Wing profile analysis method based on 3D point cloud data

Abstract

The present invention discloses a wing contour analysis method based on three-dimensional point cloud data, comprising: performing multi-level octree connected domain filtering processing on the three-dimensional point cloud data of the aircraft wing to obtain a filtered point cloud; performing distortion correction on the filtered point cloud based on gravity compensation to correct the filtered point cloud; registering the corrected filtered point cloud with the standard geometric model of the aircraft wing, and extracting the center axis of the aircraft wing by a weighted fusion local operator method; taking a reference point along the center axis of the aircraft wing, making a plane perpendicular to the center axis at the reference point as a reference section, and obtaining a wing section point cloud and a standard geometric section; respectively calculating the distance between the standard geometric section point and the corresponding wing section point, and using the largest distance to calculate the aircraft wing contour error. The wing contour analysis method of the present invention can realize fast and accurate measurement of wing contour, and has the characteristics of strong robustness.

Description

Technical Field

[0001] The present invention relates to the technical field of wing profile analysis, and in particular to a wing profile analysis method based on three-dimensional point cloud data. Background Art

[0002] In recent years, with the development and popularization of LiDAR and 3D scanning technology, this technology has been widely used in various fields, especially in the aviation field. Due to the particularity of the complex shape of the aircraft, non-contact measurement is required to prevent damage to the outer surface. At the same time, it is necessary to obtain dense and complete detection data of the aircraft surface, and to ensure sufficient detection accuracy and efficiency. The detection of 3D point clouds meets all the above requirements. In the field of aviation measurement and detection, 3D scanning technology is playing an increasingly important role.

[0003] It is of great significance to inspect the shape and structure of an aircraft, among which the wing is an important part of the aircraft. During the flight of the aircraft, the wing not only provides lift through its shape and structure, but also, for passenger aircraft, the thrust provided by the engine is transmitted through the wing. Therefore, it is of great significance to ensure the aerodynamic shape of the wing. During production and subsequent inspection, it is not allowed to have excessive deformation to ensure that the aircraft has good aerodynamic lift, drag characteristics, stability and maneuverability. Summary of the invention

[0004] The present invention proposes a wing profile analysis method based on three-dimensional point cloud data, which can realize fast and accurate measurement of wing profile and has the characteristics of strong robustness. The present invention uses three-dimensional point cloud data to analyze the local deformation changes of the wing, reflect the isotropic deformation of its complex surface, and provide early warning signals for aircraft safety.

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

[0006] A wing profile analysis method based on three-dimensional point cloud data comprises the following steps:

[0007] Step S1, obtaining three-dimensional point cloud data of an aircraft wing, and performing multi-level octree connected domain filtering processing on the three-dimensional point cloud data of the aircraft wing to obtain a filtered point cloud;

[0008] Step S2, performing distortion correction on the filtered point cloud based on gravity compensation, obtaining the lateral bending deformation of the aircraft wing, and correcting the filtered point cloud;

[0009] Step S3, using the local registration method of assembly reference registration, registering the corrected filtered point cloud with the standard geometric model of the aircraft wing, and extracting the center axis of the aircraft wing by using the weighted fusion local operator method on the center axis of the standard geometric model of the aircraft wing;

[0010] Step S4, taking reference points every 50 mm along the center axis of the aircraft wing, making a plane perpendicular to the center axis at the reference point as a reference section, intersecting the reference section with the registered three-dimensional point cloud data of the aircraft wing to obtain the wing section point cloud, and intersecting the reference section with the standard geometric model of the aircraft wing to obtain a standard geometric section line;

[0011] Step S5, respectively calculating the distances between the standard geometrical intercept points and the corresponding wing cross-section points, and using the largest distance to calculate the aircraft wing profile error.

[0012] Furthermore, step S1 includes the following sub-steps:

[0013] Step S101, the laser emitted by the 3D laser scanner is projected onto the surface of the aircraft wing, and the coordinates of each laser reflection point in space are calculated based on the laser reflection principle, and the coordinates are summarized to obtain the 3D point cloud data of the aircraft wing;

[0014] Step S102, establish a multi-level octree data structure of the three-dimensional point cloud data of the aircraft wing, divide the multi-level octree data structure into k levels from low to high, and in the process of raising the octree level from low to high, cluster according to the connectivity of the minimum bounding box of the current octree, and take the area with the most points in the clustering area as the main area. When the octree level of the main area reaches the highest level, if any bounding box in the six directions around the minimum bounding box is connected to the minimum bounding box, the score of the minimum bounding box is set to 1, and the points in the minimum bounding box are retained; otherwise, the score of the minimum bounding box is set to 0, and all points in the minimum bounding box are deleted.

[0015] Furthermore, step S2 includes the following sub-steps:

[0016] Step S201, transform the filtered point cloud into a coordinate system, take the plane where the wing box reference is located in the filtered point cloud as the XOY plane of the transformed coordinate system, and align the lower edge of the plane where the wing box reference is located in the filtered point cloud with the x-axis of the transformed coordinate system to obtain a coordinate transformed point cloud, and calculate the lateral bending deformation Δy of the wing:

[0017]

[0018] Where l is the length of the aircraft wing, G is the weight of the aircraft wing, z is the distance from the coordinate z value of the midpoint of the coordinate transformation point cloud to the y axis, E is the elastic modulus of the wing, and I is the moment of inertia;

[0019] Step 202: Add the lateral bending deformation Δy of the aircraft wing to the y coordinate in the coordinate transformation point cloud to correct the filtered point cloud.

[0020] Furthermore, step S3 includes the following sub-steps:

[0021] Step S301, placing the modified filtered point cloud and the standard geometric model of the aircraft wing in the same coordinate system, calculating the distance between each point in the modified filtered point cloud and the connection wing box benchmark, and minimizing the distance objective function of the registration method based on local feature weights, calculating the translation matrix and the rotation matrix, and realizing the registration of the modified filtered point cloud and the standard geometric model of the aircraft wing through the translation matrix and the rotation matrix, and obtaining the registered three-dimensional point cloud data of the aircraft wing;

[0022] Step S302: Randomly sample the registered aircraft wing three-dimensional point cloud data, and obtain the center axis of the aircraft fuselage component based on the center axis on the standard geometric model of the aircraft wing by weighted fusion of the energy function of the local operator.

[0023] Furthermore, the process of minimizing the distance objective function W is:

[0024]

[0025] Among them, R is the translation matrix, P 0 is the rotation matrix, N Y is the number of points in the corrected filtered point cloud, i is N Y The index of dis i is the coordinate y of the i-th point in the corrected filtered point cloud i The distance between the wing box and the reference, d(Ry i +P 0 , M(P)) is the coordinate y of the i-th point in the modified filtered point cloud i The distance between the standard geometric model M(P) of the aircraft wing.

[0026] Furthermore, the energy function E(Q) of the weighted fusion local operator is:

[0027]

[0028] Among them, E LOP (q j ) is the energy function of the LOP method, j is the jth projection point on the central axis of the standard geometric model of the aircraft wing, ||·|| is the K norm, θ(·) is the smoothing weight function, x a is the ath random sampling point, q j is the coordinate of the jth projection point on the central axis of the standard geometric model of the aircraft wing, is q after t iterations j Coordinate, E rep (q j ) is the repulsive energy function, λ j is the first equilibrium function, γ j is the second equilibrium function, Represents the coordinates of the ath projection point on the central axis of the standard geometric model of the aircraft wing after t iterations.

[0029] Furthermore, the distance d between the standard geometric intercept point and the corresponding wing section point i The calculation process is:

[0030]

[0031] Among them, (x i ,y i ) is the coordinate of the i-th wing section point, (x j ,y j ) is the coordinate of the i-th wing section point on the standard geometric section line.

[0032] Furthermore, the calculation process of the aircraft wing profile error L is:

[0033] L=2max(d i )

[0034] Among them, max(d i ) represents the maximum distance between the standard geometric intercept point and the corresponding wing section point.

[0035] Compared with the prior art, the present invention has the following beneficial effects: the wing contour analysis method based on three-dimensional point cloud data of the present invention processes the three-dimensional point cloud data of the aircraft wing through the connected domain filtering of the multi-level octree, which can strengthen the retention of the main area point cloud and increase the filtering of the discrete point cloud, and is suitable for the filtering scene of the aircraft wing point cloud; the gravity-corrected point cloud processing method is adopted to reduce the influence of the wing gravity on the measurement, thereby ensuring the accuracy of the method of extracting the central axis; the present invention extracts the central axis through the weighted fusion local operator method, thereby determining the position of the wing section point cloud. The wing contour analysis method based on three-dimensional point cloud data of the present invention can meet the requirements of wing contour error evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a flow chart of the wing profile analysis method based on three-dimensional point cloud data of the present invention;

[0037] Figure 2 The effect of coordinate transformation for the filtered point cloud. DETAILED DESCRIPTION

[0038] The technical solution of the present invention is further explained below in conjunction with the accompanying drawings.

[0039] like Figure 1 The flowchart of the wing profile analysis method based on three-dimensional point cloud data of the present invention comprises the following steps:

[0040] Step S1, obtaining the three-dimensional point cloud data of the aircraft wing, and performing a multi-level octree connected domain filtering process on the three-dimensional point cloud data of the aircraft wing to obtain a filtered point cloud. The multi-level octree connected domain filtering process can ensure the integrity of the three-dimensional point cloud data of the aircraft wing and have a good denoising effect on the discrete point cloud. Specifically, the process includes the following sub-steps:

[0041] Step S101, the laser emitted by the 3D laser scanner is projected onto the surface of the aircraft wing, and the coordinates of each laser reflection point in space are calculated based on the laser reflection principle, and the coordinates are summarized to obtain the 3D point cloud data of the aircraft wing;

[0042] Step S102, establish a multi-level octree data structure of the three-dimensional point cloud data of the wing of the aircraft, divide the multi-level octree data structure into k levels from low to high, in the process of raising the octree level from low to high, cluster according to the connectivity of the minimum bounding box of the current octree, and take the area with the most points in the cluster area as the main area. When the octree level of the main area reaches the highest level, if any of the six directions around the minimum bounding box is connected to the minimum bounding box, the score of the minimum bounding box is set to 1, and the points in the minimum bounding box are retained; otherwise, the score of the minimum bounding box is set to 0, and all points in the minimum bounding box are deleted. The filtering effect of the traditional filtering method on the main body of the point cloud and the filtering strength of the discrete points of the point cloud are the same, but in order to ensure sufficient measurement accuracy, the main point cloud corresponding to the wing of the aircraft should be retained as much as possible. The multi-level octree filtering in the method of the present invention can strengthen the retention of the main area point cloud and increase the filtering effect on the discrete point cloud, which is suitable for the filtering application scenario of the wing point cloud.

[0043] Step S2: Since the aircraft wing is a flexible structure, it will bend laterally due to its own gravity. To avoid the influence of the bending deformation, the filtered point cloud is subjected to distortion correction based on gravity compensation to obtain the lateral bending deformation of the aircraft wing, and the filtered three-dimensional point cloud data of the aircraft wing is corrected. The step S2 specifically includes the following sub-steps:

[0044] Step S201, transform the filter point cloud into a coordinate system, take the plane where the filter point cloud connected to the aircraft fuselage is located as the XOY plane of the transformation coordinate system, and align the lower edge of the plane where the filter point cloud connected to the aircraft fuselage is located with the x-axis of the transformation coordinate system to obtain a coordinate transformation point cloud. The above coordinate transformation can reduce the amount of calculation in one dimension in subsequent calculations and simplify the calculation process. The effect of coordinate transformation of the filter point cloud is as follows: Figure 2 As shown. Calculate the lateral bending deformation of the wing Δy:

[0045]

[0046] Where l is the length of the aircraft wing, G is the weight of the aircraft wing, z is the distance from the coordinate z value of the midpoint of the coordinate transformation point cloud to the y axis, E is the elastic modulus of the wing, and I is the moment of inertia;

[0047] Step 202: Add the lateral bending deformation Δy of the aircraft wing to the y coordinate in the coordinate transformation point cloud to correct the filtered three-dimensional point cloud data of the aircraft wing.

[0048] Step S3, using the local registration method of assembly reference registration, registering the corrected filtered point cloud with the standard geometric model of the aircraft wing, extracting the center axis of the aircraft wing by weighted fusion of local operators, extracting the center axis of the aircraft wing by the above method, the center axis of the aircraft wing can provide the necessary vertical reference for the generation of the section for subsequent calculation of the section profile, and facilitate the generation of the section perpendicular to the center axis; specifically including the following sub-steps:

[0049] Step S301, placing the modified filtered point cloud and the standard geometric model of the aircraft wing in the same coordinate system, calculating the distance between each point in the modified filtered point cloud and the connection wing box benchmark, and minimizing the distance objective function of the registration method based on local feature weights, calculating the translation matrix and the rotation matrix, and realizing the registration of the modified filtered point cloud and the standard geometric model of the aircraft wing through the translation matrix and the rotation matrix, and obtaining the registered three-dimensional point cloud data of the aircraft wing;

[0050] The process of minimizing the distance objective function W in the present invention is:

[0051]

[0052] Among them, R is the translation matrix, P 0 is the rotation matrix, N Y is the number of points in the corrected filtered point cloud, i is N Y The index of dis i is the coordinate y of the i-th point in the corrected filtered point cloud i The distance between the wing box and the reference, d(Ry i +P 0 , M(P)) is the coordinate y of the i-th point in the modified filtered point cloud i The distance between the standard geometric model M(P) of the aircraft wing.

[0053] Step S302: Randomly sample the registered aircraft wing three-dimensional point cloud data, and obtain the center axis of the aircraft fuselage component based on the center axis on the standard geometric model of the aircraft wing by weighted fusion of the energy function of the local operator.

[0054] Although the existing LOP operator energy function converges too fast, it will cause the problem of inaccurate convergence position. The energy function of the weighted fusion local operator of the present invention adds a repulsive energy function so that the energy function converges closer to the central axis. The energy function E(Q) of the weighted fusion local operator of the present invention is:

[0055]

[0056] Among them, E LOP (q j ) is the energy function of the LOP method, j is the jth projection point on the central axis of the standard geometric model of the aircraft wing, ||·|| is the K norm, θ(·) is the smoothing weight function, x a is the ath random sampling point, q j is the coordinate of the jth projection point on the central axis of the standard geometric model of the aircraft wing, is q after t iterations j Coordinate, E rep (q j ) is the repulsive energy function, λ j is the first equilibrium function, γ j is the second equilibrium function, Represents the coordinates of the ath projection point on the central axis of the standard geometric model of the aircraft wing after t iterations.

[0057] Step S4, taking reference points every 50 mm along the center axis of the aircraft wing, making a plane perpendicular to the center axis at the reference point as a reference section, intersecting the reference section with the registered three-dimensional point cloud data of the aircraft wing to obtain the wing section point cloud, and intersecting the reference section with the standard geometric model of the aircraft wing to obtain a standard geometric section line;

[0058] Step S5: Calculate the distance between the standard geometrical intercept point and the corresponding wing section point respectively The largest distance is used to calculate the aircraft wing profile error L = 2max (d i ), where (x i ,y i ) is the coordinate of the i-th wing section point, (x j ,y j ) is the coordinate of the i-th wing section point on the standard geometric section line, max(d i) represents the maximum distance between the standard geometrical intercept point and the corresponding wing section point. Through the geometrical intercept method of wing section comparison, for the wing, the direction of error deviation is limited to the section area, so that the measured error value has a certain standard; at the same time, the centerline of the aircraft wing and the corresponding reference section can be used for positioning, and the area of ​​the aircraft wing profile error can be quickly matched.

[0059] The wing profile analysis method based on three-dimensional point cloud data of the present invention is evaluated for its accuracy using coverage rate and root mean square error quantitative evaluation indicators. The coverage rate and root mean square error are 0.006 and 94% respectively, indicating that the wing profile analysis method of the present invention has high accuracy and strong robustness, and can be widely used in industrial inspection scenarios.

Claims

1. A wing profile analysis method based on three-dimensional point cloud data, characterized in that: The steps include: Step S1, obtaining three-dimensional point cloud data of an aircraft wing, and performing multi-level octree connected domain filtering processing on the three-dimensional point cloud data of the aircraft wing to obtain a filtered point cloud; Step S2, performing distortion correction on the filtered point cloud based on gravity compensation, obtaining the lateral bending deformation of the aircraft wing, and correcting the filtered point cloud; Step S3, using the local registration method of assembly reference registration, registering the corrected filtered point cloud with the standard geometric model of the aircraft wing, and extracting the center axis of the aircraft wing by using the weighted fusion local operator method on the center axis of the standard geometric model of the aircraft wing; Step S4, taking reference points every 50 mm along the center axis of the aircraft wing, making a plane perpendicular to the center axis at the reference point as a reference section, intersecting the reference section with the registered three-dimensional point cloud data of the aircraft wing to obtain the wing section point cloud, and intersecting the reference section with the standard geometric model of the aircraft wing to obtain a standard geometric section line; Step S5, respectively calculating the distances between the standard geometric intercept points and the corresponding wing section points, and using the largest distance to calculate the aircraft wing profile error.

2. The wing profile analysis method based on three-dimensional point cloud data according to claim 1, characterized in that: Step S1 includes the following sub-steps: Step S101, the laser emitted by the 3D laser scanner is projected onto the surface of the aircraft wing, and the coordinates of each laser reflection point in space are calculated based on the laser reflection principle, and the coordinates are summarized to obtain the 3D point cloud data of the aircraft wing; Step S102, establish a multi-level octree data structure of the three-dimensional point cloud data of the aircraft wing, divide the multi-level octree data structure into k levels from low to high, and in the process of raising the octree level from low to high, cluster according to the connectivity of the minimum bounding box of the current octree, and take the area with the most points in the clustering area as the main area. When the octree level of the main area reaches the highest level, if any bounding box in the six directions around the minimum bounding box is connected to the minimum bounding box, the score of the minimum bounding box is set to 1, and the points in the minimum bounding box are retained; otherwise, the score of the minimum bounding box is set to 0, and all points in the minimum bounding box are deleted.

3. The wing profile analysis method based on three-dimensional point cloud data according to claim 1, characterized in that: Step S2 includes the following sub-steps: Step S201, transform the filtered point cloud into a coordinate system, take the plane where the wing box reference is located in the filtered point cloud as the XOY plane of the transformed coordinate system, and align the lower edge of the plane where the wing box reference is located in the filtered point cloud with the x-axis of the transformed coordinate system to obtain a coordinate transformed point cloud, and calculate the lateral bending deformation Δy of the wing: Where l is the length of the aircraft wing, G is the weight of the aircraft wing, z is the distance from the coordinate z value of the midpoint of the coordinate transformation point cloud to the y axis, E is the elastic modulus of the wing, and I is the moment of inertia; Step 202: Add the lateral bending deformation Δy of the aircraft wing to the y coordinate in the coordinate transformation point cloud to correct the filtered point cloud.

4. The wing profile analysis method based on three-dimensional point cloud data according to claim 1, characterized in that: Step S3 includes the following sub-steps: Step S301, placing the modified filtered point cloud and the standard geometric model of the aircraft wing in the same coordinate system, calculating the distance between each point in the modified filtered point cloud and the connection wing box benchmark, and minimizing the distance objective function of the registration method based on local feature weights, calculating the translation matrix and the rotation matrix, and realizing the registration of the modified filtered point cloud and the standard geometric model of the aircraft wing through the translation matrix and the rotation matrix, and obtaining the registered three-dimensional point cloud data of the aircraft wing; Step S302: Randomly sample the registered aircraft wing three-dimensional point cloud data, and obtain the center axis of the aircraft fuselage component based on the center axis on the standard geometric model of the aircraft wing by weighted fusion of the energy function of the local operator.

5. The wing profile analysis method based on three-dimensional point cloud data according to claim 4 is characterized in that: The process of minimizing the distance objective function W is: Among them, R is the translation matrix, P0 is the rotation matrix, N Y is the number of points in the corrected filtered point cloud, i is N Y The index of dis i is the coordinate y of the i-th point in the corrected filtered point cloud i The distance between the wing box and the reference, d(Ry i +P0,M(P)) is the coordinate y of the i-th point in the modified filtered point cloud i The distance between the standard geometric model M(P) of the aircraft wing.

6. The wing profile analysis method based on three-dimensional point cloud data according to claim 4 is characterized in that: The energy function E(Q) of the weighted fusion local operator is: Among them, E LOP (q j ) is the energy function of the LOP method, j is the jth projection point on the central axis of the standard geometric model of the aircraft wing, ‖·‖ is the K norm, θ(·) is the smoothing weight function, x a is the ath random sampling point, q j is the coordinate of the jth projection point on the central axis of the standard geometric model of the aircraft wing, is q after t iterations j Coordinate, E rep (q j ) is the repulsive energy function, λ j is the first equilibrium function, γ j is the second equilibrium function, Represents the coordinates of the ath projection point on the central axis of the standard geometric model of the aircraft wing after t iterations.

7. The wing profile analysis method based on three-dimensional point cloud data according to claim 1, characterized in that: The distance d between the standard geometric intercept point and the corresponding wing section point i The calculation process is: Among them, (x i ,y i ) is the coordinate of the i-th wing section point, (x j ,y j ) is the coordinate of the i-th wing section point on the standard geometric section line.

8. The wing profile analysis method based on three-dimensional point cloud data according to claim 1, characterized in that: The calculation process of the aircraft wing profile error L is: L=2max(d i ) Among them, max(d i ) represents the maximum distance between the standard geometric intercept point and the corresponding wing section point.

Images