A method for field of view planning of blade section line laser scanning based on multi-constraints

By adopting the multivariate constraints in the blade cross-section line laser scanning field planning method in complex blade detection, the scanning data accuracy and reliability problems caused by complex geometric features are solved, and high-precision multi-field data splicing and detection are achieved.

CN118602985BActive Publication Date: 2024-11-05SICHUAN UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411073689.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-07
Publication Date
2024-11-05
Estimated Expiration
2044-08-07

AI Technical Summary

Technical Problem

The prior art is difficult to ensure the accuracy and reliability of scanning data in complex blade detection, especially the differences in imaging distances and viewing angles of different segments caused by complex geometric features, which increases the difficulty of splicing of multi-field data and reduces the detection accuracy.

Method used

Using the blade cross-section line laser scanning field of view planning method based on multivariate constraints, the potential scanning points are obtained through discrete contour data sets, the viewing angle and viewing distance constraints are realized, and the positioning is iteratively optimized, ensuring the accuracy of the scan data and the proportion of overlapping areas, and optimizing the Field of view planning.

Benefits of technology

It effectively improves the reliability of scanning data during blade detection and the accuracy of multi-field data splicing, and improves the detection accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118602985B_ABST
    Figure CN118602985B_ABST
Patent Text Reader

Abstract

The present invention belongs to the technical field of blade cross-section profile detection, and specifically relates to a blade cross-section line laser scanning field of view planning method based on multi-factor constraints. The method realizes the planning of the line laser sensor scanning field of view through adjacent field of view overlap area constraints and angle of view-distance constraints, so as to effectively ensure the reliability of scanning data during blade detection and the accuracy of multi-field of view scanning data splicing.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of blade cross-section profile detection, and in particular relates to a blade cross-section line laser scanning field of view planning method based on multi-constraints. Background Art

[0002] As the core components of aircraft engines and gas turbines, the surface accuracy of blades directly affects the energy conversion efficiency of the whole machine. Therefore, high-precision detection of blade surfaces is a prerequisite for ensuring their working reliability. In recent years, non-contact optical detection methods have shown good application prospects in blade surface detection due to their good flexibility and detection efficiency. According to the light source structure and data acquisition form, non-contact optical detection methods can be divided into point, line, and surface light source methods. Compared with point and surface light source methods, line light source methods can take into account both single measurement accuracy and multi-view detection efficiency. It is suitable for the detection of complex and highly reflective blade surfaces and has received widespread attention.

[0003] The Chinese invention patent with patent number 202110739936.6 discloses a blade multi-field point cloud stitching method based on overlapping features and local distance constraints. This method does not rely on the motion accuracy of the detection platform to achieve blade multi-field point cloud data registration, and realizes the two-dimensional cross-section detection of the blade. However, for complex feature blades, their complex geometric features such as thin walls, wide leaf spans, and strong spatial torsion can easily lead to significant differences between the imaging distance (viewing distance) and the incident angle (viewing angle) of the light source in different sections of any cross section of the blade during the scanning process, making it difficult to effectively guarantee the accuracy and reliability of the original scan data; at the same time, the complex geometric features of the blade also make it difficult to effectively ensure that there are sufficient overlapping areas between the multi-field data during the scanning process, increasing the difficulty of stitching the multi-field scan data and reducing the accuracy of blade detection. Summary of the invention

[0004] The purpose of the present invention is to provide a blade cross-section line laser scanning field of view planning method based on multi-constraints, which can effectively ensure the reliability of scanning data during blade detection and the accuracy of multi-field scanning data splicing.

[0005] To achieve the above object, the present invention adopts the following technical solution:

[0006] A blade section line laser scanning field of view planning method based on multi-constraints comprises the following steps:

[0007] Step 1: Obtaining the discrete contour data set and its normal vector of the blade section to be measured;

[0008] Extract the section to be measured from the 3D design model of the blade The theoretical contour data is obtained by taking the resolution Lx of the line laser sensor in the X direction as the point spacing, and the extracted theoretical contour data is discretized to obtain a discrete contour data set: , M is the cross section to be measured The number of points after discretization, the normal vector set , For discrete contour data sets The normal vector of the mth point in ;

[0009] Step 2: Acquisition of potential scanning point set;

[0010] Set the first point in the discrete contour data set of the section to be measured as the initial point , through translation transformation, the upper left end point P1 of the trapezoidal measurement range of the line laser sensor is aligned with the initial point Alignment; solve the end point according to the distance constraint ; With the initial point is the rotation point, and the initial point is transformed by rotation , the upper right end point P2 of the trapezoidal measurement range and the end point On the same straight line; discrete contour data within the trapezoidal measurement range are potential scanning points , C is the number of potential scanning points;

[0011] Step 3: Implementation of perspective constraints;

[0012] Calculate the potential scan point set The average normal vector , calculate the depth direction vector of the laser surface of the line laser sensor according to the coordinates of the four vertices (P1...P4) of the trapezoidal measurement range of the line laser sensor , through the rotation transformation to make the average normal vector and the depth of field direction vector Parallel implementation of perspective constraints;

[0013] Step 4: Implementation of sight distance constraints;

[0014] Calculate the potential scan point set Each point in the data coordinate system o d -x d y d x d The distance between the axes is taken as the projection distance , projection distance satisfy Realize the line-of-sight constraint; among them, is the potential scanning point set The projection distance of the cth point in , is the data coordinate system o d -x d yd Medium d The slope of the reference line corresponding to the axis in the blade coordinate system o-xyz, G is the intercept of the reference line;

[0015] When the conditions are met, the coordinates of the vertices (P1…P4) of the trapezoidal measurement range are updated, and the translation matrix T1 between the trapezoidal measurement range and the blade coordinate system o-xyz is calculated;

[0016] Step 5: Optimal pose Acquisition of

[0017] The rotation matrix R1 obtained in step 3 and the translation matrix T1 obtained in step 4 are used to update the potential scanning point set. The rotation matrix and the translation matrix are calculated again based on the updated potential scanning point set. The iteration is repeated several times until the rate of change of the number of potential scanning point sets is less than or equal to the rate of change threshold. Stop and get the current optimal pose ;

[0018] Among them, the rate of change of the number of potential scanning points , N k is the number of points in the potential scanning point set during the k-th calculation;

[0019] Step 6: Planning of the next field of view;

[0020] The optimal pose obtained according to step 5 Update data points in the field of view ,in The number of scanning points that fall into the field of view after the field of view planning is completed; according to the overlap area ratio constraint Determine the number of overlapping points According to the number of scanning points and Determine the initial point of the next field of view , repeat steps 2 to 5 to plan the next field of view, and complete the scanning field planning of all blade cross-sectional contours.

[0021] Furthermore, in step 2, the upper length A of the trapezoidal measurement range is used as the distance constraint to solve the discrete contour data with the initial point The point whose distance is just greater than the length A is taken as the end point .

[0022] Furthermore, the average normal vector in step 3 and the depth of field direction vector After parallelism, calculate the depth direction vector The angle between the blade coordinate system o-xyz and the y-axis , and then calculate the rotation matrix of field of view 1 .

[0023] Furthermore, the intercept G of the baseline in step 4 is solved using the least squares method.

[0024] The beneficial effects of the present invention are as follows: in response to the scanning field of view planning requirements during the complex blade line laser detection process, a multi-constrained field of view planning method is proposed in combination with the viewing distance, viewing angle and the proportion of overlapping area between scanning data during sensor imaging to effectively ensure the accuracy of blade scanning data. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 This is a schematic diagram of the field of view planning principle of the present invention.

[0026] Figure 2 This is a schematic diagram of the implementation principle of step 1 of the present invention.

[0027] Figure 3 This is a schematic diagram of the implementation principle of step 2 of the present invention.

[0028] Figure 4 This is a schematic diagram of the implementation principle of step 3 of the present invention.

[0029] Figure 5 This is a schematic diagram of the implementation principle of step 4 of the present invention.

[0030] Figure 6 This is a schematic diagram of the implementation principle of step 5 of the present invention.

[0031] Figure 7 This is a schematic diagram of the implementation principle of step 6 of the present invention. DETAILED DESCRIPTION

[0032] like Figure 1 As shown, this embodiment provides a blade cross-section line laser scanning field of view planning method based on multi-constraints. This method controls the relative posture between adjacent scanning fields of view by constraining the overlapping area ratio of adjacent fields of view. ,like Figure 1 As shown; that is, by minimizing the viewing angle deviation of all points , optimize the rotation matrix under field of view n , and then by minimizing the line of sight deviation d of all points L , optimize the translation parameters under field of view n , thus finally realizing the field of view planning.

[0033] Taking field of view 1 and field of view 2 as an example, the field of view planning method includes the following steps:

[0034] Step 1: Obtaining the discrete contour data set and its normal vector of the blade section to be measured;

[0035] Control the relative position between the line laser sensor and the blade so that the laser surface of the line laser sensor is aligned with the blade section to be measured. Total overlap.

[0036] like Figure 2 As shown in the figure, the section to be measured is extracted from the 3D design model of the blade The theoretical contour data is discretized with the resolution Lx of the line laser sensor in the X direction as the point spacing, and the normal vector information of each contour point is retained; the section to be measured The discrete contour data set is , M is the number of points after the profile of the cross section to be measured is discretized, is the three-dimensional coordinate of the mth point on the profile of the cross section to be measured, are the coordinate values ​​of the point respectively. Since the data is directly extracted from the blade design model, the coordinate values ​​of its contour points are based on the blade coordinate system o-xyz, and the normal vector set is ,in , It is the component of the normal vector of the point on the three axes of the coordinate system o-xyz.

[0037] Step 2: Acquisition of potential scanning points;

[0038] Set the section to be measured The first point in the discrete contour data set is the initial point , through translation transformation, the upper left end point P1 of the trapezoidal measurement range of the line laser sensor is aligned with the initial point Align, and use the upper length A of the trapezoidal measurement range as the distance constraint to solve the discrete contour data set and the initial point The point whose distance is just greater than the length A is represented as the end point ,like Figure 3 shown.

[0039] Starting point is the rotation point, and the trapezoidal measurement range is rotated by rotation transformation so that the initial point , the upper right end point P2 of the trapezoidal measurement range of the line laser sensor and the end point In the same straight line.

[0040] Preliminary extraction of initial points based on discrete contour data sets To end point All contour points between are regarded as potential scanning point sets of field of view 1 , C is the number of potential scanning points under field of view 1.

[0041] Step 3: Implementation of viewing angle constraints; Figure 4 As shown,

[0042] According to the normal vector set extracted in step 1 Calculate the potential scanning point set under field of view 1 in step 2 The average normal vector , and normalize it.

[0043] Then, the depth direction vector of the laser surface of the line laser sensor is calculated according to the coordinates of the four vertices (P1...P4) of the trapezoidal measurement range of the line laser sensor. , the depth direction vector and the data coordinate system o d -x d y d of y d Axis parallel.

[0044] like Figure 1 As shown in the figure, in order to ensure that the overall viewing angle deviation of all scanning points in the field of view is minimized during the scanning process, it is necessary to ensure that the average normal vector of the potential scanning points in the field of view is Depth direction vector of line laser sensor There is a parallel relationship between them, realizing the perspective constraint mechanism.

[0045] In order to ensure the viewing angle constraint, the initial point As the origin, rotate the linear laser sensor trapezoidal measurement range so that the average normal vector and the depth of field direction vector For a parallel relationship, after the rotation transformation is completed, Figure 4 The solid line trapezoid in the middle is the rotated trapezoidal measurement range.

[0046] After the rotation is completed, calculate the depth direction vector The angle between the blade coordinate system o-xyz and the y-axis , and then calculate the rotation matrix of field of view 1 .

[0047] Step 4: Implementation of view distance constraints

[0048] Calculate the potential scanning point set extracted in step 2 according to the rotated trapezoidal measurement range in step 3 Each point in the data coordinate system o d -x d y d x d The distance between the axes, i.e. the projection distance ,like Figure 5 shown.

[0049] To ensure the accuracy of the scanning data, the scanned point should be placed as close to the reference distance of the line laser sensor as possible during the field of view planning. However, due to the complex geometric characteristics of the blade cross section, it is difficult to effectively ensure that each scanning point is located near the reference distance of the line laser sensor. Therefore, the average projection distance of all potential scanning points is used as an indicator to minimize it to solve the scanning posture, that is, the line of sight constraint.

[0050] During the solution process, due to the potential scanning point set All coordinates are based on the blade coordinate system o-xyz, so the potential scanning point set will be calculated by The average projection distance is minimized to meet the viewing distance constraint condition, that is, it satisfies:

[0051] ,

[0052] In the formula, is the potential scanning point set The projection distance of the cth point in , is the data coordinate system o d -x d y d Medium d The slope of the reference line corresponding to the axis in the blade coordinate system o-xyz, G is the intercept of the reference line, and the equation of the reference line is: , As the initial point To axis x d The distance between. In the baseline equation, It can be solved by the vertices P1 and P2 in the trapezoidal measurement range, that is, .

[0053] Therefore, the core of the line-of-sight constraint is to solve the optimal intercept G. All coordinate values ​​of are known, and the least square method can be used to solve the optimal intercept G, that is, when the optimal viewing distance constraint condition is clearly met, the trapezoidal measurement range ( Figure 5 The spatial position relationship between the dotted trapezoid (in the middle) and the potential scanning point set is the corresponding field of view pose under the optimal "viewing angle-viewing distance" constraint in this state. Finally, update the coordinates of the vertices (P1...P4) of the trapezoidal measurement range at this time, and calculate the translation matrix T1 between the trapezoidal measurement range and the blade coordinate system o-xyz.

[0054] Step 3 and step 4 respectively obtain the trapezoidal measurement range rotation matrix R1 corresponding to the optimal viewing angle in field of view 1 and the translation matrix T1 corresponding to the optimal viewing distance from the perspectives of viewing angle and viewing distance, thus completing the "viewing angle-viewing distance" constraint.

[0055] Step 5: Optimal pose Acquisition of

[0056] Step 3 obtains the rotation matrix R1 through the viewing angle constraint, and step 4 obtains the translation matrix T1 through the viewing distance constraint. Therefore, the position and posture of the line laser sensor are changed, which makes the potential scanning point set in step 2 fall within the trapezoidal measurement range. Therefore, it is necessary to determine whether the potential scanning point set falling within the trapezoidal measurement range can meet the optimal viewing distance and viewing angle conditions in the current state.

[0057] This embodiment will use an iterative method to determine the difference between the current calculation result and the previous calculation result. If the difference is small, it is considered that the current calculation result can meet the optimal condition. Otherwise, further optimization calculation will be performed based on the current result. Figure 6 Specifically, based on the potential scanning point set Calculate R1 and T1, and update the potential scanning point set in real time based on R1 and T1 , and then recalculate the current The optimal constraints are considered to be met when the changes of the potential scanning point sets are small.

[0058] In order to avoid the difference in the number of potential scanning points in each field of view caused by the change in the curvature of the blade section, this implementation will use the rate of change in the number of potential scanning point sets between two adjacent iterations as an indicator to determine whether the current field of view is optimal.

[0059] Suppose the number of points in the potential scanning point set during the kth calculation is N k , and the calculated field of view position parameter is R 1k and T 1k At this time, based on R 1k and T 1k The number of points in the updated potential scanning point set is N k+1 , the rate of change of the number of potential scanning points .

[0060] Set the quantity change rate threshold to ; After step 4, determine the current Is it less than ;if , it means that the potential scanning point set has little effect on the field of view pose (R1, T1) after updating, that is, it meets the optimal constraint condition. , then continue to repeat steps 2 to 4 until the When , the iteration is exited and the rotation matrix R1 and translation matrix T1 of the current posture are recorded, that is, the posture of field of view 1 .

[0061] Step 6: Planning of the next field of view;

[0062] Take field of view 2 as an example: Based on the spatial position of the trapezoidal measurement range under field of view 1 obtained in steps 2 to 5, update the data points of field of view 1 that fall within this area ,in The number of scanning points that fall into the field of view after the planning of field of view 1 is completed.

[0063] Initial point search for field of view 2: When planning field of view 2, the overlap ratio constraint must be met, that is, the overlap ratio between field of view 1 and field of view 2 should be at least greater than , the end of the scan point set in field of view 1 must have points are also in field of view 2, such as Figure 7 Therefore, according to the number of scanning points and number of points Determine the initial point of field 2 , initial point The leftmost scanning point of field of view 2 is also the section to be measured. Discrete contour dataset Middle Points. Get the end point according to step 2 and the potential scanning point set of field of view 2 .

[0064] Determine the optimal pose of field of view 2 according to steps 3 to 5 For the subsequent field of view planning process, steps 6-2-3-4-5 are repeated continuously to complete the planning of all fields of view of the blade, and finally the planning of all scanning fields of view of the blade cross-sectional profile is completed.

[0065] The above is only a preferred implementation of the present invention, but the protection scope of the present invention is not limited thereto, and any modification and replacement based on the technical solution and inventive concept provided by the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for planning the field of view of blade cross-section line laser scanning based on multi-constraints, characterized in that: The steps include: Step 1: Obtaining the discrete contour data set and its normal vector of the blade section to be measured; Extract the section S to be measured from the three-dimensional design model of the blade u The theoretical contour data is obtained by taking the resolution Lx of the line laser sensor in the X direction as the point spacing, and the extracted theoretical contour data is discretized to obtain a discrete contour data set: M is the section to be measured S u The number of points after discretization, the normal vector set n m For discrete contour data sets The normal vector of the mth point in ; Step 2: Acquisition of potential scanning point set; Set the first point in the discrete contour data set of the section to be measured as the initial point Through translation transformation, the upper left end point P1 of the trapezoidal measurement range of the line laser sensor is aligned with the initial point Alignment; solve the end point according to the distance constraint Starting point is the rotation point, and the initial point is transformed by rotation The upper right end point P2 of the trapezoidal measurement range and the end point On the same straight line; discrete contour data within the trapezoidal measurement range are potential scanning points C is the number of potential scanning points; Step 3: Implementation of perspective constraints; Calculate the potential scanning point set Q V1 The average normal vector Nv 1-mean =(n1+…n c +…+n C ) / C, calculate the depth of field direction vector Np of the laser surface of the line laser sensor according to the coordinates of the four vertices (P1…P4) of the trapezoidal measurement range of the line laser sensor v1 , through the rotation transformation, the average normal vector Nv 1-mean and the depth direction vector Np v1 Parallel implementation of perspective constraints; Average normal vector Nv 1-mean and the depth direction vector Np v1 After parallelism, calculate the depth direction vector Np v1 The angle θ1 between the blade coordinate system o-xyz and the y-axis is used to calculate the rotation matrix of field of view 1. Step 4: Implementation of sight distance constraints; Calculate the potential scanning point set Q V1 Each point in the data coordinate system o d -x d y d x d The distance between the axes is taken as the projection distance d L-c , projection distance d L-C satisfy Realize the line-of-sight constraint; among them, is the potential scanning point set Q V1 The projection distance of the cth point in , is the data coordinate system o d -x d y d Medium d The slope of the reference line corresponding to the axis in the blade coordinate system o-xyz, G is the intercept of the reference line; When the conditions are met, the coordinates of the vertices (P1…P4) of the trapezoidal measurement range are updated, and the translation matrix T1 between the trapezoidal measurement range and the blade coordinate system o-xyz is calculated; Step 5: Obtaining the optimal posture V1 (R1, T1); The rotation matrix R1 obtained in step 3 and the translation matrix T1 obtained in step 4 are used to update the potential scanning point set. The rotation matrix and the translation matrix are calculated again based on the updated potential scanning point set. The iteration is repeated several times until the rate of change of the number of potential scanning point sets is less than or equal to the rate of change threshold. Stop and obtain the current optimal posture V1 (R1, T1); Among them, the rate of change of the number of potential scanning points N k is the number of points in the potential scanning point set during the k-th calculation; Step 6: Planning of the next field of view; Update the data points in the field of view according to the optimal posture V1 (R1, T1) obtained in step 5 Where C_new is the number of scanning points that fall into the field of view after the field of view planning is completed; according to the overlap area ratio constraint δ min Determine the number of overlapping points N1, and then determine the initial point of the next field of view based on the number of scanning points C_new and N1 Repeat steps 2 to 5 to plan the next field of view, and complete the planning of all scanning fields of view for the blade cross-sectional profile.

2. The method for planning the field of view of blade cross-section line laser scanning based on multi-constraints according to claim 1 is characterized in that: In step 2, the upper length A of the trapezoidal measurement range is used as the distance constraint to solve the distance between the discrete contour data and the initial point The point whose distance is just greater than the length A is taken as the end point 3. The method for planning the field of view of blade cross-section line laser scanning based on multi-constraints according to claim 1 is characterized in that: The intercept G of the baseline in step 4 is solved using the least squares method.