An automatic detection hammering alignment method and device for aircraft skin

Abstract

The application discloses an automatic detection hammering shaping method and device for an aircraft skin. In the application, point cloud data of a round-arc convex surface of a skin to be shaped is obtained through multiple laser detectors, a three-dimensional model of the skin to be shaped is established, curvature information of each data point of the round-arc convex surface of the skin to be shaped is calculated by using a curvature analysis tool, actual curvature deviations of each data point of the round-arc convex surface of the skin to be shaped and each data point of a standard skin round-arc convex surface are obtained, each shaping block of the skin to be shaped is divided according to the actual curvature deviations, the shaping blocks form a shaping area, then a target detection algorithm, a particle swarm optimization algorithm and finite element simulation are combined to obtain an optimal shaping scheme of each shaping area, the optimal shaping scheme is used to shape the positions of each shaping point, and after shaping, the above steps are repeated until the curvature of the round-arc convex surface of the skin after shaping meets the requirements. The application can realize intelligent automatic precise shaping of the skin to be shaped, and the shaping efficiency is relatively high.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of aircraft skin manufacturing, and particularly relates to a method and device for automatically detecting and hammering aircraft skin. BACKGROUND

[0002] As a core component of the outer shape structure of an aircraft, the aircraft skin has an important influence on the performance of key regions such as the fuselage, tail and wing, and its design complexity and size scale are increasing, and the requirements for precision and surface quality are also increasing. The skin shaping technology is a key technology in the skin processing technology, but the existing skin shaping device and method have low shaping efficiency and precision, and cannot realize high-precision shaping of the skin. Therefore, for skin shaping, a method capable of realizing high-precision shaping of the skin is developed, which has great significance for the development of the skin shaping process. SUMMARY

[0003] The application aims to overcome the deficiencies of the prior art and provide a method and device for automatically detecting and hammering aircraft skin.

[0004] To achieve the above-mentioned purpose, the application adopts the following technical solutions:

[0005] The application provides a method for automatically detecting and hammering aircraft skin, which specifically comprises the following steps:

[0006] S1, installing the skin to be shaped in the circular-arc-shaped groove of the calibration platform, the bottom surface of the circular-arc-shaped groove is provided with a plurality of cylindrical pits arranged in a rectangular shape, and a laser detector is fixed in each cylindrical pit; wherein the calibration platform is fixed on the bottom plate, the bottom plate is provided with a mechanical arm, the mechanical arm drives the hammering device to move and rotate, the steering wheel in the hammering device drives the cylinder group and the shaping hammer group to rotate, and each shaping hammer of the shaping hammer group is driven to move by one cylinder of the cylinder group;

[0007] S2, each laser detector performs three-dimensional laser scanning on the corresponding scanning area of the circular-arc-shaped convex surface of the skin to be shaped, obtains the point cloud data of each scanning area, and has an overlapping area between each adjacent two scanning areas, then performs preprocessing such as registration, denoising and simplification on the point cloud data of each scanning area, and obtains the point cloud data of the preprocessed circular-arc-shaped convex surface of the skin to be shaped;

[0008] S3, input the pre-processed skin arc convex surface point cloud data to be corrected and the pre-measured skin thickness parameters to be corrected into a three-dimensional modeling software, construct a three-dimensional model of the skin to be corrected along the normal direction of the curved surface by a normal offset algorithm, calculate the curvature information of the skin arc convex surface point cloud data to be corrected by using a curvature analysis tool; then compare the curvature information of the skin arc convex surface point cloud data to be corrected with the curvature information of the standard skin arc convex surface, and obtain the actual curvature deviation of the curvature of each data point of the skin arc convex surface to be corrected and the curvature of the data point at the same position of the standard skin arc convex surface;

[0009] S4, compare each actual curvature deviation with a preset curvature deviation threshold range, take the continuous data point region or the single discrete data point with an actual curvature deviation exceeding the preset curvature deviation threshold range as a correction block, and take all the correction blocks as a correction region, then select a correction hammer shape for each correction block in turn, and then obtain all the correction point positions, correction force sizes, correction force directions and correction paths of the correction region by combining a particle swarm optimization algorithm and a finite element simulation;

[0010] S5, correct each correction block in turn according to the correction hammer shape, correction point, correction path, correction force size and correction force direction of each correction block obtained in step S4; wherein when each correction block is corrected, first drive each cylinder and each correction hammer to rotate by a controller controlling a steering engine to drive a cutter tower base, so that the correction block corresponds to the correction hammer with the corresponding hammer shape, then drive the hammering device to move by a mechanical arm, so that the corresponding correction hammer moves along the correction path, and each time the end center of the corresponding correction hammer moves above a corresponding correction point, the mechanical arm stops driving the hammering device to move, drives the hammering device to deflect, so that the correction hammer deflects to the corresponding correction force direction, the corresponding cylinder drives the correction hammer to extend and retract, so that the correction hammer knocks the correction point with the corresponding correction force size and correction force direction, wherein the controller controls the air pump to change the air pressure in the corresponding cylinder to adjust the correction force size of the correction hammer; after the knocking correction of the correction point is completed, the mechanical arm drives the hammering device to continue to move, so that the correction hammer moves to the next correction point along the correction path;

[0011] S6, execute step S2 and step S3, detect the curvature of the skin arc convex surface after the actual correction, obtain the correction curvature deviation of the curvature of each data point of the skin arc convex surface after the actual correction and the curvature of the same data point of the standard skin arc convex surface, if each correction curvature deviation is within the preset curvature deviation threshold range, the correction of the skin to be corrected is completed, otherwise, step S4 is continuously executed until each correction curvature deviation is within the preset curvature deviation threshold range.

[0012] Preferably, the point cloud data of each scanning area is pre-processed by registration, denoising and simplification using Geomagic Studio 2012 software, and the registration process is as follows:

[0013] S21, importing the point cloud data of two adjacent scanning areas into Geomagic Studio 2012 software, converting into point cloud views, dragging the point cloud views of the two scanning areas into the same view window through VBScript script, selecting a plurality of common points with marks from the two point cloud views, aligning the common points in the two point cloud views, completing the initial splicing of the point cloud data in the two point cloud views, and thus completing the initial registration of the point cloud data of the two scanning areas;

[0014] S22, repeating step S21 to complete the initial registration of the point cloud data of all scanning areas;

[0015] S23, detecting the average distance and standard deviation of each group of data points corresponding to the position of the overlapping area between the two adjacent point cloud views;

[0016] S24, judging whether the average distance and standard deviation of each group of data points corresponding to the position of the overlapping area between the two adjacent point cloud views exceed the corresponding preset threshold value, if only the average distance exceeds the preset threshold value, using ICP algorithm to perform fine registration on the corresponding two scanning area point cloud data after initial registration, if the standard deviation exceeds the preset threshold value, using the "local registration" tool to frame the deviation area and re-perform initial registration, and returning to step S23 until the average distance and standard deviation of each group of data points corresponding to the position of the overlapping area between the two adjacent point cloud views do not exceed the corresponding preset threshold value, and thus completing the fine registration of the point cloud data of all scanning areas;

[0017] The denoising process is as follows: using the denoising function of Geomagic Studio 2012 software to remove the noise points of the point cloud data of all scanning areas after registration;

[0018] The simplification process is as follows: using the "uniform sampling" function of Geomagic Studio 2012 software to adjust the density of data points in the point cloud data of all scanning areas after denoising by uniform thinning, then reconstructing a uniform grid, and then interpolating and compensating the curved edges and areas with a curvature higher than the preset curvature.

[0019] Preferably, the shape correction hammer set is composed of a shape correction hammer one, a shape correction hammer two and a shape correction hammer three, and the shape correction hammer one and the shape correction hammer two are both cylindrical hammers, the end of the cylindrical hammer is cylindrical, the length of the end of the shape correction hammer one is less than the length of the end of the shape correction hammer two, and the shape correction hammer three is a spherical hammer, the end of the spherical hammer is spherical.

[0020] More preferably, the process of selecting the shape of the shaping hammer for each shaping block is:

[0021] ① Create a yolov11_data.yaml file in yolov11, adjust the value of the detection category number nc to 2, and the value of the category name name to ['circle','rectangle'];

[0022] ② Modify the folder path to the pre-labeled data set, which includes four types of annotations: circular point regions, irregular regions, long rectangular line regions, and short rectangular line regions. Then use the yolov11n.pt pre-trained model for training to obtain the best.pt weight file;

[0023] ③ Create a yolov11_predict.py file and modify the weight to the best.pt weight file. Detect each shaping block category for the un-pre-labeled validation set. If the conformity of each shaping block category to the actual value is 95% or more, the model training is complete. Otherwise, increase the number of training rounds and the amount of data in the data set, go back to step ②, and retrain the yolov11n.pt pre-trained model until the model training is complete.

[0024] ④ Import the point cloud data of the shaping block into Geomagic Studio 2012 software to convert it into a point cloud view. Perform shape target detection on the point cloud view of the shaping block using the trained model. If the point cloud view of the shaping block is detected as 'circle', it is classified as a circular point region. If the point cloud view of the shaping block is detected as'rectangle', calculate the diagonal length of the detection box. If the diagonal length is within the pre-set length range, it is classified as a short rectangular line region. If the diagonal length is greater than any value within the pre-set length range, it is classified as a long rectangular line region. If the point cloud view of the shaping block is not detected as 'circle' or'rectangle', it is classified as an irregular region.

[0025] ⑤ If the shaping block is a circular point region or an irregular region, select shaping hammer three. If the shaping block is a short rectangular line region, select shaping hammer one. If the shaping block is a long rectangular line region, select shaping hammer two.

[0026] Preferably, the process of planning the shaping points, shaping forces, and shaping paths of the shaping block by combining the particle swarm optimization algorithm and finite element simulation is:

[0027] (1) Determine the number of shaping points for each shaping block. Every 10 mm 2 Take a shaping point, less than 10 mm 2The calibration area of each calibration block has only one calibration point, and then all calibration point positions and calibration paths of the calibration area are planned through a particle swarm optimization algorithm;

[0028] (2) According to the calibration hammer shape, calibration point, calibration path, rough calibration force size and calibration force direction of each calibration block, the skin to be calibrated is sequentially subjected to finite element simulation calibration through Ansys software, to obtain the skin after simulation calibration and the curvature and normal direction of each data point on the skin after simulation calibration; wherein the initial calibration force direction is the normal direction at the calibration point;

[0029] (3) The simulation curvature deviation of the curvature of each data point on each calibration block of the simulated skin arc convex surface from the curvature at the same position of the standard skin arc convex surface, and the simulation direction deviation of the normal direction of each data point on each calibration block of the simulated skin arc convex surface from the normal direction at the same position of the standard skin arc convex surface are calculated, if each simulation curvature deviation does not exceed the preset curvature deviation threshold range, and each simulation direction deviation does not exceed the preset direction deviation threshold range, then the calibration force size and the calibration force direction of each calibration block at this time are the optimal calibration force size and the optimal calibration force direction, otherwise, the calibration force size and the calibration force direction are adjusted according to the simulation curvature deviation and the simulation direction deviation of all data points on each calibration block, and step (2) is returned; wherein the adjusted rough calibration force size and the adjusted calibration force direction are used each time step (2) is returned; the process of adjusting the calibration force size and the calibration force direction according to the simulation curvature deviation and the simulation direction deviation of all data points on each calibration block is as follows: if the simulation curvature deviation of any data point on the calibration block is higher than the preset curvature deviation threshold range, the current calibration force size is reduced by 2%, if the simulation curvature deviation of any data point on the calibration block is lower than the preset curvature deviation threshold range, the current calibration force size is increased by 2%, and if the simulation direction deviation of any data point on the calibration block exceeds the preset direction deviation threshold range, the current calibration force direction is tilted by 1° in the opposite direction of the deviation.

[0030] More preferably, the planning of all calibration point positions and calibration paths of the calibration area through the particle swarm optimization algorithm has the following specific process:

[0031] 1.1 Set the number of particles m, particle velocity, learning factors c1 and c2, and inertia weight ω;

[0032] 1.2 Randomly initialize all particles in the particle swarm; wherein each particle represents a candidate calibration scheme, the candidate calibration scheme includes a calibration point coordinate set, a rough calibration force size set and a calibration path, and the particle coding structure is

[0033]

[0034] points is a set of coordinates of all n calibration points in the calibration area, except that the initial value and subsequent iteration value of the calibration point coordinates of the calibration block with only a single discrete data point are set to the coordinates of the corresponding discrete data point and remain unchanged, the initial value of the calibration point coordinates of each calibration point is randomly generated within the calibration block range, forces is a set of n calibration point corresponding rough selection calibration force size, the initial value of each calibration point corresponding rough selection calibration force size is calculated according to the required deformation amount through the stress strain formula, path_order is the calibration path, each element in path_order represents the calibration order of each calibration point, and the calibration order of each calibration point is randomly generated;

[0035] 1.3 Calculate the fitness value of each particle according to the target function, and the target function is

[0036] Fitness = a · Δ κ + β · E a + γ · L path + δ · D e

[0037] Wherein, a, β, γ and δ are weight coefficients, Δ κ is the total curvature deviation after calibration of all calibration points, E a is the total calibration energy consumption of all calibration points, the path length L path is the total Euclidean distance of each adjacent two calibration points on the calibration path, and D e is the total deformation penalty of all calibration points;

[0038] The total curvature deviation Δ κ after calibration of all calibration points is calculated by the following process

[0039]

[0040] In the formula, is the curvature radius of the i-th calibration point after changing under the action of the corresponding rough selection calibration force, R is the curvature radius of the i-th calibration point on the standard skin at the same position, is the curvature radius of the i-th calibration point before calibration, F i is the rough selection calibration force of the i-th calibration point, r is the calibration influence radius, which is three times the diameter of the corresponding calibration hammer end, d j is the distance from the j-th calibration point to the i-th calibration point in the calibration area, λ is the attenuation coefficient, n is the number of calibration points in the calibration area, E is the elastic modulus of the skin to be calibrated, v is the Poisson's ratio of the skin to be calibrated, D is the bending stiffness of the skin to be calibrated, and h is the thickness of the skin to be calibrated;

[0041] The total calibration energy consumption Ea The calculation process is

[0042] E = E hammer + E path

[0043]

[0044] In the formula, E hammer is the total energy consumption required for all calibration points in the calibration area to be calibrated by the corresponding calibration hammer, E path is the total energy consumption required for the mechanical arm to move the corresponding calibration hammer along the calibration path in the calibration area, |Δx i | is the displacement of the i-th calibration point in the calibration area, and is represented by the absolute value of the difference between and η is the energy conversion efficiency, P arm is the rated power of the mechanical arm, t path is the motion time of the mechanical arm, and v arm is the moving speed of the mechanical arm.

[0045] The total deformation penalty D e of all calibration points is calculated by the formula

[0046]

[0047] 1.4 Update the velocity and position of the particle, and calculate the updated particle fitness value according to the objective function. If the updated particle fitness value is less than the updated particle fitness value, update the individual best position and the global best position. Otherwise, the individual best position and the global best position remain unchanged. The update formula of the particle velocity and the particle position is respectively

[0048] v sq (t+1) = ωv sq (t) + c1r1(p sq (t) - x sq (t)) + c2r2(p gq (t) - x sq (t))

[0049] x sq (t+1) = x sq (t) + v sq (t+1)

[0050] In the formula, q is the iteration dimension of the particle, and each calibration point has 5 iteration dimensions, including three direction coordinates (x, y, z), the rough calibration force size and the calibration path. The particle containing n calibration points has 5n iteration dimensions; t represents the iteration number; r1 and r2 are both uniform random numbers in the range of [0, 1]; p sq(t) is the individual optimal solution position of the s-th particle at the t-th iteration, p gq (t) is the global optimal solution position at the t-th iteration, x sq (t) is the s-th particle position at the t-th iteration, v sq (t) is the s-th particle velocity at the t-th iteration, s = 1, 2, 3, …, m;

[0051] 1.5 Repeat step 1.4, and terminate the iteration when the updated particle fitness value changes continuously for multiple times during the iteration process, and the change rate is lower than the preset change threshold, or the particle swarm speed is lower than the preset value, to obtain the optimal position of each calibration point and the calibration path in the calibration area.

[0052] More preferably, the initial values of the rough selection calibration forces of the calibration points in the same calibration block are equal, which are obtained by finite element analysis, as follows:

[0053] I Import the three-dimensional model of the skin to be calibrated into Ansys software, set the boundary conditions, set the contact area between the skin to be calibrated and the calibration platform as rough, set the friction coefficient, set the skin material parameters, and divide the grid.

[0054] II A coordinate system is established with the direction parallel to the central axis of the standard skin as the x-axis direction, the direction parallel to the radial direction of the standard skin as the z-axis direction, and the direction perpendicular to the x-axis and the z-axis as the y-axis direction; select the largest calibration block in the skin to be calibrated, and cut out a three-dimensional entity part with a fan ring cross section in the calibration block range, denoted as t1, and denoted as a1, the intersection line of the three-dimensional entity part parallel to the yoz plane and the calibration block, calculate the average curvature of each data point on the intersection line a2 of the neutral surface t2 and the neutral surface t1 The average distance of each data point on the intersection line a2 and each data point on the intersection line a1 is y1, and the central angle θ1 of the intersection line a1;

[0055] III Denote the neutral surface parallel to the standard skin arc convex surface at the same position on the standard skin as t3, and denote the intersection line a3 of the neutral surface t4 parallel to the yoz plane at the same position on the standard skin and the standard skin arc convex surface, extract the average curvature of each data point on the intersection line a4 of the neutral surface t4 and the neutral surface t3 The average distance of each data point on the intersection line a3 and each data point on the intersection line a4 is y0, and the central angle θ0 of the intersection line a3;

[0056] IV Calculate the strain ε of the three-dimensional entity part parallel to the yoz plane on the skin to be calibrated by a python script:

[0057]

[0058] wherein, is the arc length of intersection line a1, is the arc length of intersection line a3, and l is the arc length of intersection line a4;

[0059] V. According to the elastic modulus E of the skin material to be calibrated, the stress of the three-dimensional entity part parallel to the yoz plane on the skin to be calibrated is calculated by the formula σ=E·ε;

[0060] VI. The calibration force F of each calibration block is calculated according to the stress of the three-dimensional entity part parallel to the yoz plane on the skin to be calibrated by a python script, as the initial value of the rough selection calibration force of each calibration point in the corresponding calibration block, wherein the calibration force F of each calibration block is calculated as follows:

[0061] F=∫ S σdA

[0062] In the formula, S is the area of the corresponding calibration block, and dA is the micro-area.

[0063] The present application discloses an automatic detection hammer calibration device for aircraft skin, which comprises a hammering device, a mechanical arm, a calibration platform and a bottom plate.

[0064] The calibration platform is fixed on the bottom plate, and the upper surface of the calibration platform is provided with a circular-arc-shaped groove, the two side edges of the circular-arc-shaped groove are provided with positioning recesses, and the bottom surface of the circular-arc-shaped groove is provided with a plurality of cylindrical pits arranged in X rows and Y columns, and a laser detector is fixed in each cylindrical pit.

[0065] The hammering device comprises a tool turret base, a cylinder group and a calibration hammer group, the tool turret base is driven to rotate by a rudder, and the rudder is driven to move and rotate by the mechanical arm arranged on the bottom plate; the cylinder group is arranged on the tool turret base and comprises three cylinders uniformly distributed in the circumferential direction, the calibration hammer group comprises calibration hammer one, calibration hammer two and calibration hammer three, and the calibration hammer one, the calibration hammer two and the calibration hammer three are respectively fixed with the piston rods of the three cylinders, wherein the calibration hammer one and the calibration hammer two are columnar hammers, the ends of the columnar hammers are cylindrical, the length of the end of the calibration hammer one is less than the length of the end of the calibration hammer two, and the calibration hammer three is a spherical hammer, and the end of the spherical hammer is a spherical head.

[0066] Preferably, the mechanical arm comprises a large arm, a small arm, a wrist, a front end support, a connecting piece and a base, the base and the bottom plate constitute a rotating pair rotating around a vertical axis and are driven by a driving motor, the large arm and the base constitute a rotating pair rotating around a horizontal axis a and are driven by a driving motor two, the small arm and the large arm constitute a rotating pair rotating around a horizontal axis b and are driven by a driving motor three, the wrist and the small arm constitute a rotating pair and are driven by a driving part one, and the rotating center axis of the wrist and the small arm is perpendicular to the horizontal axis b, the front end support and the wrist constitute a rotating pair rotating around a horizontal axis c and are driven by a driving part two, and the connecting piece is fixed on the front end support; the horizontal axis a, the horizontal axis b and the horizontal axis c are parallel; the housing of the steering engine is fixed on the connecting piece, and the output shaft of the steering engine is fixed with the cutter tower base.

[0067] More preferably, the driving part one comprises a rotating shaft one, a stepping motor one and a cylindrical gear pair, the small arm comprises a connecting seat and a hollow shaft, the connecting seat is hinged with the large arm, the hollow shaft is fixed on the connecting seat, the rotating shaft one is located in the hollow shaft and constitutes a rotating pair with the hollow shaft, the housings of the two stepping motor ones are fixed with the connecting seat through the motor warehouse, the output shafts of the two stepping motor ones are connected with one end of the rotating shaft one through the two cylindrical gear pairs, and the other end of the rotating shaft one is fixed with the wrist; the driving part two comprises a stepping motor two, a driving bevel gear and a driven bevel gear, the front end support comprises a rotating shaft two and a connecting plate, the rotating shaft two constitutes a rotating pair with the wrist, the housing of the stepping motor two is fixed on the wrist, the output shaft of the stepping motor two is fixed with the driving bevel gear, the driven bevel gear is fixed in the middle of the rotating shaft two and is engaged with the driving bevel gear, the two ends of the rotating shaft two are fixed with one end of the two connecting plates arranged at intervals, and the other end of the two connecting plates is fixed with the two ends of the connecting piece.

[0068] The application has the following beneficial effects:

[0069] The present application can realize intelligent precise shaping of the skin; specifically, the present application carries out three-dimensional laser scanning on the arc convex surface of the skin to be shaped through multiple laser detectors, obtains the point cloud data of the arc convex surface of the skin to be shaped, and pre-processes the point cloud data, thereby improving the accuracy of subsequent calculation and further improving the shaping accuracy; a three-dimensional model of the skin to be shaped is established according to the pre-processed point cloud data, and the curvature information of each data point of the arc convex surface of the skin to be shaped is calculated by using a curvature analysis tool, and is compared with the curvature information of the same position of the standard skin arc convex surface, so as to obtain the actual curvature deviation of each data point of the arc convex surface of the skin to be shaped and each data point of the standard skin arc convex surface; each shaping block on the skin to be shaped is divided according to the actual curvature deviation, and the shaping blocks constitute a shaping area; then the shape type of each shaping block is divided by a target detection algorithm, the selection of the shaping hammer of each shaping block is carried out, the shaping points, the shaping force size, the shaping force direction and the shaping path of the shaping area are obtained by combining the particle swarm optimization algorithm and the finite element simulation, and then the best shaping scheme of each shaping point is obtained, thereby further improving the shaping accuracy; then the position of each shaping point is shaped by using the best shaping scheme, and the above steps are repeated after the shaping is completed, until the curvature of the arc convex surface of the skin after shaping meets the requirements, so as to realize intelligent and automatic precise shaping of the skin to be shaped, and the shaping efficiency is relatively high. BRIEF DESCRIPTION OF DRAWINGS

[0070] Figure 1 It is the overall structure schematic diagram of the present application shaping device;

[0071] Figure 2 It is the structure schematic diagram of the mechanical arm in the present application;

[0072] Figure 3 It is the structure schematic diagram of the forearm, wrist, front end support and connecting piece in the present application;

[0073] Figure 4 It is the upper surface schematic diagram of the calibration platform in the present application;

[0074] Figure 5 It is the structure schematic diagram of the cylindrical pit and laser detector in the present application;

[0075] Figure 6 It is the structure schematic diagram of the hammering device in the present application;

[0076] Figure 7 It is the structure schematic diagram of the cylinder in the present application;

[0077] Figure 8 It is the sectional view of the cylinder in the present application;

[0078] Figure 9 It is the flow chart of the present application;

[0079] Figure 10 Flow chart for determining the calibration point, calibration force size and calibration path in the present application;

[0080] Figure 11 Working schematic diagram of the calibration device in the present application;

[0081] Figure 12 Coordinate setting schematic diagram in the present application;

[0082] Figure 13 End surface schematic diagram of the three-dimensional entity part intercepted on the skin to be calibrated and the corresponding position on the standard skin. DETAILED DESCRIPTION

[0083] The present application will be further described below in combination with the drawings.

[0084] As shown in the drawings, Figure 1 The automatic detection hammer calibration device for aircraft skin in the present application comprises a hammer device A1, a mechanical arm A2, a calibration platform A3 and a base plate A4.

[0085] As shown in the drawings, Figure 2 and Figure 3 The mechanical arm A2 comprises a large arm B1, a small arm B2, a wrist B4, a front end support B5, a connecting piece B6 and a base B7. The base B7 and the base plate A4 constitute a rotating pair rotating around a vertical axis and are driven by a driving motor one, the large arm B1 and the base B7 constitute a rotating pair rotating around a horizontal axis a and are driven by a driving motor two B3, the small arm B2 and the large arm B1 constitute a rotating pair rotating around a horizontal axis b and are driven by a driving motor three F6, the wrist B4 and the small arm B2 constitute a rotating pair and are driven by a driving piece one, and the rotating center axis of the wrist B4 and the small arm B2 is perpendicular to the horizontal axis b, the front end support B5 and the wrist B4 constitute a rotating pair rotating around a horizontal axis c and are driven by a driving piece two; the horizontal axis a, the horizontal axis b and the horizontal axis c are parallel; the connecting piece B6 is fixed on the front end support B5 and is used for connecting the hammer device A1.

[0086] As shown in the drawings, Figure 4 and Figure 5 The calibration platform A3 is fixed on the base plate A4, the circular arc upper surface of the calibration platform A3 is provided with a circular arc groove, the two side edges of the circular arc groove are provided with positioning depressions, and the bottom surface of the circular arc groove is provided with a plurality of cylindrical pits D2 arranged in X rows and Y columns, and each cylindrical pit D2 is fixed with a laser detector D1. Among them, the radius of the bottom surface of the circular arc groove is consistent with the radius of the standard skin circular arc convex surface; in this embodiment, X = 20, Y = 40, the diameter of the cylindrical pit D2 is in the range of 29.95-30.05 mm, and the depth is in the range of 24.95-25.05 mm.

[0087] AsFigure 6 As shown in the figure, the hammering device A1 includes a tool turret base E5, a cylinder group and a profile correcting hammer group. The tool turret base E5 constitutes a rotary pair with the connecting piece B6 and is driven to rotate by the steering engine E2. The cylinder group is arranged on the tool turret base E5 and is composed of three cylinders arranged in the circumferential direction. The profile correcting hammer group is composed of a profile correcting hammer one E3, a profile correcting hammer two E7 and a profile correcting hammer three E8. The profile correcting hammer one E3, the profile correcting hammer two E7 and the profile correcting hammer three E8 are respectively fixed with the piston rods E10 of the three cylinders. The profile correcting hammer one E3 and the profile correcting hammer two E7 are both cylindrical hammers, the ends of which are in cylindrical shape and are suitable for profile correction of square linear areas. The length of the end of the profile correcting hammer one E3 is less than that of the profile correcting hammer two E7. The profile correcting hammer three E8 is a spherical hammer, the end of which is in spherical shape and is suitable for profile correction of circular point areas and irregular areas.

[0088] As a preferred embodiment, the driving member one includes a rotating shaft one, two step motors one F1 and two cylindrical gear pairs F2. The small arm B2 includes a connecting seat and a hollow shaft. The connecting seat is hinged with the large arm B1. The hollow shaft is fixed on the connecting seat. The rotating shaft one is located in the hollow shaft and constitutes a rotary pair with the hollow shaft. The housings of the two step motors one F1 are fixed with the connecting seat through the motor bin F7. The output shafts of the two step motors one F1 are connected with one end of the rotating shaft one through the two cylindrical gear pairs F2. The other end of the rotating shaft one is fixed with the wrist B4.

[0089] As a preferred embodiment, the driving member two includes a step motor two F3, a driving bevel gear F4 and a driven bevel gear F5. The front end support B5 includes a rotating shaft two and two connecting plates. The rotating shaft two constitutes a rotary pair with the wrist B4. The housing of the step motor two F3 is fixed on the wrist B4. The output shaft of the step motor two F3 is fixed with the driving bevel gear F4. The driven bevel gear F5 is fixed on the middle part of the rotating shaft two and is engaged with the driving bevel gear F4. The two ends of the rotating shaft two are fixed with one end of the two connecting plates. The other end of the two connecting plates is fixed with the two ends of the connecting piece B6.

[0090] As a preferred embodiment, as shown in the figure, Figure 7 The cylinder includes a cylinder cover E9, a piston, a piston rod E10 and a cylinder body E11. The cylinder body E11 is fixed on the tool turret base E5 through a cylinder base E14. The cylinder body E11 is provided with a gas pipe which is integrally formed and communicates with the inner cavity of the cylinder body E11 at one end and is open at the other end and is provided with a gas hole. The gas pipe is connected with a sealing rubber pipe E1. The open end is fixed with the cylinder cover E9 through a bolt E6. The piston is arranged in the cylinder body E11 and constitutes a sliding pair with the cylinder body E11. The piston rod E10 penetrates into the round hole of the cylinder cover E9, is fixed with the piston and constitutes a sliding pair with the round hole. The piston will not pass through the gas hole when moving in the cylinder body E11.

[0091] More preferably, a sealing rubber ring E12 is embedded on the gas pipe to ensure the air tightness between the gas pipe and the sealing rubber pipe E1.

[0092] More preferably, as shown in Figure 8 The piston is embedded with a plurality of wear rings E13 arranged at an axial distance.

[0093] Wherein, each cylinder can be connected with a gas pump (not shown in the figure) through a gas pipe E4, the steering engine, the driving motor one, the driving motor two B3, the driving motor three F6, the stepping motor one F1, the stepping motor two F3 and the gas pump are all controlled by the controller.

[0094] As shown in Figure 9 The automatic detection hammering shaping method of the aircraft skin of the application is as follows:

[0095] S1, as shown in Figure 11 The edge protrusion of the skin to be shaped is embedded in the positioning depression of the circular arc-shaped groove, so as to limit the position of the skin to be shaped and prevent the position of the skin to be shaped from shifting during the shaping process.

[0096] S2, each laser detector D1 performs three-dimensional laser scanning on the corresponding scanning area of the circular arc convex surface of the skin to be shaped, obtains the point cloud data of each scanning area, and each adjacent two scanning areas have an overlapping area. In this embodiment, the overlapping area is 20% of the area of the scanning area. After the scanning is completed, if there is local data missing or abnormal, the corresponding scanning area is re-measured to ensure the integrity and usability of the point cloud data. Then the point cloud data of each scanning area is stored in a file, and the Geomagic Studio 2012 software is used to preprocess the point cloud data of each scanning area by registration, denoising and simplification, and the preprocessed point cloud data of the circular arc convex surface of the skin to be shaped is obtained. The registration process is as follows:

[0097] S21, import the point cloud data of the adjacent two scanning areas into the Geomagic Studio 2012 software and convert it into a point cloud view. Drag the point cloud views of the two scanning areas into the same view window through the VBScript script. Select a plurality of common points (such as inflection points) with obvious marks from the two point cloud views, align the common points in the two point cloud views, complete the initial splicing of the point cloud data in the two point cloud views, and thus complete the initial registration of the point cloud data of the two scanning areas.

[0098] S22, repeat step S21 to complete the initial registration of the point cloud data of all scanning areas.

[0099] S23, detect the average distance and standard deviation of each group of data points corresponding to the position of the overlapping area point cloud in each adjacent two point cloud views.

[0100] S24, judge whether the distance average value and the standard deviation of each group of data points corresponding to the position of the overlapping region point cloud in each adjacent two point cloud views exceed the corresponding preset threshold value, if only the distance average value exceeds the preset threshold value, then use the ICP algorithm to perform fine registration on the corresponding two scanning region point cloud data after initial registration, if the standard deviation exceeds the preset threshold value, then use the "local registration" tool, frame the deviation region to re-perform initial registration, and return to step S23, until the distance average value and the standard deviation of each group of data points corresponding to the position of the overlapping region point cloud in all adjacent point cloud views do not exceed the corresponding preset threshold value, and then complete the fine registration of all scanning region point cloud data. In the embodiment, the preset threshold value of the distance average value is 0.008 m, the preset threshold value of the standard deviation is 0.019 m, and the maximum iteration number is set to 30 times and the distance tolerance is set to 0.008 m when the ICP algorithm is used to perform fine registration on the corresponding two scanning region point cloud data after initial registration.

[0101] The denoising process is: using the denoising function of Geomagic Studio 2012 software to remove the noise points of the point cloud data of all scanning regions after registration, and iterating 3 times.

[0102] The simplification process is: using the "uniform sampling" function of Geomagic Studio 2012 software, adjusting the density of data points in all scanning region point cloud data after denoising by uniform thinning, randomly removing part of the data points that are too dense, and using the absolute interval mode to set the interval between adjacent data points to 0.010 m, to reconstruct a uniform grid, ensure the uniformity of the point cloud, achieve the purpose of data simplification, and then perform interpolation compensation on the curved edge and the area higher than the preset curvature to avoid feature loss.

[0103] S3, input the point cloud data of the to-be-corrected skinning arc convex surface after pre-processing and the to-be-corrected skinning thickness parameter measured in advance into a three-dimensional modeling software, construct a three-dimensional model of the to-be-corrected skinning along the normal direction of the curved surface by using the normal offset algorithm, and calculate the curvature information of the point cloud data of the to-be-corrected skinning arc convex surface by using a curvature analysis tool (a tool provided by the three-dimensional modeling software); then compare the curvature information of the point cloud data of the to-be-corrected skinning arc convex surface with the curvature information of the standard skinning arc convex surface, to obtain the actual curvature deviation of the curvature of each data point of the to-be-corrected skinning arc convex surface and the curvature of the data point at the same position of the standard skinning arc convex surface.

[0104] S4, compare each actual curvature deviation with a preset curvature deviation threshold range, take a continuous data point region or a single discrete data point whose actual curvature deviation exceeds the preset curvature deviation threshold range as a correction block, and all correction blocks form a correction region, in the embodiment, the preset curvature deviation threshold range is -2×10 -3 μm -1~2x10 -3 μm -1 ; then the selection of the shaping hammer of each shaping block is carried out in turn, and then all the shaping point positions, shaping force sizes, shaping force directions and shaping paths of the shaping area are obtained by combining the particle swarm optimization algorithm and the finite element simulation, as shown in Figure 10

[0105] Among them, the process of selecting the shaping hammer of each shaping block is:

[0106] ① Create a yolov11_data.yaml file in yolov11, adjust the value of the detection category number nc to 2, and the value of the category name name to ['circle','rectangle'].

[0107] ② Modify the folder path to the pre-labeled data set, which includes four types of annotations: circular point area, irregular area, long rectangular line area (longer square line area), and short square line area. Then use the yolov11n.pt pre-training model for training to obtain the best.pt weight file.

[0108] ③ Create a yolov11_predict.py file and modify the weight to the best.pt weight file. Detect each shaping block category (circular point area, irregular area, long rectangular line area) in the un-pre-labeled validation set. If the compliance of each shaping block category with the actual value is more than 95%, the model training is completed, otherwise increase the training rounds and the data amount in the data set, go back to step ②, and retrain the yolov11n.pt pre-training model until the model training is completed.

[0109] ④ Import the point cloud data of the shaping block into the Geomagic Studio 2012 software to convert it into a point cloud view. Perform shape target detection on the point cloud view of the shaping block using the trained model. If the point cloud view of the shaping block is detected as 'circle', it is classified as a circular point area. If the point cloud view of the shaping block is detected as'rectangle', calculate the diagonal length of the detection box. If the diagonal length is 50mm-90mm, it is classified as a short square line area. If the diagonal length is greater than 90mm, it is classified as a long square line area. If the point cloud view of the shaping block is not detected as 'circle' or'rectangle', it is classified as an irregular area.

[0110] ⑤ If the shaping block is a circular point area or an irregular area, select the shaping hammer three E8. If the shaping block is a short square line area, select the shaping hammer one E3. If the shaping block is a long square line area, select the shaping hammer two E7. ​

[0111] As Figure 10 shown, the process of planning each calibration point, calibration force and calibration path of the calibration area is as follows by combining particle swarm optimization algorithm and finite element simulation:

[0112] (1) Determine the number of calibration points for each calibration block. Every 10mm 2 of the calibration block takes one calibration point. The calibration block less than 10mm 2 has only one calibration point. Then, the particle swarm optimization algorithm is used to plan the positions of all calibration points and the calibration path of the calibration area:

[0113] 1.1 Set the number of particles m = 50, the range of particle velocity [-10m / s, 10m / s], learning factor c1 = c2 = 2, and the initial value of inertia weight ω is 0.9, which linearly decreases to 0.4 with the iteration number;

[0114] 1.2 Randomly initialize all particles in the particle swarm; each particle represents a candidate calibration scheme, which includes a set of calibration point coordinates, a set of rough calibration force sizes and a calibration path. The particle coding structure is as follows:

[0115]

[0116] points is the set of calibration point coordinates of all n calibration points in the calibration area. Except for the calibration point coordinates of the calibration block with only one discrete data point, which are set as the coordinates of the corresponding discrete data point and remain unchanged, the calibration point coordinates of the other calibration points are randomly generated within the range of the calibration block. forces is the set of rough calibration force sizes corresponding to the n calibration points. The initial value of the rough calibration force size corresponding to each calibration point is calculated according to the required deformation by the stress-strain formula. path_order is the calibration path, and each element in path_order represents the calibration order of each calibration point. The initial value of the calibration order of each calibration point is randomly generated.

[0117] 1.3 Calculate the fitness value of each particle according to the objective function, which is as follows:

[0118] Fitness = α·Δ κ + β·E a + γ·L path + δ·D e

[0119] Wherein, α, β, γ and δ are weight coefficients, Δ κ is the total curvature deviation after calibration at all calibration points, E a is the total calibration energy consumption of all calibration points, and the path length L pathThe sum of the Euclidean distances of every two adjacent shape correction points on the shape correction path, D e The sum of the deformation penalties of all shape correction points; wherein the deformation penalty is a strong inhibition to curvature deterioration, and if the curvature of the shape correction area deteriorates after the shape correction, the deformation penalty is applied;

[0120] The sum of the curvature deviation after the shape correction of all shape correction points Δ κ The calculation process is

[0121]

[0122]

[0123] In the formula, is the curvature radius of the i-th shape correction point after the change under the action of the corresponding rough selection shape correction force, is the curvature radius of the i-th shape correction point on the standard skin at the same position, is the curvature radius of the i-th shape correction point before the shape correction, F i is the rough selection shape correction force of the i-th shape correction point, r is the shape correction influence radius, which is three times the diameter of the end of the corresponding shape hammer, d j is the distance from the j-th shape correction point to the i-th shape correction point in the shape correction area, λ is the attenuation coefficient, which is 10 mm, n is the number of shape correction points in the shape correction area, E is the elastic modulus of the skin to be shaped, v is the Poisson's ratio of the skin to be shaped, D is the bending stiffness of the skin to be shaped, and h is the thickness of the skin to be shaped.

[0124] The sum of the shape correction energy consumption of all shape correction points E a The calculation process is

[0125] E a = E hammer + E path

[0126]

[0127] In the formula, E hammer is the total energy consumption required for the shape correction of all shape correction points in the shape correction area using the corresponding shape hammer, E path is the total energy consumption required for the movement of the mechanical arm along the shape correction path with the corresponding shape hammer in the shape correction area, |Δx i | is the displacement of the i-th shape correction point during the shape correction in the shape correction area, which is represented by the absolute value of the difference between and η is the energy conversion efficiency, which is an empirical value of 0.7, P arm is the rated power of the mechanical arm, t path is the motion time of the mechanical arm, v arm is the moving speed of the mechanical arm, which is artificially set.

[0128] the sum of deformation penalty of all the shaping points D e The calculation formula is

[0129]

[0130] 1.4 Update the velocity and position of the particle, and calculate the updated particle fitness value according to the objective function. If the updated particle fitness value is less than the particle fitness value before updating, update the individual optimal solution position and the global optimal solution position, otherwise, the individual optimal solution position and the global optimal solution position remain unchanged; wherein the update formula of the particle velocity and the particle position are respectively

[0131] v sq (t+1) = ωv sq (t) + c1r1(p sq (t) - x sq (t)) + c2r2(p qq (t) - x sq (t)

[0132] x sq (t+1) = x sq (t) + v sq (t+1)

[0133] In the formula, q is the iteration dimension of the particle, each shaping point has 5 iteration dimensions, which are three direction coordinates (x, y, z), rough selection shaping force size and shaping path, so the particle containing n shaping points has 5n iteration dimensions; t represents the iteration number, r1 and r2 are uniform random numbers in the range of [0, 1]; p sq (t) is the individual optimal solution position of the s-th particle at the t-th iteration, p gq (t) is the global optimal solution position at the t-th iteration, x sq (t) is the position of the s-th particle at the t-th iteration, v sq (t) is the velocity of the s-th particle at the t-th iteration, s = 1, 2, 3, …, m.

[0134] 1.5 Repeat step 1.4, and terminate the iteration when the change rate of the updated particle fitness value in the iteration process is continuously lower than the preset change rate threshold for multiple times, or the particle swarm speed is lower than the preset value, to obtain the optimal shaping point position and shaping path of the shaping area.

[0135] Wherein, the initial values of the rough selection shaping force size of each shaping point in the same shaping block are equal, which are obtained by finite element analysis calculation, as follows:

[0136] I Import the three-dimensional model of the skin to be calibrated into Ansys software, set the boundary conditions, set the contact area between the skin to be calibrated and the calibration platform (the circular arc convex surface of the skin to be calibrated) as rough, set the friction coefficient to 0.2, set the skin material parameters as aluminum lithium alloy (2060), and divide the grid by python code.

[0137] IIAs shown in Figure 12 and Figure 13 , a coordinate system is established with the direction parallel to the central axis of the standard skin as the x-axis direction, the direction parallel to the radial direction of the standard skin as the z-axis direction, and the direction perpendicular to the x-axis and the z-axis as the y-axis direction; select the largest area calibration block in the skin to be calibrated, and cut out a three-dimensional entity part with a fan ring cross section from the skin to be calibrated within the calibration block range, denote the neutral surface parallel to the calibration block in the three-dimensional entity part as t1, denote the intersection line a1 of the neutral surface t2 parallel to the yoz plane in the three-dimensional entity part and the calibration block, and calculate the average curvature of each data point on the intersection line a2 of the neutral surface t2 and the neutral surface t1. The average distance y1 of each data point on the intersection line a2 and each data point on the intersection line a1, and the central angle θ1 included by the intersection line a1.

[0138] III Denote the neutral surface parallel to the standard skin circular arc convex surface at the same position as the three-dimensional entity part on the skin to be calibrated as t3, denote the intersection line a3 of the neutral surface t4 parallel to the yoz plane at the same position as the three-dimensional entity part on the skin to be calibrated and the standard skin circular arc convex surface, and extract the average curvature of each data point on the intersection line a4 of the neutral surface t4 and the neutral surface t3. The average distance y0 of each data point on the intersection line a3 and each data point on the intersection line a4, and the central angle θ0 included by the intersection line a3.

[0139] IV Calculate the strain ε of the three-dimensional entity part parallel to the yoz plane on the skin to be calibrated by python script:

[0140]

[0141] Wherein, is the arc length of the intersection line a1, is the arc length of the intersection line a3, and l is the arc length of the intersection line a4.

[0142] V According to the elastic modulus E of the skin to be calibrated, the stress of the three-dimensional entity part parallel to the yoz plane on the skin to be calibrated is calculated by the formula σ=E·ε (i.e. stress-strain formula).

[0143] VI Calculate the shape correction force F of each shape correction block according to the stress of the three-dimensional entity part parallel to the yoz plane on the skin to be corrected by a python script, as the initial value of the rough selection shape correction force size of each shape correction point in the corresponding shape correction block, wherein the shape correction force F of each shape correction block is calculated as follows:

[0144] F = ∫ S σdA

[0145] In the formula, S is the area of the corresponding shape correction block, and dA is the microelement area.

[0146] (2) According to the shape correction hammer shape, shape correction point, shape correction path, rough selection shape correction force size and shape correction force direction of each shape correction block, the Ansys software is used to perform finite element simulation shape correction on each shape correction block of the skin to be corrected in turn, and the skin after simulation shape correction and the curvature and normal direction of each data point on the skin after simulation shape correction are obtained; wherein the initial shape correction force direction is the normal direction at the shape correction point.

[0147] (3) Calculate the simulation curvature deviation of the curvature of each data point on each shape correction block of the simulation shape correction skin arc convex surface and the curvature of the same position of the standard skin arc convex surface, and the simulation direction deviation of the normal direction of each data point on each shape correction block of the simulation shape correction skin arc convex surface and the normal direction of the same position of the standard skin arc convex surface, if each simulation curvature deviation does not exceed the preset curvature deviation threshold range, and each simulation direction deviation does not exceed the preset direction deviation threshold range, then the shape correction force size and the shape correction force direction of each shape correction block are the best shape correction force size and the shape correction force direction, otherwise, adjust the shape correction force size and the shape correction force direction according to the simulation curvature deviation and the simulation direction deviation of all data points on each shape correction block, and return to step (2); wherein the adjusted rough selection shape correction force size and the shape correction force direction are used each time the step (2) is returned; the process of adjusting the shape correction force size and the shape correction force direction according to the simulation curvature deviation and the simulation direction deviation of all data points on each shape correction block is as follows: if the simulation curvature deviation of any data point on the shape correction block is higher than the preset curvature deviation threshold range, the current shape correction force size is reduced by 2%, if the simulation curvature deviation of any data point on the shape correction block is lower than the preset curvature deviation threshold range, the current shape correction force size is increased by 2%, and if the simulation direction deviation of any data point on the shape correction block exceeds the preset direction deviation threshold range, the current shape correction force direction is inclined by 1° in the opposite direction of the deviation. Wherein, after optimization by step (1), there will be no situation that some data points on the shape correction block have simulation curvature deviation higher than the preset curvature deviation threshold range, some data points have simulation curvature deviation lower than the preset curvature deviation threshold range, and different data points have opposite simulation direction deviation.

[0148] S5, the corresponding shaping hammer shape, shaping point, shaping path, shaping force size and shaping force direction of each shaping block are sequentially shaped according to the shaping blocks obtained in step S4; wherein, when shaping each shaping block, first, the controller controls the steering engine E2 to drive the cutter tower base E5 to rotate with each cylinder and each shaping hammer, so that the shaping block corresponds to the shaping hammer with the corresponding hammer shape, then the mechanical arm A2 drives the hammering device A1 to move, so that the corresponding shaping hammer moves according to the shaping path, and each time the corresponding shaping hammer moves to the corresponding shaping point, the mechanical arm A2 stops driving the hammering device A1 to move, drives the hammering device A1 to deflect, so that the shaping hammer deflects to the corresponding shaping force direction, the corresponding cylinder drives the shaping hammer to extend and retract, so that the shaping hammer knocks the shaping point with the corresponding shaping force size and shaping force direction, and the shaping force size of the shaping hammer is adjusted by controlling the air pressure in the corresponding cylinder, after the knocking shaping of the shaping point is completed, the mechanical arm A2 drives the hammering device A1 to continue to move, so that the shaping hammer moves to the next shaping point according to the shaping path.

[0149] S6, the curvature of the actual shaped skin arc convex surface is detected after the actual shaping, the curvature deviation of the actual shaped skin arc convex surface is obtained, and the curvature deviation of the actual shaped skin arc convex surface is obtained. If each shaping curvature deviation is within the preset curvature deviation threshold range, the shaping work of the skin to be shaped is completed, otherwise, step S4 is continued to be executed until each shaping curvature deviation is within the preset curvature deviation threshold range, and the shaping work of the skin to be shaped is completed.

Claims

1. An automatic detection of hammer straightening method for aircraft skin, characterized in that: Specifically as follows: S1, install the skin to be calibrated in the arc-shaped groove of the calibration platform, the bottom surface of the arc-shaped groove is provided with a plurality of cylindrical pits arranged in a rectangular shape, each cylindrical pit is fixed with a laser detector; wherein the calibration platform is fixed on the bottom plate, the bottom plate is provided with a mechanical arm, the mechanical arm drives the hammering device to move and rotate, the rudder in the hammering device drives the cutter tower base to drive the cylinder group and the calibration hammer group to rotate, each calibration hammer of the calibration hammer group is driven to move by one cylinder of the cylinder group; S2, each laser detector performs three-dimensional laser scanning on the corresponding scanning area of the arc convex surface of the skin to be calibrated, obtains the point cloud data of each scanning area, and each adjacent two scanning areas have an overlapping area, then the point cloud data of each scanning area is preprocessed by registration, denoising and simplification to obtain the preprocessed point cloud data of the arc convex surface of the skin to be calibrated; S3, input the preprocessed point cloud data of the arc convex surface of the skin to be calibrated and the thickness parameters of the skin to be calibrated measured in advance into a three-dimensional modeling software, construct a three-dimensional model of the skin to be calibrated along the normal direction of the curved surface by a normal offset algorithm, and calculate the curvature information of the point cloud data of the arc convex surface of the skin to be calibrated by using a curvature analysis tool; then compare the curvature information of the point cloud data of the arc convex surface of the skin to be calibrated with the curvature information of the standard skin arc convex surface, and obtain the actual curvature deviation of the curvature of each data point of the arc convex surface of the skin to be calibrated and the curvature of the data point at the same position of the standard skin arc convex surface; S4, compare each actual curvature deviation with a preset curvature deviation threshold range, take the continuous data point region or the single discrete data point whose actual curvature deviation exceeds the preset curvature deviation threshold range as a calibration block, all calibration blocks constitute a calibration region, then select the hammer shape of the calibration hammer for each calibration block in turn, and then obtain all calibration point positions, calibration force sizes, calibration force directions and calibration paths of the calibration region by combining a particle swarm optimization algorithm and finite element simulation; S5, calibrate each calibration block in turn according to the corresponding calibration hammer shape, calibration point, calibration path, calibration force size and calibration force direction of each calibration block obtained in step S4; wherein when calibrating each calibration block, first control the rudder to drive the cutter tower base to drive each cylinder and each calibration hammer to rotate, so that the calibration block corresponds to the calibration hammer with the corresponding hammer shape, then drive the hammering device to move by the mechanical arm, so that the corresponding calibration hammer moves along the calibration path, and each time the end center of the corresponding calibration hammer moves above a corresponding calibration point, the mechanical arm stops driving the hammering device to move, deflects the hammering device, deflects the direction of the calibration hammer to the corresponding calibration force direction, the corresponding cylinder drives the calibration hammer to extend and retract, so that the calibration hammer knocks the calibration point with the corresponding calibration force size and calibration force direction, wherein the controller controls the air pump to change the air pressure in the corresponding cylinder to adjust the calibration force size of the calibration hammer; after completing the knocking calibration of the calibration point, the mechanical arm drives the hammering device to continue to move, so that the calibration hammer moves to the next calibration point along the calibration path. S6, performing step S2 and step S3, the actual skin after the arc convex surface of the skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after the arc convex surface is detected, the actual skin after 2. The method of claim 1, wherein: ​ ​ ​ ​ ​ ​ ​ 3. The method of claim 1, wherein: ​ 4. The method of claim 3, wherein: ​ ① Create a yolov11_data.yaml file in yolov11, adjust the value of the detection category number nc to 2, and the value of the category name name to ['circle','rectangle']; ② Modify the folder path to the pre-labeled dataset, which includes four types of annotations: circular point regions, irregular regions, long rectangular line regions, and short rectangular line regions. Then use the yolov11n.pt pre-trained model for training to obtain the best.pt weight file; ③ Create a yolov11_predict.py file and modify the weight to the best.pt weight file. Perform category detection on each school shape block in the validation set that has not been pre-labeled. If the conformity of each school shape block category to the actual value is 95% or higher, the model training is complete. Otherwise, increase the number of training rounds and the amount of data in the dataset, go back to step ②, and retrain the yolov11n.pt pre-trained model until the model training is complete; ④ Import the point cloud data of the school shape block into Geomagic Studio 2012 software to convert it into a point cloud view. Perform shape target detection on the point cloud view of the school shape block using the trained model. If the point cloud view of the school shape block is detected as 'circle', it is classified as a circular point region. If the point cloud view of the school shape block is detected as'rectangle', calculate the diagonal length of the detection box. If the diagonal length is within the pre-set length range, it is classified as a short rectangular line region. If the diagonal length is greater than any value within the pre-set length range, it is classified as a long rectangular line region. If the point cloud view of the school shape block is not detected as 'circle' or'rectangle', it is classified as an irregular region; ⑤ If the school shape block is a circular point region or an irregular region, select school hammer three. If the school shape block is a short rectangular line region, select school hammer one. If the school shape block is a long rectangular line region, select school hammer two.

5. The method of claim 1, wherein: Through the combination of particle swarm optimization algorithm and finite element simulation, the process of planning each school shape point, school shape force, and school shape path of the school shape region is as follows: (1) Determine the number of shaping points for each shaping block, and each 10mm 2 Take a shaping point, less than 10mm 2 The shaping block has only one shaping point, and then all the shaping point positions and shaping paths in the shaping area are planned through the particle swarm optimization algorithm; (2) According to the school hammer shape, school shape point, school shape path, rough selection school shape force size, and school shape force direction of each school shape block, perform finite element simulation school shape on each school shape block of the to-be-schooled skin in Ansys software to obtain the simulated school shape skin and the curvature and normal direction of each data point on the simulated school shape skin. The initial school shape force direction is the normal direction at the school shape point. (3) calculating the simulation curvature deviation of each data point on the skin arc convex surface of each calibration block after the simulation calibration from the curvature of the same position of the standard skin arc convex surface, and the simulation direction deviation of the normal direction of each data point on the skin arc convex surface of each calibration block after the simulation calibration from the normal direction of the same position of the standard skin arc convex surface, if each simulation curvature deviation does not exceed the preset curvature deviation threshold range, and each simulation direction deviation does not exceed the preset direction deviation threshold range, then the size and direction of the calibration force of each calibration block at this time are the optimal size and direction of the calibration force, otherwise, the size and direction of the calibration force are adjusted according to the simulation curvature deviation and the simulation direction deviation of all data points on each calibration block, and step (2) is returned; wherein the adjusted rough selection calibration force size and calibration force direction are used each time step (2) is returned; the process of adjusting the size and direction of the calibration force according to the simulation curvature deviation and the simulation direction deviation of all data points on each calibration block is as follows: if the simulation curvature deviation of any data point on the calibration block is higher than the preset curvature deviation threshold range, the current calibration force size is reduced by 2%, if the simulation curvature deviation of any data point on the calibration block is lower than the preset curvature deviation threshold range, the current calibration force size is increased by 2%, and if the simulation direction deviation of any data point on the calibration block exceeds the preset direction deviation threshold range, the current calibration force direction is tilted by 1° in the opposite direction of the deviation.

6. The method of claim 5, wherein: The particle swarm optimization algorithm is used to plan the positions of all calibration points and the calibration path of the calibration area, and the specific process is as follows: 1).1 setting the number of particles m, particle velocity, learning factors c1 and c2, and inertia weight ω; 1).2 randomly initializing all particles in the particle swarm; wherein each particle represents a candidate calibration scheme, the candidate calibration scheme includes a set of calibration point coordinates, a set of rough selection calibration force sizes, and a calibration path, and the particle coding structure is points is a set of calibration point coordinates of all n calibration points in the calibration area, except that the initial value and the subsequent iterative value of the calibration point coordinates of the calibration block with only a single discrete data point are set as the coordinates of the corresponding discrete data point and remain unchanged, the initial value of the calibration point coordinates of each calibration point is randomly generated within the range of the calibration block, forces is a set of rough selection calibration force sizes corresponding to the n calibration points, the initial value of the rough selection calibration force size corresponding to each calibration point is calculated according to the required deformation variable through the stress-strain formula, path_order is a calibration path, and each element in path_order represents the calibration order of each calibration point, and the initial value of the calibration order of each calibration point is randomly generated; 1).3 calculating the fitness value of each particle according to the objective function, and the objective function is Fitness = a · Δ κ + β · E a + γ · L path + δ · D e wherein α, β, γ and δ are weight coefficients, Δ κ is the total curvature deviation after correction at all correction points, E a is the total energy consumption of correction at all correction points, the path length L path is the total Euclidean distance between every two adjacent correction points on the correction path, D e is the total deformation penalty of all correction points; Sum of curvature deviation after correction at all correction points Δ κ The calculation process is In the formula, is the curvature radius of the i-th shaping point after the change under the action of the corresponding rough selection shaping force, is the curvature radius of the i-th shaping point after the change under the action of the corresponding rough selection shaping force, is the curvature radius of the i-th shaping point after the change under the action of the corresponding rough selection shaping force, i is the rough selection shaping force of the i-th shaping point, r is the shaping influence radius, which is three times the diameter of the end of the shaping hammer, d j is the distance from the j-th shaping point to the i-th shaping point in the shaping area, λ is the attenuation coefficient, n is the number of shaping points in the shaping area, E is the elastic modulus of the material of the skin to be shaped, v is the Poisson's ratio of the material of the skin to be shaped, D is the bending stiffness of the skin to be shaped, and h is the thickness of the skin to be shaped. The total energy consumption E of all the straightening points a The calculation process is E = E hammer + E path In the formula, E hammer is the total energy consumption required for all the calibration points in the calibration area to be calibrated by the corresponding calibration hammer, E path is the total energy consumption required for the mechanical arm to move along the calibration path with the corresponding calibration hammer in the calibration area, |Δx i is the displacement of the i-th calibration point in the calibration area, |Δx and is the absolute value of the difference, η is the energy conversion efficiency, P arm is the rated power of the mechanical arm, t path is the motion time of the mechanical arm, v arm is the moving speed of the mechanical arm; The sum of the deformation penalties D of all the alignment points e The calculation formula is 1).4update the velocity and position of the particle, and calculate the fitness value of the updated particle according to the objective function, if the fitness value of the updated particle is less than the fitness value of the particle before updating, then update the individual optimal position and the global optimal position, otherwise, the individual optimal position and the global optimal position remain unchanged; wherein the update formula of the particle velocity and the particle position are respectively sq (t+1) = ωv sq (t) + c1r1(p sq (t) - x sq (t)) + c2r2(p gq (t) - x sq (t)) x sq (t+1) = x sq (t) + v sq (t+1) In the formula, q is the iteration dimension of the particle, each calibration point has 5 iteration dimensions, which are three direction coordinates (x, y, z), rough selection calibration force size and calibration path, and a particle containing n calibration points has 5n iteration dimensions; t represents the iteration number; r1 and r2 are both uniform random numbers in the range of [0, 1]; p sq (t) is the individual optimal solution position of the s-th particle at the t-th iteration, p gq (t) is the global optimal solution position at the t-th iteration, x sq (t) is the s-th particle position at the t-th iteration, v sq (t) is the s-th particle velocity at the t-th iteration, s=1, 2, 3, …, m; 1).5 repeating step 1.4, and terminating the iteration when the change rate of the updated particle fitness value in the iteration process is lower than the preset change threshold for a plurality of times in succession, or the particle swarm velocity is lower than the preset value, to obtain the optimal positions of all calibration points and the calibration path of the calibration area.

7. The method of claim 6, wherein: The initial values of the rough selection calibration force sizes of all calibration points in the same calibration block are equal and are calculated through finite element analysis, and the specific process is as follows: I import the three-dimensional model of the skin to be calibrated into Ansys software, set boundary conditions, set the contact area between the skin to be calibrated and the calibration platform as rough, set the friction coefficient, set the skin material parameters, and divide the grid; II a coordinate system is established with the direction parallel to the central axis of the standard skin as the x-axis direction, the direction parallel to the radial direction of the standard skin as the z-axis direction, and the direction perpendicular to the x-axis and the z-axis as the y-axis direction; the largest area of the skin to be corrected is selected as a correction block, and a three-dimensional entity part with a cross section of a fan ring is cut from the skin to be corrected within the range of the correction block, the neutral surface parallel to the correction block in the three-dimensional entity part is denoted as t1, the intersection line of the neutral surface t2 parallel to the yoz plane and the correction block is denoted as a1, and the average curvature of each data point on the intersection line a2 of the neutral surface t2 and the neutral surface t1 is calculated the average distance y1 of each data point on the intersection line a2 and each data point on the intersection line a1, and the central angle θ1 of the intersection line a1 III the neutral surface t3 parallel to the circular-arc convex surface of the standard skin at the same position of the three-dimensional entity part on the standard skin and the skin to be corrected, the intersection a3 of the neutral surface t4 parallel to the yoz plane at the same position of the three-dimensional entity part on the standard skin and the circular-arc convex surface of the standard skin, and the mean value of the curvatures of the data points on the intersection a4 of the neutral surface t4 and the neutral surface t3 the mean values y0 of the distances of the data points on the intersection a3 and the data points on the intersection a4 and the central angle θ0 of the intersection a3 IV Calculate the strain ε of the three-dimensional entity part parallel to the yoz plane on the skin to be calibrated through a python script: wherein is the arc length of intersection line a1, is the arc length of intersection line a3, and l is the arc length of intersection line a4; V Calculate the stress of the three-dimensional entity part parallel to the yoz plane on the skin to be calibrated according to the elastic modulus E of the skin material to be calibrated through the formula σ=E·ε; VI Calculate the calibration force F of each calibration block according to the stress of the three-dimensional entity part parallel to the yoz plane on the skin to be calibrated through a python script, which is used as the initial value of the rough selection calibration force of each calibration point in the corresponding calibration block, wherein the calibration force F of each calibration block is calculated as follows: S = ∫ s σdA In the formula, S is the area of the corresponding calibration block, and dA is the microelement area.

8. A hammer straightening device for use in a method of automatic detection of hammer straightening of an aircraft skin according to any one of claims 1 to 7, characterized in that: The calibration platform is fixed on the bottom plate, and an arc-shaped groove is formed on the upper surface of the calibration platform. Positioning recesses are formed on the two side edges of the arc-shaped groove. The bottom surface of the arc-shaped groove is provided with a plurality of cylindrical pits arranged in X rows and Y columns. A laser detector is fixed in each cylindrical pit. The hammering device includes a tool turret base, a cylinder group and a calibration hammer group. The tool turret base is driven to rotate by a rudder, and the rudder is driven to move and rotate by a mechanical arm arranged on the bottom plate. The cylinder group is arranged on the tool turret base and is composed of three cylinders uniformly distributed in the circumferential direction. The calibration hammer group is composed of calibration hammer one, calibration hammer two and calibration hammer three, and the piston rods of the three cylinders are respectively fixed with calibration hammer one, calibration hammer two and calibration hammer three. Among them, calibration hammer one and calibration hammer two are columnar hammers, the ends of which are cylindrical, and the length of the end of calibration hammer one is less than that of the end of calibration hammer two. Calibration hammer three is a spherical hammer, and the end of which is a spherical head.

9. A device for use in a method of automatic detection of hammering of a skin of an aircraft according to claim 8, characterized in that: The mechanical arm includes a large arm, a small arm, a wrist, a front end support, a connecting piece and a base. The base and the bottom plate constitute a rotating pair rotating around a vertical axis and are driven by a driving motor one. The large arm and the base constitute a rotating pair rotating around a horizontal axis a and are driven by a driving motor two. The small arm and the large arm constitute a rotating pair rotating around a horizontal axis b and are driven by a driving motor three. The wrist and the small arm constitute a rotating pair and are driven by a driving piece one. The rotation center axis of the wrist and the small arm is perpendicular to the horizontal axis b. The front end support and the wrist constitute a rotating pair rotating around a horizontal axis c and are driven by a driving piece two. The connecting piece is fixed on the front end support. The horizontal axis a, the horizontal axis b and the horizontal axis c are parallel. The housing of the rudder is fixed on the connecting piece, and the output shaft of the rudder is fixed with the tool turret base.

10. A device for use in a method of automatic detection of hammering of a skin of an aircraft according to claim 9, characterized in that: The driving member one comprises a rotating shaft one, a step motor one and a cylindrical gear pair, the small arm comprises a connecting seat and a hollow shaft, the connecting seat is hinged with the large arm, the hollow shaft is fixed on the connecting seat, the rotating shaft one is located in the hollow shaft and forms a rotating pair with the hollow shaft, the housings of the two step motor ones are fixed with the connecting seat through the motor storeroom, the output shafts of the two step motor ones are connected with one end of the rotating shaft one through the two cylindrical gear pairs, the other end of the rotating shaft one is fixed with the wrist, the driving member two comprises a step motor two, a driving bevel gear and a driven bevel gear, the front end support comprises a rotating shaft two and a connecting plate, the rotating shaft two forms a rotating pair with the wrist, the housing of the step motor two is fixed on the wrist, the output shaft of the step motor two is fixed with the driving bevel gear, the driven bevel gear is fixed on the middle part of the rotating shaft two and is engaged with the driving bevel gear, the two ends of the rotating shaft two are fixed with one end of the two connecting plates, the other end of the two connecting plates is fixed with the two ends of the connecting piece.

Images