Method for detecting blade profile tolerance of precision casting blade with notch position

Abstract

The invention relates to the technical field of aero-engine blade detection, in particular to a blade profile tolerance detection method for a precision casting blade with a notch position, and the method comprises the following steps: S1, importing actual measurement point cloud data and a theoretical CAD model of a blade profile; s2, pre-processing the actually measured point cloud data, and converting the point cloud data into an NURBS (Non-Uniform Rational B-Spline) curved surface; s3, aligning the theoretical CAD model of the blade with the coordinate system of the NURBS curved surface by using an improved ICP algorithm; s4, taking spatial residual and curvature continuity as composite convergence conditions, and performing iterative optimization on the curve of the processing area by adopting a dynamic interruption strategy and incremental fitting to obtain an optimized curve; and S5, generating a profile report of the blade profile according to the optimization curve. According to the method, through gradual alignment of a theoretical model and compensation measurement data, self-adaptive segmentation of a curvature sensitive area is realized; and meanwhile, iterative fitting is carried out by adopting a dynamic interruption strategy and incremental fitting, so that the accuracy of a detection result is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of aero-engine blade detection, in particular to a method for detecting the profile of a precision-cast blade with a notch position. Background Art

[0002] The quality of the exhaust edge of an aircraft engine blade directly affects the performance, reliability, and safety of the engine. In particular, the shape and dimensional accuracy of the exhaust edge directly affect the aerodynamic performance of the blade. If there are defects in the exhaust edge (such as cracks, deformation, dimensional deviation, etc.), it will cause airflow separation, reduced efficiency, and even cause vibration and stall, seriously affecting the engine's thrust and fuel efficiency.

[0003] CN113393428A discloses a method for detecting the shape of the intake and exhaust edges of aircraft engine blades, comprising: obtaining theoretical point cloud data of the blade profile, measured point cloud data of the blade profile, and a deviation value; performing curve fitting based on the theoretical point cloud data of the blade profile to obtain a theoretical blade profile curve and a theoretical blade profile mid-camber line, and performing curve fitting based on the measured point cloud data of the blade profile to obtain a measured blade profile curve and a measured blade profile mid-camber line; calculating based on the theoretical blade profile curve and the theoretical mid-camber line to obtain a theoretical leading edge vertex and a theoretical trailing edge vertex; calculating based on the measured blade profile curve and the measured mid-camber line to obtain a measured leading edge vertex and a measured trailing edge vertex; calculating the curvature radius of the measured leading edge vertex and the curvature radius of the measured trailing edge vertex; and comparing and determining the leading edge shape and the trailing edge shape of the blade.

[0004] Based on the above detection methods, the following problems exist: 1. Low data utilization rate of traditional detection methods: Traditional methods use a discrete processing method, and each iteration requires re-operation of the sampled point cloud data, resulting in less than 40% utilization of the original data, which increases the detection cost and time; 2. Limited accuracy of existing fitting algorithms: The commonly used RANSAC algorithm is easily interfered by outliers when processing complex surfaces, with a maximum deviation of up to 0.15mm, which cannot meet high-precision manufacturing requirements; 3. Commercial software processes are irreversible: Commercial software lacks a traceability mechanism for the iterative process and cannot achieve parameterized reverse verification, resulting in the need to redo the entire detection process for adjustment and verification, wasting time and resources; 4. Multi-version data storage redundancy: Traditional methods lack effective management in data storage, and the amount of multi-version intermediate data storage can reach 3-5 times that of the original data, increasing storage costs and management difficulty. Summary of the Invention

[0005] In view of the problem that blade profile detection in the prior art cannot meet the required accuracy, the present invention provides a method for detecting the blade profile of a precision-cast blade with a notch position.

[0006] The present invention is achieved through the following technical solutions: A method for detecting the profile of a precision-cast blade having a notch position comprises the following steps: S1, import measured point cloud data and theoretical CAD model of blade profile; S2, pre-processing the measured point cloud data and converting it into a NURBS surface; S3, using the improved ICP algorithm to align the theoretical CAD model of the blade with the coordinate system of the NURBS surface to obtain the adaptive segmented curve of the curvature sensitive area; where the curvature sensitive area is the intake edge and the exhaust edge; S4, with spatial residual and curvature continuity as the composite convergence conditions, the curve of the processing area is iteratively optimized using a dynamic interruption strategy and incremental fitting to obtain the optimized curve. The processing area is all areas outside the unprocessed area, and the unprocessed area is the exhaust edge area of the blade basin segment or the middle area of the exhaust edge segment; S5, generating a blade profile report according to the optimization curve.

[0007] Preferably, in S1, the measured point cloud data of the blade profile includes the blade inlet edge, the blade outlet edge and the blade necking; and radius compensation is used to pre-process the measured point cloud data.

[0008] Preferably, in S2, the subdivision level of the NURBS surface is Lv5.

[0009] Preferably, in S3, the steps of using the improved ICP algorithm to align the coordinate system are: S301, executing the improved ICP algorithm, assigning feature weights: leading edge weight 0.6, exhaust edge weight 0.3, blade back / blade basin weight 0.1; S302 , using KD-Tree to accelerate the nearest point search, progressively aligning the theoretical CAD model of the blade with the pre-processed point cloud data to obtain adaptive segmentation of the curvature-sensitive area.

[0010] Preferably, in S4, in the composite convergence condition, the spatial residual is ≤ 0.03 mm and the curvature C² is continuous.

[0011] Preferably, in S4, in the dynamic interruption strategy, root mean square error (RMS) and curvature calculation are used to perform dynamic interruption and exclude unprocessed areas.

[0012] Preferably, in S4, the specific operation of the incremental fitting is: taking the actually measured point cloud data as input, retaining the Jacobian matrix condition number of the previous fitting, and using the LM algorithm to update the parameters, wherein the damping factor μ=0.01.

[0013] A system for detecting the profile of a precision-cast blade with a notch position comprises a data acquisition module, a data processing module, a data alignment module, an optimization iteration module, and a data output module, wherein the data acquisition module is used to import the measured point cloud data of the blade profile and the theoretical CAD model of the blade profile; The data processing module is used to pre-process the measured point cloud data and convert the pre-processed blade profile point cloud data into a NURBS surface; The data alignment module is used to align the theoretical CAD model of the blade with the coordinate system of the measured point cloud data using the improved ICP algorithm to obtain adaptive segmentation of the curvature-sensitive area; The optimization iteration module is used to iteratively optimize irregular and unpredictable curves in the processing area using spatial residual and curvature continuity as composite convergence conditions, and adopt dynamic interruption strategy and incremental fitting to obtain the optimized curve; The data output module is used to generate a blade profile report based on the optimization curve.

[0014] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method when executing the computer program.

[0015] A storage medium stores a computer program thereon, and the steps of the method are performed when the computer program is executed by a processor.

[0016] Compared with the prior art, the present invention has the following beneficial effects: This method, developed to detect the profile of notched precision blades, uses progressive alignment of a theoretical model with compensated measurement data to achieve adaptive segmentation of curvature-sensitive regions. It also employs a dynamic interruption strategy and incremental fitting for iterative fitting, ensuring the accuracy of the test results. Testing has shown that this method can reduce the processing time for a single blade from one minute to 10 seconds, while reducing the residual error in the exhaust edge region to 0.012 mm (RMS), significantly better than the 0.15 mm achieved by existing RANSAC algorithms. Furthermore, the number of stored versions is only one-seventh of that required by traditional methods.

[0017] Furthermore, the system error of the optical measurement equipment is corrected by the radius compensation algorithm, eliminating the influence of the inherent error of the measurement equipment on the edge contour, making the measured point cloud closer to the actual geometric shape.

[0018] Furthermore, the improved ICP algorithm introduces feature weight distribution (0.6 for the leading edge, 0.3 for the exhaust edge, and 0.1 for the blade back / base), and combines it with KD-Tree to accelerate the nearest point search and achieve adaptive segmentation of curvature-sensitive areas.

[0019] Furthermore, root mean square error (RMS) and curvature calculation are used for dynamic interruption to ensure that the fitting results meet high standards in terms of accuracy and surface quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a flow chart of the blade profile detection method of the present invention; Figure 2 Schematic diagram of the exhaust edge position before dynamic interruption of the present invention; Figure 3 Schematic diagram of the exhaust edge position after dynamic interruption of the present invention; Figure 4 Schematic diagram of the blade outlet edge shape before dynamic interruption of the present invention; Figure 5 Schematic diagram of the blade outlet edge shape after dynamic interruption of the present invention; Figure 6 Schematic diagram of the exhaust edge position of the blade basin of the present invention before dynamic interruption; Figure 7 Schematic diagram of the exhaust edge position of the blade basin of the present invention after dynamic interruption. DETAILED DESCRIPTION

[0021] The present invention will be further described in detail below with reference to specific embodiments, which are intended to explain the present invention rather than to limit it.

[0022] The present invention discloses a method for detecting the profile of a precision casting blade with a notch position, referring to Figure 1 , including the following steps: S1 imports measured point cloud data and a theoretical CAD model of the blade profile. The measured point cloud data includes the blade inlet edge, blade outlet edge, and blade neck. The measured point cloud data is actually cross-sectional information. The data is continuous, high-density, and uniformly high. Radius compensation is used to pre-process the measured point cloud data.

[0023] S2 preprocesses the measured point cloud data and converts it into a NURBS surface with a subdivision level of Lv5. The specific process is as follows: First, the measured point cloud data is preprocessed through radius compensation and denoising. Then, the NURBS surface parameters are determined and fitted. Finally, the surface is evaluated and optimized until the accuracy requirements are met.

[0024] S3, using the improved ICP algorithm, aligns the theoretical CAD model of the blade with the coordinate system of the NURBS surface to obtain an adaptive segmented curve in the curvature-sensitive area; the curvature-sensitive area is the intake edge and the exhaust edge; the CAD model coordinate system is established based on the design standard, and the measured data coordinate system may be determined based on the measurement benchmark.

[0025] The specific operations of alignment are: S301, executing the improved ICP algorithm, assigning feature weights: leading edge weight 0.6, exhaust edge weight 0.3, blade back / blade basin weight 0.1; S302 , using KD-Tree to accelerate the nearest point search, progressively aligning the theoretical CAD model of the blade with the pre-processed point cloud data to obtain an adaptive segmented curve in the curvature-sensitive area.

[0026] S4: Using spatial residuals and curvature continuity as composite convergence criteria, the curve in the machining area is iteratively optimized using a dynamic interruption strategy and incremental fitting to obtain the optimized curve. The optimized curve has minimal spatial residuals and good curvature continuity. The machining area is all areas outside the unmachined area, which is the exhaust edge of the blade basin segment or the middle area of the exhaust edge segment.

[0027] Among them, in the composite convergence conditions, the spatial residual is ≤ 0.03 mm and the curvature C² is continuous.

[0028] In the dynamic interruption strategy, root mean square error and curvature calculation are used for dynamic interruption to exclude the unprocessed area, where the unprocessed area is the exhaust edge area of the blade basin segment or the middle area of the exhaust edge segment.

[0029] Specifically, the root mean square (RMS) error is given by the following formula:

[0030] Where, n : Indicates the sample size, that is, the number of data points.

[0031] i : is the index variable for summation, used to traverse from 1 to n Each sample of .

[0032] :Represents i An actual measured value or an actual observed value.

[0033] Representative i A predicted value, that is, the value obtained by the model, algorithm, etc. The corresponding prediction results.

[0034] The curvature calculation formula is:

[0035] Where, It represents curvature, which is used to measure the degree of curvature of the blade profile curve at a certain point, and the degree to which the reaction curve deviates from a straight line; is the parametric equation of the curve Parameters The first derivative of the curve is the tangent vector of the curve at a certain point, and its direction represents the tangent direction of the curve at that point; the parametric equation of the curve Used to describe the blade profile curve through parameters The change of points at different positions on the top curve; is the parametric equation of the curve Parameters The second derivative of , reflecting the tangent vector changes in the situation.

[0036] The specific operation of incremental fitting is: taking the actual measured point cloud data as input, retaining the Jacobian matrix condition number of the previous fitting, and using the LM algorithm to update the parameters, where the damping factor μ=0.01.

[0037] S5 generates a blade profile report based on the optimized curve. When generating a profile report, the fitting parameters, deviation values, and other data are summarized. Then, the test results and analysis are clearly presented using charts and text. The profile report includes the blade profile deviation values and a graphical report.

[0038] According to the method for detecting the blade profile of a precision casting blade with a notch position proposed by the present invention, a blade is detected, and the comparison results before and after the dynamic interruption of the corresponding area are as follows: Figures 2 to 7 As shown in the figure, the blade processing time is shortened from 1 minute of the traditional method to 10 seconds, and the exhaust edge area fitting residual is reduced to 0.012mm (RMS), which is significantly better than the 0.15mm of the existing RANSAC algorithm.

[0039] The present invention provides a method for detecting the blade profile of a precision-cast blade with a notch position. By improving the ICP algorithm with feature weights, high-order NURBS dynamic optimization, and composite convergence conditions, it achieves a breakthrough in blade profile detection with high precision, high efficiency, and intelligence. The method is particularly suitable for fields such as aircraft engines that have strict requirements on surface accuracy and reliability.

[0040] The present invention discloses a precision casting blade profile contour detection system with a notch position, comprising a data acquisition module, a data processing module, a data alignment module, an optimization iteration module and a data output module, wherein the data acquisition module is used to import the measured point cloud data of the blade profile and the theoretical CAD model of the blade profile; the data processing module is used to preprocess the measured point cloud data and convert the preprocessed blade profile point cloud data into a NURBS surface; the data alignment module is used to align the theoretical CAD model of the blade with the coordinate system of the measured point cloud data using an improved ICP algorithm to obtain adaptive segmentation of the curvature sensitive area; the optimization iteration module is used to use spatial residual and curvature continuity as composite convergence conditions, adopt a dynamic interruption strategy and incremental fitting to iteratively optimize irregular and unpredictable curves in a processing area to obtain an optimized curve; the data output module is used to generate a blade profile contour report according to the optimized curve.

[0041] The present invention discloses an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method when executing the computer program.

[0042] The present invention discloses a storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the method are performed.

[0043] The above description is merely a preferred embodiment of the present invention and is not intended to impose any limitation on the technical solution of the present invention. Those skilled in the art should understand that, without departing from the spirit and principles of the present invention, the technical solution can also be subjected to several simple modifications and replacements, and these modifications and replacements are also within the scope of protection covered by the claims.

Claims

1. A method for detecting the profile of a precision-cast blade with a notch position, characterized in that: The following steps are involved: S1, import measured point cloud data and theoretical CAD model of blade profile; S2, pre-processing the measured point cloud data and converting it into a NURBS surface; S3, using the improved ICP algorithm to align the theoretical CAD model of the blade with the coordinate system of the NURBS surface to obtain the adaptive segmented curve of the curvature sensitive area; where the curvature sensitive area is the intake edge and the exhaust edge; S4, with spatial residual and curvature continuity as the composite convergence conditions, the curve of the processing area is iteratively optimized using a dynamic interruption strategy and incremental fitting to obtain the optimized curve. The processing area is all areas outside the unprocessed area, and the unprocessed area is the exhaust edge area of the blade basin segment or the middle area of the exhaust edge segment; S5, generating a blade profile report according to the optimization curve.

2. The method for detecting the profile of a precision-cast blade with a notch position according to claim 1, characterized in that: In S1, the measured point cloud data of the blade profile includes the blade inlet edge, the blade outlet edge and the blade neck; the measured point cloud data is preprocessed using radius compensation.

3. The method for detecting the profile of a precision-cast blade with a notch position according to claim 1, characterized in that: In S2, the subdivision level of NURBS surfaces is Lv5.

4. The method for detecting the profile of a precision-cast blade with a notch position according to claim 1, characterized in that: In S3, the steps for coordinate system alignment using the improved ICP algorithm are: S301, executing the improved ICP algorithm, assigning feature weights: leading edge weight 0.6, exhaust edge weight 0.3, blade back / blade basin weight 0.1; S302 , using KD-Tree to accelerate the nearest point search, the theoretical CAD model of the blade is progressively aligned with the pre-processed point cloud data to obtain an adaptive segmented curve in the curvature-sensitive area.

5. The method for detecting the profile of a precision-cast blade with a notch position according to claim 1, characterized in that: In S4, the composite convergence conditions are that the spatial residual is ≤ 0.03 mm and the curvature C² is continuous.

6. The method for detecting the profile of a precision-cast blade with a notch position according to claim 1, characterized in that: In S4, in the dynamic interruption strategy, root mean square error and curvature calculation are used for dynamic interruption to exclude unprocessed areas.

7. The method for detecting the profile of a precision-cast blade with a notch position according to claim 1, characterized in that: In S4, the specific operation of incremental fitting is: taking the actual measured point cloud data as input, retaining the Jacobian matrix condition number of the previous fitting, and using the LM algorithm to update the parameters, where the damping factor μ=0.

01.

8. A system for detecting the profile of a precision-cast blade with a notch position, characterized in that: It includes a data acquisition module, a data processing module, a data alignment module, an optimization iteration module and a data output module, wherein the data acquisition module is used to import the measured point cloud data of the blade profile and the theoretical CAD model of the blade profile; The data processing module is used to pre-process the measured point cloud data and convert the pre-processed blade profile point cloud data into a NURBS surface; The data alignment module is used to align the theoretical CAD model of the blade with the coordinate system of the measured point cloud data using the improved ICP algorithm to obtain adaptive segmentation of the curvature-sensitive area; The optimization iteration module is used to iteratively optimize the curve of the curvature-sensitive area in the processing area using the spatial residual and curvature continuity as the composite convergence conditions, and adopt the dynamic interruption strategy and incremental fitting to obtain the optimized curve; The data output module is used to generate a blade profile report based on the optimization curve.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images