A deterministic robotic grinding and polishing method for blades based on hybrid B-spline curves
By using trajectory planning and speed management based on hybrid B-spline curves, the problems of low efficiency and poor consistency in blade grinding and polishing were solved, achieving efficient and high-precision blade grinding and polishing, and improving the performance and lifespan of the steam turbine.
Patent Information
- Application Number
- CN202511530673.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-24
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2045-10-24
AI Technical Summary
Traditional manual blade polishing suffers from low efficiency and poor quality consistency. In particular, over-polishing or under-polishing is prone to occur in the transition zone at the blade edge, affecting the turbine's performance and service life.
Hybrid B-spline curves are used for grinding and polishing trajectory planning. By sharing spline curve parameters to manage velocity boundaries, deterministic grinding and polishing of velocity is achieved, improving fitting accuracy and consistency.
This improved the efficiency and quality of blade polishing, reduced over-polishing or under-polishing, ensured the overall consistency of blade polishing, and enhanced the working performance and service life of the steam turbine.
Smart Images

Figure CN121245648B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of intelligent manufacturing and industrial robot technology, specifically relating to a deterministic robotic grinding and polishing method for blades based on hybrid B-spline curves. Background Technology
[0002] Steam turbine blades are key components of steam turbines in marine propulsion systems, and their manufacturing quality directly determines the turbine's performance and service life. Traditional manual polishing operations suffer from low efficiency and inconsistent quality, making robotic polishing the mainstream approach due to its flexibility and ease of configuration. For the complex curved surfaces of blades, the curvature changes dramatically across a polishing section. Areas with high curvature or large curvature variations, especially the blade edge transition zone, often experience over-polishing or under-polishing. Therefore, the overall polishing trajectory planning and speed configuration are crucial for improving the quality and efficiency of polishing operations. Summary of the Invention
[0003] To address the aforementioned problems, the present invention aims to propose a deterministic robotic blade grinding and polishing method based on hybrid B-spline curves. This invention innovatively proposes a method to improve the fitting accuracy of the grinding and polishing trajectory to the original model by introducing hybrid B-spline curves. By sharing spline curve parameters to manage the velocity boundaries of different curve segments and pre-setting regional velocities, deterministic blade grinding and polishing is achieved.
[0004] The specific technical solution is as follows:
[0005] A deterministic robotic grinding and polishing method for blades based on hybrid B-spline curves includes the following steps:
[0006] Step 1: Determine the distribution interval of the turbine blade profile section using SolidWorks software, extract the two-dimensional profile section line in CAD and process irrelevant lines;
[0007] Step 2: In CAD, find the projection of the tool coordinate system origin onto each blade profile section;
[0008] Step 3: In the projected coordinate system, set the interpolation points for the cubic non-uniform rational B-spline curve for each contour section line. By adjusting and supplementing the interpolation points, improve the fitting accuracy between the fitted curve and the contour section line, and obtain the trajectory curve of the cubic B-spline curve prefit.
[0009] Step 4: Perform curvature analysis on the cumulative arc length of the prefitted trajectory curve, calculate the average curvature, determine the high curvature threshold, and select the 20% convex surface region or the 10% concave surface region with the largest curvature change as the high curvature region.
[0010] Step 5: Perform 5-fold non-uniform rational B-spline curve interpolation fitting on the high curvature region, and keep the B-spline curve fitting on the other regions at 3-fold, to obtain the trajectory curve of the mixed B-spline curve fitting.
[0011] Step 6: Perform uniform resampling of the trajectory curve fitted by the mixed B-spline curve with equal arc length to determine the final programming grinding and polishing contact point;
[0012] Step 7: Using the curvature information of the trajectory curve, preset the data points differently according to different curvatures, manage the speed boundary, and obtain the speed distribution curve of the grinding and polishing trajectory tool contact point and the processing trajectory with a determined speed.
[0013] Further, the specific operation process of step 1 is as follows: import the three-dimensional model of the turbine blade into SolidWorks software, determine the distribution interval of the outline section according to the blade length and the width of the belt sander, set the section lines, export the two-dimensional drawing, process the two-dimensional drawing in CAD, remove irrelevant lines, and only retain the outline section lines.
[0014] Further, the specific operation process of step 2 is as follows: According to the tool coordinate system predetermined in CAD, set a tiny circle with a diameter of 0.2-1mm with the origin of the coordinate system as the center, and pass through the entire blade along the Z-axis of the coordinate system. The center of the circle on each cross section is the projection of the origin of the tool coordinate system on each cross section.
[0015] Furthermore, the specific operation process of step 3 is as follows: In CAD, the Point tool is used to initially select points at equal intervals, generate splines, analyze errors, and perform local point deletion and supplementation. The process of interpolation and fitting is continuously iterated to infinitely approximate the original curve, thereby obtaining the fitted curve and interpolation point values.
[0016] Further, the specific operation process of step 4 is as follows: In Matlab, the obtained interpolation points are used to generate a 3rd B-spline curve, and the curvature of the spline curve on the cumulative arc length is analyzed to calculate the average curvature, determine the high curvature threshold, and screen out the 20% convex surface region or the 10% concave surface region with the largest curvature change as the high curvature region.
[0017] Furthermore, the specific operation process of step 5 is as follows: In view of the characteristics of the 5th order B-spline curve, the interpolation points are supplemented, adjusted and refitted in the high curvature region to further improve the fitting accuracy of the curve in the high curvature region, so that the end point of the 3rd order B-spline curve and the beginning point of the 5th order B-spline curve coincide. At this point, the second derivatives of the two orders of curves are equal, ensuring a smooth transition of different orders of curves at this point.
[0018] Further, the specific operation process of step 7 is as follows: Hertz contact theory is introduced to obtain the cumulative arc length-curvature-velocity relationship model. Based on the cumulative arc length parameter of the hybrid B-spline curve trajectory, according to the cumulative arc length-curvature-velocity relationship model, the corresponding feed speed is assigned to different curvature regions on the trajectory, thereby determining the velocity distribution of the entire machining trajectory, and obtaining the velocity distribution curve of the grinding and polishing trajectory tool contact point and the machining trajectory with determined velocity.
[0019] The beneficial effects of this invention are as follows:
[0020] This invention provides an efficient and high-precision solution for grinding and polishing components with complex curved surfaces, such as steam turbine blades, aero-engine blades, and wind turbine blades. This invention improves the grinding and polishing effect at the blade edge of steam turbine blades, effectively reduces over-polishing and under-polishing in this area, and ensures the consistency of grinding and polishing of the entire blade body. Attached Figure Description
[0021] Figure 1 This is a schematic diagram of the deterministic robotic grinding and polishing method and process for blades based on hybrid B-spline curves provided in an embodiment of the present invention.
[0022] Figure 2 This is a schematic diagram of the cross-sectional profile of a steam turbine blade.
[0023] Figure 3 This is a schematic diagram of the tool coordinate system mapping on blade section A;
[0024] Figure 4 This is a schematic diagram showing the distribution of interpolation points on the concave and convex surfaces of the blade profile section line A.
[0025] Figure 5 This is a schematic diagram showing the fitting of a mixed B-spline curve of the concave and convex surfaces of the blade profile section line A and the change of curvature with cumulative arc length.
[0026] Figure 6 A schematic diagram of the machining trajectory curves for the convex and concave surfaces of a steam turbine blade;
[0027] Figure 7 The cumulative arc length-velocity relationship is shown in the trajectory curves of the concave and convex surfaces of the blade profile section line A.
[0028] Figure 8 A velocity distribution diagram of the machining trajectory on the concave surface of the blade;
[0029] Figure 9 This is a velocity distribution diagram of the machining trajectory on the outer convex surface of the blade. Detailed Implementation
[0030] The present invention will be further described below with reference to the accompanying drawings and embodiments, but the scope of protection of the present invention is not limited thereto.
[0031] Example 1
[0032] like Figure 1 As shown, a deterministic robotic grinding and polishing method for blades based on hybrid B-spline curves includes the following process:
[0033] Import the 3D model of the turbine blade into Solidworks software. Measurements in Solidworks show the blade length is approximately 120mm, and the belt width is 20mm. The interval for taking the blade profile section lines is set to 10mm. In CAD, the 2D drawing is processed to remove irrelevant lines. Considering the blade tip is still present (which needs to be removed during actual processing), and to avoid collisions between the belt and the blade tip during grinding, a profile section line is taken every 10mm, starting from the blade root, for a total of 11 profile section lines with a total interval of 100mm. Figure 2 As shown.
[0034] In CAD, locate the projection of the tool coordinate system origin onto each blade profile section. Set interpolation points on the tool coordinate system projection of each profile section. For cubic B-spline curves, use the "arc length parameterization" method, arranging the density of interpolation points based on the curvature along the arc length. Taking blade profile section line A as an example, the projections of the X and Y axes of the workpiece coordinate system onto this plane are known as follows: Figure 3 As shown, the coordinates of each point on curve A can be obtained in this coordinate system.
[0035] First, define the parameter step function:
[0036]
[0037] in, These are parameter variables used for the "parametric description" of the turbine blade profile curve projected in the tool coordinate system.
[0038] curvature function , Let arc length be the function of the curve to be fitted. These are the first and second derivatives of the arc length function of the curve to be fitted, respectively.
[0039] The global scale factor is constrained by the total length. ;
[0040] This represents the total number of interpolation points.
[0041] Then, the parameter points are determined using differential equations:
[0042]
[0043] Discretization solution:
[0044]
[0045] in, It is a parameterized description of the parameter points. It is the value of adjacent parameter points projected onto the tool coordinate system in the discretization solution;
[0046] For the curvature density integral, is the continuous integral variable in the curvature density integral.
[0047] Then, set the mandatory containment point set:
[0048]
[0049] This indicates that a parameter value in the point set M must be included, for each... Insert the nearest neighbor interval ,like Then replace for .
[0050] Finally, for a P=3 B-spline curve, the nodal vectors are constructed as follows:
[0051]
[0052] Interpolation point coordinate calculation:
[0053]
[0054] in, The original curve function, For arc length parameters;
[0055] It is a function of arc length. Its inverse function, ;
[0056] This represents the total length of the curve.
[0057] Through the above method, the interpolation points of the cubic B-spline curve interpolation fitting for the "concave surface" and "convex surface" of contour line A can be obtained, such as... Figure 4 As shown, Figure 4 In the diagram, (a) represents the B-spline curve interpolation point setting for the concave surface, and (b) represents the B-spline curve interpolation point setting for the convex surface.
[0058] The obtained interpolation point data are used as the original data points for cubic B-spline curve interpolation fitting. In MATLAB, the `spcrv` function (Spline curve) is used to generate the required uniform B-spline curve, and then the `cscvn` function (CubicSpline curve) is used to generate a cubic B-spline curve passing through the given interpolation points, which will be used as the object for subsequent curvature analysis. Then, three-point curvature calculations are performed on the generated cubic B-spline curve: for the three points... The curvature is approximately:
[0059]
[0060] Obtain curvature Next, Gaussian filtering is performed to eliminate high-frequency noise from discrete differentiation, preserving the true curvature characteristics. This allows us to obtain the change in curvature over the cumulative arc length. Then, using the percentile method, non-parametric statistics based on the data distribution are performed to select the top 20% (convex surface) or 10% (concave surface) regions with the largest rate of curvature change as high curvature regions. Finally, 3rd and 5th order B-spline curve interpolation fitting are performed on the regions with low and high curvature change rates, respectively. At the junction of the 3rd and 5th order B-spline curves, the continuity of the second derivative of different curves at that point must be ensured, i.e.:
[0061]
[0062] in, This enables piecewise interpolation fitting of 3rd and 5th order B-spline curves, such as... Figure 5 As shown, both the concave surface (a) and the convex surface (b) underwent quadratic fitting using fifth-order B-spline curves in regions with large curvature changes. The curvature changes with cumulative arc length as follows: Figure 5 As shown in (c) (concave surface) and (d) (convex surface).
[0063] Then, a uniform resampling operation was performed on the trajectory curves fitted by the obtained mixed B-spline curves. The 11 trajectory curves fitted by the convex and concave surfaces were uniformly resampled. Considering that the leading edge region of the blade is a transverse straight line, two adjacent machining trajectory curves were set to be connected by a straight line from beginning to end. Finally, the machining trajectory curves of the entire blade's convex surface (a) and concave surface (b) were obtained, as shown below. Figure 6 As shown.
[0064] In the engineering scenario of robotic belt polishing of blades, the physical contact relationship between the belt and the blade can be effectively described by Hertz's contact theory. This is fundamentally because the contact process aligns closely with the core assumptions of Hertz's theory. First, both the belt and the blade can essentially be considered elastic bodies: the belt is typically made of flexible materials such as rubber or fabric, capable of elastic deformation in the stressed area under tension; the blade, being a metal part, also exhibits elastic response on its surface at the microscopic level. Second, during the polishing process, the contact area between the belt and the blade is much smaller than their overall dimensions, representing a typical localized contact situation. This aligns with Hertz's theory's applicability condition that "the contact area is much smaller than the overall size of the object." Third, in actual polishing parameter design, the contact pressure between the belt and the blade is much lower than the material's yield strength, and the entire contact deformation process remains in the elastic stage, making it suitable for analysis using Hertz's elastic contact model. Finally, the sand belt can be approximated as a cylinder, and the blade surface can be simplified locally as a curved surface with a certain radius of curvature. The contact between the two on the spatial trajectory can be reduced to a local line contact or point contact model, which allows the equivalent radius of curvature in Hertz theory to be directly applied.
[0065] According to Hertzian contact theory, the amount of material removed by grinding and polishing can be approximately expressed as:
[0066]
[0067] in, This refers to the amount of material removed per unit time. The contact pressure is determined by both curvature and normal load. The grinding and polishing speed; This is an empirical constant. The contact pressure is:
[0068]
[0069] in, ; For normal loads; The equivalent radius of curvature, , The radius of curvature of the sand belt, The radius of curvature of the current point of the workpiece trajectory curve (i.e. ).so, Since the actual grinding and polishing contact point is set on the straight line of the belt abrasive, the actual radius of curvature of the belt abrasive can be approximated as infinite. Therefore, the local radius of curvature of the workpiece is much smaller than the radius of curvature of the belt abrasive. .
[0070] To maintain a constant amount of material removed, combining the above formula with empirical formulas for grinding and polishing, we have:
[0071]
[0072] Thus, the relationship between velocity and curvature can be solved:
[0073]
[0074] Taking the A-B spline fitting curve of the blade profile section line as an example, its cumulative arc length is sampled densely at equal intervals to obtain two sets of cumulative arc length-curvature data points for the fitting trajectory curves of the concave and convex surfaces. Then, the curvature data is substituted into the above curvature-velocity relationship to obtain the curvature-velocity relationship curve as shown in the figure. Figure 7 As shown, (a) represents the velocity-to-curvature relationship of the concave surface, and (b) represents the velocity-to-curvature relationship of the convex surface. This method is then applied to the trajectory fitting curves of the other 10 contour section lines. By sharing the resampled data points of the hybrid B-spline curve as the grinding and polishing tool contact points, it is convenient to preset the speed in segments and manage the speed boundaries, thus obtaining the velocity distribution map of the entire machining trajectory, as shown below. Figure 8 and Figure 9 As shown.
[0075] The implementation scheme and theoretical support of the deterministic robotic grinding and polishing method for blades based on hybrid B-spline curves provided by this invention have been described in detail above. The principles and methods used herein can be implemented in other examples (such as aero-engine blades, wind turbine blades, etc.) without departing from the spirit and scope of this invention. Therefore, the embodiments described herein are merely preferred embodiments, and this invention is not limited to the embodiments shown herein. Any improvements or modifications based on this invention should also fall within the protection scope of this application.
Claims
1. A deterministic robotic grinding and polishing method for blades based on hybrid B-spline curves, characterized in that, Includes the following steps: Step 1: Determine the distribution interval of the turbine blade profile section using SolidWorks software, extract the two-dimensional profile section line in CAD and process irrelevant lines; Step 2: In CAD, find the projection of the origin of the tool coordinate system onto each blade profile section; Step 3: In the projected coordinate system, set the interpolation points for the cubic non-uniform rational B-spline curve for each contour section line. By adjusting and supplementing the interpolation points, improve the fitting accuracy between the fitted curve and the contour section line, and obtain the trajectory curve of the cubic B-spline curve prefit. Step 4: Perform curvature analysis on the cumulative arc length of the prefitted trajectory curve, calculate the average curvature, determine the high curvature threshold, and select the 20% convex surface region or the 10% concave surface region with the largest curvature change as the high curvature region. Step 5: Perform 5-fold non-uniform rational B-spline curve interpolation fitting on the high curvature region, and keep the B-spline curve fitting on the other regions at 3-fold, to obtain the trajectory curve of the mixed B-spline curve fitting. Step 6: Perform uniform resampling of the trajectory curve fitted by the mixed B-spline curve with equal arc length to determine the final programming grinding and polishing contact point; Step 7: Using the curvature information of the trajectory curve, preset the data points differently according to different curvatures, manage the speed boundary, and obtain the speed distribution curve of the grinding and polishing trajectory tool contact point and the processing trajectory with a determined speed.
2. The deterministic robotic grinding and polishing method for blades based on hybrid B-spline curves as described in claim 1, characterized in that, The specific operation process of step 1 is as follows: Import the three-dimensional model of the turbine blade into SolidWorks software, determine the distribution interval of the outline section according to the blade length and the width of the belt sander, set the section lines, export the two-dimensional drawing, process the two-dimensional drawing in CAD, remove irrelevant lines, and only retain the outline section lines.
3. The deterministic robotic grinding and polishing method for blades based on hybrid B-spline curves as described in claim 2, characterized in that, The specific operation process of step 2 is as follows: According to the tool coordinate system predetermined in CAD, set a tiny circle with a diameter of 0.2-1mm with the origin of the coordinate system as the center. This circle runs through the entire blade along the Z-axis of the coordinate system. The center of the circle on each cross section is the projection of the origin of the tool coordinate system onto each cross section.
4. The deterministic robotic grinding and polishing method for blades based on hybrid B-spline curves as described in claim 2, characterized in that, The specific operation process of step 3 is as follows: In CAD, the Point tool is used to initially select points at equal intervals, generate splines, analyze errors, and perform local point deletion and supplementation. The process is to continuously interpolate and iterate to infinitely approximate the original curve, thereby obtaining the fitted curve and interpolation point values.
5. The deterministic robotic grinding and polishing method for blades based on hybrid B-spline curves as described in claim 4, characterized in that, The specific operation process of step 4 is as follows: In Matlab, the obtained interpolation points are used to generate a 3rd B-spline curve, and the curvature of the spline curve on the cumulative arc length is analyzed to calculate the average curvature, determine the high curvature threshold, and screen out the 20% convex surface region or the 10% concave surface region with the largest curvature change as the high curvature region.
6. The deterministic robotic grinding and polishing method for blades based on hybrid B-spline curves as described in claim 5, characterized in that, The specific operation process of step 5 is as follows: Based on the characteristics of the 5th order B-spline curve, interpolation points are supplemented, adjusted, and refitted in the high curvature region to further improve the fitting accuracy of the curve in the high curvature region, so that the end point of the 3rd order B-spline curve coincides with the beginning point of the 5th order B-spline curve. At this point, the second derivatives of the two orders of curves are equal, ensuring a smooth transition between different orders of curves at this point.
7. The deterministic robotic grinding and polishing method for blades based on hybrid B-spline curves as described in claim 6, characterized in that, The specific operation process of step 7 is as follows: Hertz contact theory is introduced to obtain the cumulative arc length-curvature-velocity relationship model. Based on the cumulative arc length parameter of the hybrid B-spline curve trajectory, according to the cumulative arc length-curvature-velocity relationship model, the corresponding feed speed is assigned to different curvature regions on the trajectory, thereby determining the velocity distribution of the entire machining trajectory, and obtaining the velocity distribution curve of the grinding and polishing trajectory tool contact point and the machining trajectory with determined velocity.
Citation Information
Patent Citations
High-efficiency processing method for complex curved surface part
CN106125672A
Numerical control machining tool path partitioning mapping method considering error regional distribution
CN107145126A
