Large-bearing automatic aligning and leveling device and method for large disc part of large rotary equipment
By combining a sphere-pair three-parameter constraint model with a nonlinear optimization algorithm, the problem of multiple error sources not being comprehensively considered in existing blade inspection is solved, improving the accuracy and simplicity of the measurement system and realizing efficient calibration of aero-engine blades.
Patent Information
- Application Number
- CN202510973358.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-15
- Publication Date
- 2025-11-11
AI Technical Summary
Existing blade inspection methods suffer from low measurement efficiency, easy surface damage, and failure to comprehensively consider multiple error sources in large-scale online or field inspections, resulting in insufficient measurement accuracy and repeatability.
A sphere-pair three-parameter constraint model is adopted, which uses the radii of the two spheres and the distance between their centers to construct a triple geometric constraint. By combining ellipsoid fitting, linear least squares method and Cholesky decomposition, nonlinear optimization algorithms (such as LM algorithm) are used to achieve joint calibration of the system attitude parameters and compensate for multiple error sources.
It improves the accuracy and ease of operation of the measurement system, reduces the complexity and cost of the calibration process, and achieves effective compensation for multiple error sources.
Smart Images
Figure CN120926874A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of aero-engine technology, specifically relating to a high-load automatic self-aligning and leveling device and method for large disk parts of large rotating equipment. Background Technology
[0002] Engine blades must possess precise geometric dimensions, rigorous profile shapes, and excellent surface integrity to ensure stable engine operation. To improve blade manufacturing quality and processing efficiency, high-precision, rapid profile inspection is crucial. Typically, the blade profile is an irregular free-form surface formed by fitting multiple cross-sectional curves, exhibiting strong distortion and thin-walled characteristics. This not only increases the difficulty of inspection but also places higher demands on the accuracy, stability, and spatial resolution of the measurement system. Currently, blade inspection still mainly relies on contact measurement methods, including the standard template method, inductive measurement method, and contact coordinate measuring machine (CMM). These methods have high measurement accuracy and reliability in laboratory environments, but they suffer from problems such as low measurement efficiency and easy damage to the blade surface. Therefore, they are mainly used for sampling inspection and are not suitable for online or field inspection of large batches of blades.
[0003] Currently, blade inspection still primarily relies on contact measurement methods, including the standard template method, inductive measurement, and contact coordinate measuring machine (CMM). These methods offer high accuracy and reliability in laboratory environments, but suffer from low efficiency and potential damage to the blade surface. Therefore, they are mainly used for random sampling and are not suitable for large-scale online or field inspection of blades. Non-contact measurement is now widely used in airfoil 3D profile measurement. Common non-contact methods include laser triangulation, fringe projection, and stereo vision. Sun et al. proposed a laser displacement sensor-based method for aero-engine blade shape detection, achieving efficient inspection of engine blades. Chen used binocular structured light to acquire point clouds of aero-engine blades, developed an algorithm based on rotation and projection to extract the chord length of turbine blades, and proposed an algorithm for extracting leading and trailing edges and a torsional analysis method for airfoil surfaces based on the angle between the edge and center connection line of the feature section. This method can effectively evaluate the machining accuracy of turbine blades. In practical measurement systems, factors such as sensor installation errors and structural errors often cause systematic deviations, which in turn affect measurement accuracy and repeatability. To reduce the impact of errors, researchers have proposed various calibration methods to compensate for and optimize the system. Among them, Sun et al., building on previous research, proposed an error compensation model that can effectively reduce errors caused by changes in the incident angle during laser displacement sensor measurement; Wang et al. developed a measurement system based on a four-axis platform and a line laser scanning sensor. They introduced a calibration method based on the characteristics of the blade reference surface, eliminating the system's pose error without using additional calibration objects. He et al., based on the principle of spectral confocality, developed a system for in-situ measurement of five-axis CNC machine tools and proposed a calibration method combining a standard sphere and spherical constraint equations, effectively improving system accuracy. Xue et al. established a blade profile measurement system based on a spectral confocal sensor, designed a probe sampling strategy, and used a standard sphere to calibrate the sensor probe's light output direction and the precision turntable's rotation center.
[0004] While existing research can effectively improve the calibration accuracy of measurement systems, it still faces many challenges in practical applications of blade measurement. On the one hand, existing methods often only model and compensate for single error sources (such as sensor installation deviation or platform pose error), lacking comprehensive consideration and collaborative modeling of multiple error sources such as guide rail error, sensor attitude error, and turntable rotation axis error. On the other hand, some methods rely on complex or diverse calibration objects, such as multi-ball arrays and feature plates, resulting in complex calibration operations and strong dependence on field conditions, limiting their promotion and application in engineering practice. Therefore, there is an urgent need to propose an efficient calibration method with a simple structure, clear geometric constraints, and the ability to simultaneously compensate for multiple error sources, in order to achieve joint estimation and accurate calibration of system attitude parameters. Summary of the Invention
[0005] This invention provides a high-load automatic self-aligning and leveling device and method for large disc parts of large rotating equipment. The constructed "sphere-pair three-parameter constraint model" uses the radii of the two spheres and the distance between their centers to construct triple geometric constraints, realizing the joint calibration of the system's attitude parameters and effectively improving the measurement accuracy of the measurement system.
[0006] This invention is achieved through the following technical solution: A high-load automatic self-aligning and leveling device and method for large disk parts of large rotating equipment is disclosed. The method is as follows: first, the initial value of the tilt angle of the guide rail is obtained by using ellipsoidal fitting, linear least squares method and Cholesky decomposition; then, the tilt angle between the guide rails, the installation error angle of the sensor and the position of the rotation axis are accurately estimated by joint nonlinear optimization, so as to achieve effective compensation for measurement error and thus realize the calibration of the aero-engine blade measuring instrument.
[0007] Furthermore, the calibration method includes the following steps: Step 1. Place the standard ball pair on the turntable; Step 2. The turntable rotates the standard ball pair; Step 3. Use a three-coordinate guide rail to drive the spectral confocal sensor to collect data; Step 4. Obtain initial calibration values using ellipse fitting and the SVD method; Step 5. Calculate the accurate guide rail tilt angle, turntable rotation axis, and spectral confocal sensor tilt angle using the LM algorithm.
[0008] Furthermore, step 2 specifically involves using a pair of standard spheres with known radii and a known distance between their centers as standard objects for calibration; placing the standard spheres on a turntable, rotating the turntable m-1 times, and taking n points on the surface of each standard sphere.
[0009] Furthermore, step 3 specifically involves having a 2m standard sphere in space, and a point on the standard sphere 1 is represented as { A point on a standard sphere 2 is represented as { The coordinates of the center of the standard sphere 1 are: The coordinates of the center of the standard sphere 2 are: The radii of the two standard spheres are respectively The distance between the centers of the two standard spheres is D.
[0010] Furthermore, step 4 specifically involves performing ellipse fitting on the acquired points to obtain the sphere center and the Q matrix, and then performing Cholesky decomposition on the Q matrix to obtain the initial values of the tilt parameter and the sphere center. The coordinates of a point on the sphere can be approximated as:
[0011]
[0012] in, d represents the reading from the xyz guide rail, and d represents the reading from the sensor. These are spherical coordinates with errors. The ellipsoid formula is then written as:
[0013]
[0014]
[0015] in, The coordinates of the center of the sphere are the true coordinates; the coordinates of the center of the ellipsoid are the minimum points of the function. :
[0016]
[0017]
[0018] After obtaining the center of the ball, the axis of rotation is calibrated.
[0019] Furthermore, we analyze the points on standard sphere 1; and obtain the center of standard sphere 1. Given a coordinate set, assuming there are m spheres, the coordinates of the center of the j-th sphere are... :
[0020]
[0021] Calculate the centroid of the point set
[0022]
[0023]
[0024] Calculate the covariance matrix and fit the plane normal vector:
[0025] in Let be the coordinates of the center of the j-th ball, and m be the number of standard balls. This is the decentralized coordinate matrix of the sphere's center. for The transpose of the matrix, The difference in the notation used to solve for the normal vector is true; Perform singular value decomposition on the covariance matrix:
[0026] The eigenvector corresponding to the minimum singular value, i.e., the normal vector of the fitting plane, has a unit vector along its rotation axis:
[0027] Take normal vector Two basis vectors of the orthogonal complement space:
[0028] The basis transformation matrix can then be expressed as:
[0029] Map the projected 3D points to The coordinates of the location are in a two-dimensional coordinate system. :
[0030] Based on two-dimensional point sets The center and radius of the circle are determined using the least squares method. Finally, the two-dimensional center coordinates are converted to three-dimensional coordinates, yielding the final three-dimensional coordinates of the center.
[0031] in, The basis transformation matrix, Project the center of the j-th ball onto The coordinates are in a two-dimensional coordinate system.
[0032] Furthermore, step 5 specifically involves setting the radii of the two standard spheres to be respectively... and The distance between the centers of the balls is For a point on standard sphere 1, the following conditions are met: The center of standard ball 1 is... and radius This is obtained using the least squares method for spheres; and the distance between the centers of the two standard spheres should satisfy:
[0033] Based on the above constraints, the objective function for nonlinear optimization is constructed as follows:
[0034] in These are weighting coefficients used to balance the impact of the residual term on the overall optimization result. For rotation matrix, For rotation angle, Let J be the coordinates of the center of sphere 2 at the j-th position. Let the coordinates of the standard sphere 2 be at the initial center coordinates. Let the coordinates of the standard sphere 1 be at its initial center. The LM algorithm is used to obtain the final calibration result.
[0035] The beneficial effects of this invention are: The calibration model of this invention takes into account various errors.
[0036] This invention uses the same standard part for calibration, reducing costs and operational difficulty.
[0037] This invention takes into account various constraints to improve calibration accuracy.
[0038] This invention employs the LM algorithm to further improve calibration accuracy. Attached Figure Description
[0039] Figure 1This is a flowchart of the method of the present invention.
[0040] Figure 2 This is a schematic diagram of the data acquisition process of the present invention. Detailed Implementation The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.
[0041] Many specific details are set forth in the following description in order to provide a full understanding of this application. However, this application may also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the spirit of this application. Therefore, this application is not limited to the specific embodiments disclosed below.
[0042] This invention proposes a system calibration model that comprehensively considers guide rail errors, sensor installation deviations, and turntable rotation axis errors, and designs a corresponding calibration method based on sphere pair geometric constraints. The constructed "sphere pair three-parameter constraint model" uses the radii of the two spheres and the distance between their centers to build triple geometric constraints, realizing the joint calibration of the system's attitude parameters and effectively improving the measurement accuracy of the measurement system.
[0043] A high-load automatic self-aligning and leveling device and method for large disk parts of large rotating equipment is disclosed. The method is as follows: first, the initial value of the tilt angle of the guide rail is obtained by using ellipsoidal fitting, linear least squares method and Cholesky decomposition; then, the tilt angle between the guide rails, the installation error angle of the sensor and the position of the rotation axis are accurately estimated by joint nonlinear optimization, so as to achieve effective compensation for measurement error and thus realize the calibration of the aero-engine blade measuring instrument.
[0044] Furthermore, the calibration method includes the following steps: Step 1. Place the standard ball pair on the turntable; Step 2. The turntable rotates the standard ball pair; Step 3. Use a three-coordinate guide rail to drive the spectral confocal sensor to collect data; Step 4. Obtain initial calibration values using ellipse fitting and the SVD method; Step 5. Calculate the accurate guide rail tilt angle, turntable rotation axis, and spectral confocal sensor tilt angle using the LM algorithm.
[0045] Furthermore, step 2 specifically involves using a pair of standard spheres with known radii and a known distance between their centers as standard objects for calibration; placing the standard spheres on a turntable, rotating the turntable m-1 times, and taking n points on the surface of each standard sphere.
[0046] Furthermore, step 3 specifically involves considering that there is a 2m standard sphere in space, and the point on the standard sphere 1 is represented as { A point on a standard sphere 2 is represented as { The coordinates of the center of the standard sphere 1 are: The coordinates of the center of the standard sphere 2 are: The radii of the two standard spheres are respectively The distance between the centers of the two standard spheres is D.
[0047] Furthermore, step 4 specifically involves performing ellipse fitting on the acquired points to obtain the sphere center and the Q matrix, and then performing Cholesky decomposition on the Q matrix to obtain the initial values of the tilt parameter and the sphere center. The coordinates of a point on the sphere can be approximated as:
[0048]
[0049] in, d represents the reading from the xyz guide rail, and d represents the reading from the sensor. These are spherical coordinates with errors.
[0050] The ellipsoid formula can then be written as:
[0051]
[0052]
[0053] These are the true coordinates of the ellipsoid's center. The coordinates of the ellipsoid's center are the minimum points of this function. :
[0054]
[0055]
[0056] After obtaining the center of the ball, the axis of rotation is calibrated.
[0057] Furthermore, we analyze the points on standard sphere 1; and obtain the center of standard sphere 1. Given a coordinate set, assuming there are m spheres, the coordinates of the center of the j-th sphere are... :
[0058]
[0059] Calculate the centroid of the point set
[0060]
[0061]
[0062] Calculate the covariance matrix and fit the plane normal vector:
[0063] in Let be the coordinates of the center of the j-th ball, and m be the number of standard balls. This is the decentralized coordinate matrix of the sphere's center. for The transpose of the matrix, The difference in the notation used to solve for the normal vector Perform singular value decomposition on the covariance matrix:
[0064] The eigenvector corresponding to the minimum singular value, i.e., the normal vector of the fitting plane, has a unit vector along its rotation axis:
[0065] Take normal vector Two basis vectors of the orthogonal complement space:
[0066] The basis transformation matrix can then be expressed as:
[0067] Map the projected 3D points to The coordinates of the location are in a two-dimensional coordinate system. :
[0068] Based on two-dimensional point sets The center and radius of the circle are determined using the least squares method. Finally, the two-dimensional center coordinates are converted to three-dimensional coordinates, yielding the final three-dimensional coordinates of the center.
[0069] in, The basis transformation matrix, Project the center of the j-th ball onto The coordinates are in a two-dimensional coordinate system.
[0070] Furthermore, step 5 specifically involves recognizing that the calibration result obtained through the linear least squares method is only an approximate solution. To further improve measurement accuracy, a nonlinear optimization method needs to be introduced to obtain more accurate calibration parameters. The first step of the nonlinear optimization is to construct a corresponding loss function based on the error model. The radii of the two standard spheres are respectively... and The distance between the centers of the balls is For a point on standard sphere 1, the following conditions are met: The center of standard ball 1 is... and radius This is obtained using the least squares method for spheres; and the distance between the centers of the two standard spheres should satisfy:
[0071] Based on the above constraints, the objective function for nonlinear optimization is constructed as follows:
[0072] in These are weighting coefficients used to balance the impact of the residual term on the overall optimization result. For rotation matrix, For rotation angle, Let J be the coordinates of the center of sphere 2 at the j-th position. Let the coordinates of the standard sphere 2 be at the initial center coordinates. Let the coordinates of the standard sphere 1 be at its initial center. The LM algorithm is used to obtain the final calibration result.
Claims
1. A high-load-bearing automatic self-aligning and leveling device and method for large disc-shaped parts in large rotating equipment, characterized in that, The method is as follows: First, the initial value of the tilt angle of the guide rail is obtained by using ellipsoid fitting, linear least squares method and Cholesky decomposition; then, the tilt angle between the guide rails, the installation error angle of the sensor and the position of the rotation axis are accurately estimated by joint nonlinear optimization, so as to achieve effective compensation for measurement error and thus realize the calibration of the aero-engine blade measuring instrument.
2. The calibration method according to claim 1, characterized in that, The calibration method includes the following steps: Step 1. Place the standard ball pair on the turntable; Step 2. The turntable rotates the standard ball pair; Step 3. Use a three-coordinate guide rail to drive the spectral confocal sensor to collect data; Step 4. Obtain initial calibration values using ellipse fitting and the SVD method; Step 5. Calculate the accurate guide rail tilt angle, turntable rotation axis, and spectral confocal sensor tilt angle using the LM algorithm.
3. The calibration method according to claim 2, characterized in that, Step 2 specifically involves using a pair of standard spheres with known radii and a known distance between their centers as standard objects for calibration; placing the standard spheres on a turntable, rotating the turntable m-1 times, and taking n points on the surface of each standard sphere.
4. The calibration method according to claim 2, characterized in that, Step 3 specifically involves having a 2m standard sphere in space, and a point on standard sphere 1 represented as { A point on a standard sphere 2 is represented as { The coordinates of the center of the standard sphere 1 are: The coordinates of the center of the standard sphere 2 are: The radii of the two standard spheres are respectively The distance between the centers of the two standard spheres is D.
5. The calibration method according to claim 2, characterized in that, Step 4 specifically involves fitting an ellipse to the acquired points to obtain the sphere center and the Q matrix, and then performing Cholesky decomposition on the Q matrix to obtain the initial values of the tilt parameter and the sphere center. The coordinates of a point on the sphere can be approximated as: in, d represents the reading from the xyz guide rail, and d represents the reading from the sensor. These are spherical coordinates with errors. The ellipsoid formula is then written as: in, The coordinates of the center of the sphere are the true coordinates; the coordinates of the center of the ellipsoid are the minimum points of the function. : After obtaining the center of the ball, the axis of rotation is calibrated.
6. The calibration method according to claim 5, characterized in that, Analyze the points on standard sphere 1; obtain the center of standard sphere 1. Given a coordinate set, assuming there are m spheres, the coordinates of the center of the j-th sphere are... : Calculate the centroid of the point set Calculate the covariance matrix and fit the plane normal vector: in Let be the coordinates of the center of the j-th ball, and m be the number of standard balls. This is the decentralized coordinate matrix of the sphere's center. for The transpose of the matrix, The difference in the notation used to solve for the normal vector is true; Perform singular value decomposition on the covariance matrix: The eigenvector corresponding to the minimum singular value, i.e., the normal vector of the fitting plane, has a unit vector along its rotation axis: Take normal vector Two basis vectors of the orthogonal complement space: The basis transformation matrix can then be expressed as: Map the projected 3D points to The coordinates of the location are in a two-dimensional coordinate system. : Based on two-dimensional point sets The center and radius of the circle are determined using the least squares method. Finally, the two-dimensional center coordinates are converted to three-dimensional coordinates, yielding the final three-dimensional coordinates of the center. in, The basis transformation matrix, Project the center of the j-th ball onto The coordinates are in a two-dimensional coordinate system.
7. The calibration method according to claim 6, characterized in that, Step 5 specifically involves setting the radii of the two standard spheres as follows: and The distance between the centers of the balls is For a point on standard sphere 1, the following conditions are met: The center of standard ball 1 is... and radius This is obtained using the least squares method for spheres; and the distance between the centers of the two standard spheres should satisfy: Based on the above constraints, the objective function for nonlinear optimization is constructed as follows: in These are weighting coefficients used to balance the impact of the residual term on the overall optimization result. For rotation matrix, For rotation angle, Let J be the coordinates of the center of sphere 2 at the j-th position. Let the coordinates of the standard sphere 2 be at the initial center coordinates. Let the coordinates of the standard sphere 1 be at its initial center. The LM algorithm is used to obtain the final calibration result.