calibration.

[0162] Step S14: Using the design coordinate system of the workpiece model as the model coordinate system, register the workpiece CAD model with the measured point cloud of the workpiece, and map the machining points {PL}d in the model to the measured point cloud using the nearest neighbor search method to generate the sequence of points {PL} to be machined.

[0163] {PL}=[R|T]*{PL}d

[0164] In the formula, [R|T] is the transformation matrix between the workpiece CAD model and the measured point cloud coordinates of the workpiece.

[0165] Step S2 specifically includes:

[0166] Step S21: As Figure 3 As shown, a standard sphere is fixed on the motion axis. Within the effective measurement range of the vision system, as the standard sphere moves with the rotation axis, the structured light measurement system synchronously measures the pre-given trajectory of the standard sphere and fits the coordinates of the sphere's center. Figure 4 As shown, by fitting the spatial center (x,y,z) and the axis direction (n1,n2,n3), two sets of spatial straight line expressions corresponding to the rotation axis C1 and C2 are constructed. Then, the least squares solution is used to solve the intersection point of the spatial straight lines as the origin of the reference coordinate system, and the direction of the reference coordinate system (X,Y,Z) is defined to be consistent with the machine tool motion.

[0167]

[0168] In the formula, Indicates the first Group 1 Given a set of spatial plane equation parameters, two intersecting planes can represent a spatial straight line. Combined with the given spatial center of the fitted circle, (x,y,z) are the coordinates of the intersection point of the two spatial straight lines.

[0169] Step S22: As Figure 5 As shown, within the effective visual measurement range, the machine tool is moved and the position of the center of the standard sphere is visually measured. Then, the displacement axis direction is fitted by least squares through the trajectory of the sphere center in the visual coordinate system. The transformation matrix [R|T]v_r between the visual coordinate system and the reference coordinate system is solved by the three-dimensional vector relationship.

[0170]

[0171] In the formula, The directions of the XYZ axes in the reference coordinate system, The vector fitted to the XYZ axes in the visual coordinate system. These are the spatial coordinate components of a point in the target coordinate system after rigid body transformation.

[0172] Step S23: As Figure 6 As shown, a silicon wafer of fixed thickness is marked at the laser focal point, and the position of the focal point in the vision system is obtained by three-dimensional vision scanning. Then, the rigid transformation matrix [R|T]v_r between the reference coordinate system and the vision coordinate system is combined to solve the rigid transformation matrix [R|T]r_l between the origin of the reference coordinate system and the laser processing focal point. Here, the laser processing axis is defined to be consistent with the Z-axis of the machine tool. Therefore, there is only a displacement vector between the laser focal point and the reference coordinate system.

[0173]

[0174] In the formula, This represents the displacement relationship between the laser focus and the origin of the reference coordinate system.

[0175] Step 3 specifically includes:

[0176] Step S31: Use binocular vision structured light to measure the contour of the surface to be processed at the scanning position;

[0177] Step S32: As Figure 7 As shown, point cloud registration is used to register the theoretical model with the measured point cloud. Then, the coordinate relationship after precise registration is used to transform the machining coordinates in the CAD model from the model coordinate system to the visual coordinate system, thus mapping the machining position to be processed into the measured point cloud. Figure 7 In the diagram, green dots represent points to be processed.

[0178] Step S33: Construct a sequence of points to be processed {PL} based on the generated points to be processed, and sample points from it as pre-processing points {pl}, trying to make the sampling points cover the entire area to be processed. Figure 7 In the diagram, the red dashed line represents the selected pre-processing point.

[0179] Step S4 specifically includes:

[0180] Step S41: Based on the pre-processed point {pl} in the vision measurement system, calculate the coordinates {pl}r in the reference coordinate system using the transformation matrix [R|T]v_r between the reference coordinate system and the vision coordinate system;

[0181]

[0182] Step S42: Further transform the machining point to the laser focus position according to the transformation matrix [R|T]r_l between the laser focus position and the origin of the reference coordinate system, and calculate and generate the machine tool machining coordinates {pl}l;

[0183]

[0184] Step S43: After calculating the motion coordinates of the pre-processing points using the computer, mark points are processed on the workpiece surface using a laser.

[0185] Step S5 specifically includes:

[0186] Step S51: After processing black marker points on the workpiece surface using laser, the workpiece surface contour is reconstructed again using binocular vision structured light at the scanning position, and the black marker points are reconstructed in three dimensions at the same time. At this time, the marker points and the workpiece surface contour point cloud are in the same coordinate system.

[0187] In step S51, the method for three-dimensional reconstruction of the marker points specifically includes:

[0188] Step S511: Detect the marker points and obtain the coordinates of the circle center in the image by fitting the circle center;

[0189] Step S512: As Figure 8 As shown, stereo matching of marker points in an image is performed based on the minimum error of epipolar constraints. The marker points in the stereo view should satisfy the following:

[0190]

[0191] In the formula and The coordinates of the detection center for the marker points in the right and left views of the stereo vision system. This is the fundamental matrix that describes the epipolar constraint relationship between corresponding points in the left and right views of a stereo vision system.

[0192] In actual matching, due to system errors and image noise, However, a single marker point in the left view may simultaneously exist in multiple marker points in the right view, satisfying the condition. , This is the threshold for the epipolar constraint error; at this point, the left and right marker point pairs with the smallest limit constraint error are taken as the stereo matching result, i.e., the corresponding left and right view marker points satisfy:

[0193]

[0194] Perform 3D reconstruction on the matching point pairs of the left and right views:

[0195]

[0196] In the formula, Let be the three-dimensional coordinates to be solved. These are the pixel coordinates of the left camera and the pixel coordinates of the right camera, respectively. These are the intrinsic and extrinsic parameter matrices for the left and right cameras, respectively.

[0197] Step S513: Reconstruct markers based on least squares, such as... Figure 9 As shown, this is the reconstruction effect of the marker points and the workpiece surface point cloud after processing. At this time, the marker points and the workpiece surface contour point cloud are in the same coordinate system.

[0198] Step S52: As Figure 10 As shown, under the same measurement position, the point clouds on the workpiece surface before and after processing should completely overlap. However, due to systematic errors, there will be deviations between the theoretical processing position and the actual processing position during actual processing. By using nearest neighbor point search and evaluating the nearest neighbor distance between the point cloud after processing and the point cloud before processing, the repeatability accuracy of the system can be evaluated to ensure the consistency of the machine tool vision system.

[0199] Step S53: When the average deviation of the workpiece surface contour before and after processing is small, that is, when the repeatability measurement error is less than the threshold th1 (<0.01mm), the visual guidance is effective; otherwise, it is necessary to return to step S2 and recalibrate the system.

[0200] Step S6 specifically includes:

[0201] Step S61: When the workpiece surface contour before and after processing is consistent, directly calculate the coordinate deviation between the pre-processing point and the processing black mark point. When it is less than the threshold th2 (<0.02mm), no error correction is required, and formal processing can be carried out directly.

[0202] Step S62: When the coordinate deviation between the pre-processing point and the processing marker point is greater than the threshold th2, fit the deviation function Δ(x,y,z) between the processed marker point and the theoretical coordinate in the XYZ directions according to the sampling points.

[0203]

[0204] Step S63: Based on the fitted deviation function, predict the positioning error of all points to be processed, and directly correct the actual processing position based on the error amount.

[0205] Actual machining coordinates = theoretical machining coordinates - Δ(x,y,z).

[0206] Furthermore, the automatic coordinate system calibration mechanism based on a standard sphere in this invention effectively solves the problem of high-precision spatial alignment between the vision system and the five-axis machine tool. Traditional methods rely on manual intervention or offline calibration, which is difficult to adapt to dynamic errors such as thermal deformation and mechanical drift during processing. This invention, however, controls the standard sphere to move along a known trajectory (including rotation axes C1 / C2 and linear axes), and uses a structured light system to collect the sphere's center point cloud in real time. Combined with spatial circle fitting and least-squares line intersection calculation, it automatically and robustly calculates the rigid transformation matrix between the vision coordinate system and the machine tool reference coordinate system, while accurately calibrating the position of the laser focus relative to this reference system. This process requires no external measuring equipment, achieving "in-machine self-calibration," significantly improving the long-term stability and repeatability of the system. The combination of pre-processed feature markers and high-precision 3D reconstruction technology constructs an observable channel for processing errors. Since the surface of turbine blades is a complex free-form surface, directly evaluating the deviation of laser processing points is extremely difficult. This invention innovatively samples theoretical processing points for low-energy pre-processing, forming high-contrast micro-marker points (such as black ablation spots), which are then scanned across the entire field using a binocular structured light system. Specifically, for stereo matching of these marker points, a matching strategy based on minimum epipolar constraint error is introduced, enabling sub-pixel-level center localization and highly reliable 3D reconstruction even in the presence of image noise and occlusion. This transforms the previously "invisible" processing positioning error into a "measurable and quantifiable" spatial coordinate deviation, providing a reliable data foundation for subsequent corrections.

[0207] An online error compensation mechanism based on full-field point cloud registration and deviation function fitting overcomes the limitations of traditional point-to-point correction. Unlike methods that rely on only a few sampling points for rigid body pose adjustment, this invention utilizes deviation data from multiple pre-processing points distributed throughout the entire processing area. Through spatial interpolation (such as radial basis functions or polynomial fitting), a continuous three-dimensional deviation field function Δ(x,y,z) is constructed, thereby enabling position prediction and dynamic correction for all points to be processed. This method not only compensates for systematic errors (such as calibration residuals and optical distortion) but also partially suppresses local nonlinear errors (such as micro-deformation during blade clamping and thermal expansion of materials), significantly improving the consistency and accuracy of the entire processing. A point cloud consistency verification mechanism before and after processing constitutes a self-diagnostic closed loop for the system. By comparing the surface point clouds of the same area before and after pre-processing and calculating their nearest neighbor distance statistics (such as root mean square deviation), the reliability of the vision-machine tool system can be determined in real time. If the deviation exceeds a threshold, a recalibration process is automatically triggered to prevent batch processing failures due to system drift. This online feedback logic of "measurement-processing-remeasurement-correction" enables the entire laser processing process to have a near-closed-loop control capability, achieving highly reliable precision manufacturing even on open-loop laser equipment.

[0208] Although embodiments of the present invention have been described above in conjunction with the accompanying drawings, the present invention is not limited to the specific embodiments and application fields described above. The specific embodiments described above are merely illustrative and instructive, and not restrictive. Those skilled in the art can make many other forms based on the guidance of this specification and without departing from the scope of protection of the claims of the present invention, and all of these are within the scope of protection of the present invention.

Claims

1. A method for online correction of turbine blades using laser processing based on three-dimensional vision, characterized in that, Includes the following steps: Step S1: Construct a five-axis kinematic model based on the geometry of the five-axis machine tool, and define the reference coordinate system, the vision coordinate system, and the model coordinate system; Step S2: Fix a standard ball on the motion platform of the five-axis machine tool, obtain the five-axis motion trajectory by visual measurement of the standard ball, define the center of the rotation axis as the origin of the reference coordinate system, calculate the rigid transformation matrix [R|T]v_r between the reference coordinate system and the visual coordinate system, and obtain the rigid transformation matrix [R|T]r_l between the laser processing focus position and the origin of the reference coordinate system using visual measurement; Step S3: Use binocular vision structured light to measure the contour of the surface to be processed at the scanning position, and use point cloud registration to map the processing hole positions in the model coordinate system to the measured point cloud in the vision coordinate system to generate the processing point position sequence {PL}, and sample from it as pre-processing point positions {pl}. Step S4: Generate five-axis machine tool machining coordinates based on the pre-machining point {pl}, rigid transformation matrix [R|T]v_r and rigid transformation matrix [R|T]r_l, and perform marker machining at the pre-machining point on the workpiece surface; Step S5: The surface contour of the processed workpiece is reconstructed again at the scanning position using binocular vision structured light, and the marker points of the pre-processed points are reconstructed at the same time, so that the marker points are merged with the point cloud of the surface of the processed workpiece; and the surface contour of the workpiece before and after processing is searched and evaluated. If the deviation of the surface contour of the workpiece before and after processing is less than the threshold th1, the visual guidance is effective and deviation correction is performed; otherwise, return to step S2 for recalibration. Step S6: When the workpiece surface contour before and after machining is consistent, directly evaluate the coordinate deviation between the pre-machining point and the marker point after machining, and fit the deviation functions ΔX(x,y,z), ΔY(x,y,z), and ΔZ(x,y,z) between the marker point after machining and the theoretical coordinates according to the sampling points. Based on the deviation functions, predict the positioning error of all points to be machined and directly correct it in the formal machining.

2. The online correction method for laser processing of turbine blades based on three-dimensional vision according to claim 1, characterized in that, Preferably, step S1 includes: Step S11: The three-dimensional measurement uses a structured light measurement system in stereo vision; Step S12: Construct a five-axis kinematic model (X, Y, Z, C1, C2) based on the geometry of the five-axis machine tool, and define the intersection of axis C1 and axis C2 as the origin of the reference coordinate system. The direction of the reference coordinate system is consistent with the axis (X, Y, Z) of the five-axis machine tool. Step S13: Define the visual coordinate system and measurement position. The vision module is fixed to the Z-axis of the five-axis machine tool. The center of the optimal depth range of the vision measurement system relative to the reference coordinate system in the machine tool is taken as the measurement position. The visual coordinate system is the camera coordinate system after binocular vision calibration. Step S14: Using the design coordinate system of the workpiece model as the model coordinate system, register the workpiece CAD model with the measured point cloud of the workpiece, and map the machining points {PL}d in the model to the measured point cloud using the nearest neighbor search method to generate the sequence of points {PL} to be machined. {PL}=[R|T]*{PL}d, where [R|T] is the coordinate transformation matrix between the workpiece CAD model and the measured point cloud of the workpiece.

3. The online correction method for laser processing of turbine blades based on three-dimensional vision according to claim 1, characterized in that, Step S2 includes: Step S21: The standard ball is fixed on the motion axis. During the motion of the standard ball with the rotation axis, the structured light measurement system synchronously measures the pre-given trajectory of the standard ball and fits the coordinates of the ball's center. By fitting the spatial center (x, y, z) and the axis direction (n1, n2, n3), two sets of spatial straight line expressions corresponding to the rotation axes C1 and C2 are constructed. Then, the least squares solution is used to solve the intersection point of the spatial straight lines as the origin of the reference coordinate system, and the direction of the reference coordinate system (X, Y, Z) is defined to be consistent with the machine tool motion. In the formula, Indicates the first Group 1 Given a set of spatial plane equation parameters, two intersecting planes represent a spatial straight line. Combined with the given spatial center of the fitted circle, (x,y,z) are the coordinates of the intersection point of the two spatial straight lines. Step S22: Move the machine tool and visually measure the position of the center of the standard sphere. Then, in the visual coordinate system, fit the direction of the displacement axis by least squares through the trajectory of the sphere center. Solve the transformation matrix [R|T]v_r between the visual coordinate system and the reference coordinate system through the three-dimensional vector relationship. ， In the formula, The directions of the XYZ axes in the reference coordinate system, The vector fitted to the XYZ axes in the visual coordinate system. These are the spatial coordinate components of a point in the target coordinate system after rigid body transformation. Step S23: By marking a silicon wafer of fixed thickness at the laser focal point, and using 3D vision scanning to obtain the position of the focal point in the vision system, the rigid transformation matrix [R|T]v_r between the reference coordinate system and the vision coordinate system is then used to solve for the rigid transformation matrix [R|T]r_l between the origin of the reference coordinate system and the laser processing focal point. The laser processing axis is consistent with the Z-axis direction of the machine tool, and there is only a displacement vector between the laser focal point and the reference coordinate system. ， In the formula, This represents the displacement relationship between the laser focus and the origin of the reference coordinate system.

4. The online correction method for laser processing of turbine blades based on three-dimensional vision according to claim 1, characterized in that, Step 3 includes: Step S31: Use binocular vision structured light to measure the contour of the surface to be processed at the scanning position; Step S32: Use point cloud registration to register the theoretical model with the measured point cloud, and use the registered coordinate relationship to transform the machining coordinates in the CAD model from the model coordinate system to the visual coordinate system, so that the position to be processed is mapped to the measured point cloud; Step S33: Construct a sequence of points to be processed {PL} based on the generated points to be processed, and sample points from it as pre-processed points {pl}.

5. The online correction method for laser processing of turbine blades based on three-dimensional vision according to claim 1, characterized in that, Step S4 includes: Step S41: Based on the pre-processed point {pl} in the vision measurement system, calculate the coordinates {pl}r in the reference coordinate system using the transformation matrix [R|T]v_r between the reference coordinate system and the vision coordinate system; ， Step S42: Based on the transformation matrix [R|T]r_l between the laser focus position and the origin of the reference coordinate system, transform the machining point to the laser focus position, and calculate and generate the machine tool machining coordinates {pl}l; ， Step S43: After calculating the motion coordinates of the pre-processing points using the computer, mark points are processed on the workpiece surface using a laser.

6. The online correction method for laser processing of turbine blades based on three-dimensional vision according to claim 1, characterized in that, Step S5 includes: Step S51: After processing black marker points on the workpiece surface using a laser, the workpiece surface contour is reconstructed again using binocular vision structured light at the scanning position, and the black marker points are simultaneously reconstructed in three dimensions. At this time, the marker points and the workpiece surface contour point cloud are in the same coordinate system. The three-dimensional reconstruction of the marker points includes: Step S511: Detect the marker points and obtain the coordinates of the circle center in the image by fitting the circle center; Step S512: Perform stereo matching of marker points in the image based on the minimum error of epipolar constraints. The marker points in the stereo view satisfy the following: In the formula and The coordinates of the detection center for the marker points in the right and left views of the stereo vision system. To describe the epipolar constraint relationship between corresponding points in the left and right views of a stereo vision system, the left and right view marker pairs satisfy: , This is the epipolar constraint error threshold; Perform 3D reconstruction on the matching point pairs of the left and right views: ， In the formula, Let be the three-dimensional coordinates to be solved. These are the pixel coordinates of the left camera and the pixel coordinates of the right camera, respectively. These are the intrinsic and extrinsic parameter matrices for the left and right cameras, respectively. Step S513: Reconstruct the marker points based on least squares, with the marker points and the workpiece surface contour point cloud in the same coordinate system; Step S52: By using nearest neighbor search and evaluating the nearest neighbor distance between the processed point cloud and the point cloud before processing, the repeatability accuracy of the system is evaluated to ensure the consistency of the machine tool vision system. Step S53: When the repeated measurement error of the workpiece surface contour before and after processing is less than the threshold th1, the visual guidance is effective; otherwise, it is necessary to return to step S2 and recalibrate the system.

7. The online correction method for laser processing of turbine blades based on three-dimensional vision according to claim 1, characterized in that, Step S6 includes: Step S61: When the workpiece surface contour before and after processing is consistent, directly calculate the coordinate deviation between the pre-processing point and the processing black mark point. When it is less than the threshold th2, proceed directly to formal processing. Step S62: When the coordinate deviation between the pre-processing point and the processing marker point is greater than the threshold th2, fit the deviation function Δ(x,y,z) between the processed marker point and the theoretical coordinate in the XYZ directions according to the sampling points. ， Step S63: Based on the fitted deviation function, predict the positioning error of all points to be processed, and directly correct the formal processing position based on the error amount: Actual processing coordinates = theoretical processing coordinates - Δ(x,y,z).

8. A system for performing the method as described in any one of claims 1-7, characterized in that, It includes: Five-axis linkage CNC machine tool; A vision system, which is fixed to the Z-axis of a five-axis CNC machine tool; The laser processing head has its focal point in the same field of view as the vision system. The control unit is configured to perform fully automated operations, including point cloud acquisition, coordinate system calibration, machining point mapping, pre-machining control, marker point reconstruction, deviation fitting, and online correction of machining coordinates.

9. A computer storage medium, characterized in that, The storage medium includes computer instructions that, when executed on a computer, cause the computer to perform the method as described in any one of claims 1-7.

10. An electronic device, characterized in that, The electronic device includes: Memory, processor, and computer programs stored in memory and executable on the processor, wherein, When the processor executes the program, it implements the method as described in any one of claims 1-7.

Images