Method for predicting deformation of curved surface conformal antenna in assembly process

Abstract

The invention provides a curved surface conformal antenna assembly process deformation prediction method, and relates to the field of assembly process intelligent prediction. Comprising the following steps: establishing a workpiece profile library based on actually measured profile data and simulated profile data, and carrying out finite element simulation analysis on the workpiece profile library in the assembly process with different process parameters to obtain a deformation result after assembly; by taking the curved surface manufacturing error and the process parameters as input and the post-assembly deformation result as output, training and building an assembly deformation prediction model based on the GAN; and importing the profile data detected in real time into the assembly deformation prediction model, predicting assembly deformation images under different process parameters, and taking the result with the minimum error between the assembly deformation images and the ideal curved surface model as the optimal assembly process parameter meeting the current condition. According to the method, deformation in the curved surface assembly process is predicted in real time, optimal assembly process parameters can be determined according to actual curved surface manufacturing errors, and the problems that the curved surface antenna is invisible in the assembly process and difficult to install and adjust are solved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent prediction in the assembly process, and particularly to a method for predicting the deformation of a curved conformal antenna during the assembly process. Background Art

[0002] Due to conforming to the surface of the carrier platform, the curved conformal antenna does not damage the external shape structure and aerodynamic characteristics of the carrier, and can effectively expand the radar aperture, increase the detection range, and expand the detection scope, becoming one of the research hotspots in the antenna field. However, due to the complex structure of the curved conformal antenna, it is difficult to form by traditional manufacturing methods. Therefore, additive manufacturing methods are mostly used to separately process multi-layer heterogeneous structures, and then assembled and formed through processes such as gluing and riveting. However, due to the problems of low manufacturing stiffness and poor manufacturing accuracy in the additive manufacturing method itself, and the characteristics of a large number and small size of the characteristic structures that undertake the electrical interconnection function on the antenna structure, the deformation during the assembly process has become an important factor affecting the assembly performance. In the existing assembly process, most are manual assembly, with high labor costs and low efficiency. And due to the invisibility of the interlayer structure, the assembly process is completed solely based on manual experience, requiring repeated attempts, making it difficult to ensure the assembly accuracy and quality. Often, only post-detection means can be used to evaluate the assembly quality, resulting in a low yield rate and low efficiency.

[0003] Therefore, an online curved surface deformation prediction method for the assembly process is needed to achieve online regulation of the assembly process.

[0004] The Chinese invention patent with the publication number CN109918755A proposes a method for predicting the assembly deformation of low-rigidity parts based on point cloud data. A three-dimensional laser scanner is used to scan the low-rigidity part to be measured, the surface information point cloud data is obtained and preprocessed, the three-dimensional transformation of the point cloud coordinates is completed in the three-dimensional coordinate system, and two-dimensional mesh division is completed using the x and y plane coordinates to obtain a quadrilateral mesh composed of ordered representative points of the point cloud, and it is input into the finite element analysis software for constraint and displacement loading to obtain the shape change of the low-rigidity part after assembly. This method may consider the actual manufacturing errors of the workpiece and rely solely on finite element simulation analysis for deformation prediction, but does not consider the real-time regulation of the assembly process, resulting in low prediction efficiency.

[0005] The Chinese invention patent with the publication number CN116108762A proposes a method for predicting the assembly deformation of large composite components using force sensors. The external force and deformation data are obtained through force sensors and three-dimensional laser scanning equipment, and the inner surface deformation of the panel after applying force during the assembly process is predicted using a neural network. This method builds a neural network prediction model based on the finite element analysis method to predict the true state of the workpiece shape, but does not consider the influence of the initial deformation of the workpiece and the coupling of multi-parameter assembly processes. Summary of the Invention

[0006] Objective of the Invention: To propose a method for predicting deformation during the assembly process of a curved conformal antenna to solve the above problems existing in the prior art.

[0007] The present invention proposes a method for predicting deformation during the assembly process of a curved conformal antenna, including the following steps:

[0008] Based on the measured surface data and the simulated surface data, establish a workpiece surface library, and conduct finite element simulation analysis on the assembly process with different process parameters for the workpiece surface library to obtain the deformation result after assembly;

[0009] Using the surface manufacturing error and process parameters as inputs and the deformation result after assembly as the output, train and build an assembly deformation prediction model based on GAN;

[0010] Import the real-time detected surface data into the assembly deformation prediction model to predict the assembly deformation images under different process parameters, and use the result with the minimum error between the assembly deformation image and the ideal surface model as the optimal assembly process parameters under the current conditions.

[0011] In a further embodiment, the measured surface data is obtained through the following method:

[0012] Use a point cloud data acquisition device to obtain a number of measured point cloud data of the assembly surface, filter and splice the measured point cloud data to obtain the overall surface as the measured surface data.

[0013] In a further embodiment, the simulated surface data is obtained through the following method:

[0014] Collect a number of groups of measured surface data, register them with the ideal surface model to obtain the manufacturing error law of the surface, and construct the simulated surface data based on the manufacturing error law.

[0015] In a further embodiment, assume the ideal surface mathematical model is , for the measured surface data and the ideal surface mathematical model Perform ICP registration with the screw holes as features, align the measured surface data with the ideal surface model and perform coordinate transformation;

[0016] Project the measured points in the measured surface data along the surface normal direction of the ideal surface model to calculate the projection distance to generate an error field; where represents the measured point in the measured point cloud data , represents the corresponding point in the ideal surface model corresponding to the measured point , represents Point normal direction vector;

[0017] Divide the ideal surface model along the surface contour into meshes of specifications, and statistically calculate the mean value , maximum value and standard deviation ;

[0018] Repeat the above steps multiple times to obtain multiple sets of manufacturing error data.

[0019] In a further embodiment, construct simulated surface data based on the manufacturing error law, specifically including:

[0020] Use the semivariogram to calculate and analyze the spatial distribution characteristics of each set of manufacturing error data in the X, Y, and Z directions respectively:

[0021]

[0022]

[0023]

[0024] In the formula, , , are the X, Y, and Z axis coordinates of the measured points respectively; , , are the point spacings in the X, Y, and Z axis directions respectively; , , are the number of point pairs with spacings of , , respectively;

[0025] Fit the covariance model, calculate the correlation length and variance parameters, and establish an error model;

[0026] Generate a random simulation error field based on Kriging interpolation, and generate M sets of simulated surface data containing simulated manufacturing errors.

[0027] In a further embodiment, use the surface manufacturing error and process parameters as inputs and the deformation result after assembly as the output to train and build a GAN-based assembly deformation prediction model, specifically including:

[0028] Taking the contour corner points and screw hole features as the reference, grids with the same number of nodes are respectively arranged for the ideal surface model and the measured surface data. The mapping relationship between the ideal surface model and the measured surface data is established using the nodes, where the node set of the ideal surface model is PI, and the node set of the measured surface data is PR;

[0029] Taking the ideal surface model as the reference, the predicted coordinates X D , Y D , Z D of the surface nodes after assembly deformation and the actual coordinates X I , Y I , Z I of the surface nodes are respectively mapped to the red, green, and blue numerical values in the RGB color space through normalization processing corresponding to the coordinate deviations from the theoretical surface corresponding contour node coordinates X R , Y R , Z R to prepare the input and output images for the training of the generative adversarial network;

[0030] The data set is divided into a training set, a validation set, and a test set. The generator inputs the surface manufacturing error image and the loading conditions and outputs the predicted surface deformation result. The discriminator inputs the simulation deformation result or the deformation result predicted by the generator and outputs the discrimination result, and they are alternately trained.

[0031] In a further embodiment, taking the ideal surface model as the reference, the actual coordinates X I , Y I , Z I of the surface nodes and the coordinate deviations from the theoretical surface corresponding contour node coordinates X R , Y R , Z R are normalized to [0, 255] and respectively mapped to the R, G, B numerical values to obtain the RGB error image:

[0032]

[0033]

[0034]

[0035] Wherein, , , are respectively the numerical values mapped to red, green, and blue for the i-th point; , , are respectively the coordinate deviations of the i-th point in the X, Y, and Z directions; , , They are respectively the minimum values of the coordinate deviations in the X, Y, and Z directions; , , They are respectively the maximum values of the coordinate deviations in the X, Y, and Z directions.

[0036] In a further embodiment, the shape surface data detected in real time is imported into the assembly deformation prediction model to predict the assembly deformation images under different process parameters, specifically including:

[0037] The shape surface data of the current curved surface conformal antenna is obtained in real time by using a point cloud data acquisition device and registered with the ideal curved surface model to obtain the simulated shape surface data containing manufacturing errors. The simulated shape surface data is reduced to a two-dimensional image, which is used as one of the inputs of the assembly deformation prediction model;

[0038] Taking the minimization of the deformation error after assembly as the optimization goal, defining the process parameter range, calling the assembly deformation prediction model, calculating the objective function value, updating the process parameters, and repeating the calculation and iterative optimization until the convergence condition is met.

[0039] In a further embodiment, the assembly deformation prediction model includes a generator and a discriminator; the generator takes a random RGB error image with a resolution and a process parameter vector as inputs and outputs a predicted deformation field RGB image. The discriminator inputs real samples or generated samples and outputs the probability that the sample is true;

[0040] Inputting real data and generated data, updating the discriminator parameters to maximize the discrimination accuracy; deceiving the discriminator with the generated data and updating the generator parameters to minimize the discrimination probability of the discriminator for the generated samples;

[0041] Training the generator once every predetermined number of times of training the discriminator; alternately and iteratively training the generator and the discriminator to realize inputting a random manufacturing error image and a process parameter vector and outputting an assembly deformation image;

[0042] Performing multi-constraint optimization and solution for the screw tightening sequence and tightening torque to ensure that the gap between the assembly deformation image after assembly and the ideal curved surface model is minimized.

[0043] In addition, the present invention also provides an electronic device, which includes: a processor and a memory storing computer program instructions; when the processor executes the computer program instructions, the deformation prediction method for the assembly process of the above-mentioned curved surface conformal antenna is realized.

[0044] Compared with the prior art, the present invention has at least the following beneficial effects:

[0045] (1) The present invention realizes real-time prediction of deformation during the curved surface assembly process, and can determine the optimal assembly process parameters according to the actual curved surface manufacturing errors, solving the problems of invisibility and difficult installation and adjustment during the curved surface antenna assembly process.

[0046] (2) The present invention uses the multi-resolution registration method for high-precision point cloud stitching, effectively improving the speed of point cloud stitching and providing a solid foundation for the real-time prediction of assembly deformation.

[0047] (3) The present invention predicts the curved surface deformation based on GAN using images. Compared with the traditional finite element analysis prediction method, the prediction speed is increased by 20%, and it is more suitable for the real-time requirements of actual production.

[0048] (4) The present invention reduces the three-dimensional space point cloud data to the two-dimensional color image space for processing, and the visualization difficulty of the prediction process is low and the effect is remarkable. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a flowchart of the deformation prediction method for the curved surface conformal antenna assembly process in the embodiment.

[0050] Figure 2 It is a schematic diagram of the multi-resolution point cloud registration method proposed for high-density point cloud stitching in the embodiment.

[0051] Figure 3 It is a schematic diagram of the error mapping model from three-dimensional point cloud to two-dimensional image in the embodiment.

[0052] Figure 4 It is an example diagram of the RGB error image in the embodiment.

[0053] Figure 5 The training flowchart of the assembly deformation prediction model in the embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0054] In the following description, numerous specific details are given to provide a more thorough understanding of the present invention. However, it will be apparent to one of ordinary skill in the art that the present invention may be practiced without one or more of these details. In other instances, well-known technical features have not been described in order to avoid obscuring the present invention.

[0055] This embodiment discloses a deformation prediction method for the curved surface conformal antenna assembly process considering manufacturing errors. The flowchart is shown in Figure 1 and the specific steps are as follows:

[0056] Based on the visual guidance of a large field-of-view camera, move a high-precision laser scanner or a high-precision binocular camera, and scan multiple times to obtain T groups of original three-dimensional curved surface point cloud data containing manufacturing errors :

[0057]

[0058] Among them, ; N is the number of point cloud data points obtained by a single scan or shot.

[0059] For the original data of the 3D surface point cloud perform outlier removal. Based on statistical filtering, remove the points that deviate from the mean by k times the standard deviation, and retain the points that meet the conditions . Among them, represents the average value of the z coordinates of all point cloud data in a single scan or shot, represents the standard deviation of the z coordinates of all point cloud data in a single scan or shot.

[0060] Adopt voxel grid filtering, retain the centroid points inside the grid, define the voxel size as v, and initialize v according to the empirical formula . Among them, is the average density of the original point cloud (points / mm²).

[0061] Calculate the index of each voxel from the point coordinates :

[0062] , , Among them, is the minimum vertex coordinate of the point cloud bounding box, represents rounding down.

[0063] Assign all points to the corresponding voxels according to their coordinates. Each voxel contains the following set of points: ;

[0064] For each non-empty voxel, calculate the centroid of all points inside it as the representative point: , where n is the number of points in the voxel, and further calculate by weighting: , retaining the normal vector feature.

[0065] Output the filtered point cloud, that is, the set of all voxel representative points: , , where M is the number of points retained after filtering.

[0066] Perform rough registration according to the positions of the high-precision scanner or binocular camera target points under the field of view of the large field-of-view camera;

[0067] For T groups of filtered point cloud sets construct Gaussian pyramids respectively, generate L layers of point cloud layers with different resolutions, and the voxel size of the l layer , and the resolution , where is the initial voxel size, initialized according to the empirical formula , the set of voxel representative points in the l-th layer is , .

[0068] For adjacent point clouds and the l-th layer point cloud and feature extraction is performed using the FPFH descriptor respectively, and the feature correspondence is calculated to obtain the matching score , and the matching pairs with scores higher than the threshold are retained. Randomly sample the matching pairs, estimate the rigid body transformation, and calculate the initial transformation matrix for each layer .

[0069] Starting from the coarsest layer l = 0, optimize the transformation parameters layer by layer from bottom to top. The number of iterations decreases for each layer. Let the optimized transformation in the l-th layer be , and use it as the initial value for the (l + 1)-th layer. In the l-th layer, the registration error between the source point cloud and the target point cloud is: , where , is the corresponding point pair, is the weight, , where , is the outlier threshold.

[0070] Introduce a regularization term to limit the mutation of the inter-layer transformation, , where is the balance factor, is the Frobenius norm. Use optimization algorithms such as SVD or Levenberg-Marquardt to solve the optimal transformation: .

[0071] Integrate the registration results of each layer, introduce the adjacent layer motion consistency constraint to avoid registration mutations caused by resolution changes. . Construct the global error function :

[0072]

[0073] where represents mapping the SE(3) transformation matrix to the Lie algebra space se(3), represents the weighted norm of the Lie algebra vector, represents the weight of the ICP error term for each layer, which increases with the increase of the resolution, Represents the weight of the smoothing term, balancing the registration accuracy and the continuity of the transformation;

[0074] Regarding each layer of transformation as the nodes for graph optimization and the differences between layers as edge constraints, and using a sparse solver to solve for the optimal of adjacent point cloud patches .

[0075] Repeat the above steps to obtain the complete point cloud data of the surface antenna with manufacturing errors :

[0076]

[0077] Among them, ; N is the number of points in the point cloud data obtained from a single scan or shot.

[0078] The above process is shown in Figure 2 .

[0079] Assume the ideal surface mathematical model is , perform ICP registration on the measured point cloud data and the ideal surface mathematical model using the screw holes as features, align the point cloud data to the ideal surface model and perform coordinate transformation;

[0080] Project the measured points in the measured point cloud data along the surface normal direction of the ideal surface model to calculate the projection distance and generate an error field;

[0081] Divide the ideal surface model along the surface contour into -sized grids, and statistically calculate the mean , maximum value and standard deviation of all point cloud errors within each grid;

[0082] Repeat the above steps n times to obtain multiple sets of measured surface error data.

[0083] Use the semi-variogram to calculate and analyze the spatial distribution characteristics of each set of errors in the X, Y, and Z directions respectively:

[0084]

[0085]

[0086]

[0087] Among them, , , They are the X, Y, and Z axis coordinates of the measured points respectively; , , They are the point spacings in the X, Y, and Z axis directions respectively; , , They are respectively the number of point pairs with spacings of , , .

[0088] Fit the covariance model, estimate the correlation length and variance parameters, and establish the error model;

[0089] Generate a random simulation error field based on Kriging interpolation, and generate M sets of surface models containing simulated manufacturing errors.

[0090] Import the surface model containing manufacturing errors into Abaqus, define the material parameters, including density, Young's modulus, Poisson's ratio, etc., and set the material direction separately for composite materials;

[0091] Assemble the component instances to ensure effective contact between the models;

[0092] Use the translator to equivalently replace the screw connection, and create an analysis step for each screw tightening process;

[0093] Apply the load and boundary conditions, divide the mesh, and submit the job;

[0094] Use the Python interface provided by Abaqus for parametric modeling, modify the screw tightening sequence and related boundary conditions, and automatically solve cyclically to obtain the simulation deformation data of multiple sets of surfaces with different manufacturing errors under different process parameters, forming a simulation database;

[0095] Based on features such as contour corner points and screw holes, arrange the same number of meshes on the surface after assembly deformation as those on the ideal surface and the measured surface, and generate the assembly deformation error field with reference to the manufacturing error processing method of the measured surface;

[0096] Establish the node mapping relationship from the deformation error to the image space, normalize the deviation values of the measured surface and the assembled deformed surface from the ideal surface in the x, y, and z axis directions to [0, 255], map the deviations in the x, y, and z axis directions to the R, G, and B values respectively, and the conversion process is shown in Figure 3 as shown, and the converted RGB error image is as Figure 4 shown.

[0097]

[0098]

[0099]

[0100] The process of training the assembly deformation prediction model is shown in Figure 5 . The generator takes a random RGB error image at resolution and a process parameter vector as inputs, and outputs a predicted deformation field RGB image. The discriminator takes a real sample or a generated sample as input and outputs the probability that the sample is real;

[0101] Input real data and generated data, update the discriminator parameters to maximize the discrimination accuracy; deceive the discriminator with the generated data and update the generator parameters to minimize the discrimination probability of the discriminator for the generated samples;

[0102] Every time the discriminator is trained 5 times, the generator is trained 1 time. Alternately and iteratively train the generator and the discriminator to realize inputting a random manufacturing error image and a process parameter vector and outputting an assembly deformation image;

[0103] Perform multi-constraint optimization and solution for process parameters such as screw tightening sequence and tightening torque to ensure that the difference between the deformation after assembly and the ideal surface is minimized.

[0104] The technical process of the deformation prediction method for the conformal antenna assembly process disclosed in the above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination.

[0105] When implemented using hardware, the above embodiments can, in whole or in part, run the working logic and calculation process on an electronic device after being compiled by software. The electronic device includes a processor, a memory, a communication interface and a communication bus. The processor, the memory and the communication interface complete communication with each other through the communication bus. The memory is used to store at least one executable instruction, and the executable instruction causes the processor to execute the technical process disclosed in the above embodiments.

[0106] When implemented using software, the above embodiments may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. If the above method is implemented in the form of software function modules and sold or used as an independent product, it may also be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the embodiments of the present application, in essence or the part that contributes to the related art, may be embodied in the form of a software product. The software product is stored in a storage medium and includes several instructions for causing an electronic device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM), magnetic disks, or optical discs. Thus, the embodiments of the present application are not limited to any specific hardware, software, or firmware, or any combination among hardware, software, and firmware.

[0107] In addition, it should be understood that although this specification is described according to embodiments, not every embodiment only contains an independent technical solution. This narrative way of the specification is only for clarity. Those skilled in the art should regard the specification as a whole, and the technical solutions in the various embodiments may also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A deformation prediction method for a curved conformal antenna assembly process, characterized in that: The steps include: A workpiece surface library is established based on measured surface data and simulated surface data, and a finite element simulation analysis of the assembly process with different process parameters is performed on the workpiece surface library to obtain a deformation result after assembly; With surface manufacturing error and process parameters as input and deformation results after assembly as output, train and build a GAN-based assembly deformation prediction model; The real-time detected surface data is imported into the assembly deformation prediction model to predict the assembly deformation image under different process parameters, and the result with the minimum error between the assembly deformation image and the ideal surface model is used as the optimal assembly process parameter that meets the current conditions.

2. The method for predicting deformation during the assembly process of a curved conformal antenna according to claim 1, characterized in that: The measured surface data is obtained by: Point cloud data acquisition equipment is used to obtain several measured point cloud data of the assembly surface, and the measured point cloud data are filtered and spliced ​​to obtain the overall surface as the measured shape surface data.

3. The method for predicting deformation during the assembly process of a curved conformal antenna according to claim 1, characterized in that: The simulated surface data is obtained in the following manner: Several sets of measured surface data are collected and aligned with the ideal surface model to obtain the manufacturing error law of the surface, and simulated surface data are constructed based on the manufacturing error law.

4. The method for predicting deformation during the assembly process of a curved conformal antenna according to claim 3, characterized in that: Assume that the ideal surface mathematical model is , for the measured surface data and ideal surface mathematical model ICP registration is performed using the screw hole as a feature, the measured surface data is aligned with the ideal surface model and the coordinate system is transformed; The measured surface data The measured points in the ideal surface model Projection in the direction of the surface normal and calculation of the projection distance Generate an error field; where Represents measured point cloud data The measured points in Represents an ideal surface model Center and measured point The corresponding point of express Point normal direction vector; Ideal surface model Divide along the surface contour into The grid of the specification is used to calculate the mean error of all point clouds in each grid. , maximum value and standard deviation ; Repeat the above steps multiple times to obtain multiple sets of manufacturing error data.

5. The method for predicting deformation during the assembly process of a curved conformal antenna according to claim 4, characterized in that: Constructing simulated surface data based on manufacturing error rules, specifically including: The semivariogram function is used to calculate and analyze the spatial distribution characteristics of each group of manufacturing error data in the X, Y, and Z directions: ； ； ； In the formula, , , are the X, Y, and Z axis coordinates of the measured point respectively; , , They are the point spacing in the X, Y and Z axis directions respectively; , , The spacing is , , The point logarithm of ; Fit the covariance model, calculate the correlation length and variance parameters, and establish the error model; A random simulation error field is generated based on Kriging interpolation, and M groups of simulation surface data containing simulated manufacturing errors are generated.

6. The method for predicting deformation during the assembly process of a curved conformal antenna according to claim 1, characterized in that: With surface manufacturing error and process parameters as input and deformation results after assembly as output, a data set is constructed, and a GAN-based assembly deformation prediction model is built and trained, including: Based on the contour corner points and screw hole features, grids with the same number of nodes are arranged for the ideal surface model and the measured surface data, and the mapping relationship between the ideal surface model and the measured surface data is established using the nodes, where the node set of the ideal surface model is PI and the node set of the measured surface data is PR. Based on the ideal surface model, the predicted surface node coordinates X D , Y D , Z D and the actual surface node coordinates X I , Y I , Z I The coordinates of the contour nodes corresponding to the theoretical surface are X R , Y R , Z R The coordinate deviations are normalized and mapped to the red, green, and blue values ​​of the RGB color space to prepare input and output images for training the generative adversarial network. The dataset is divided into training set, validation set and test set. The generator inputs the surface manufacturing error image and loading conditions and outputs the predicted surface deformation result. The discriminator inputs the simulation deformation result or the deformation result predicted by the generator and outputs the discrimination result. The training is repeated alternately.

7. The method for predicting deformation during assembly of curved conformal antenna according to claim 1, characterized in that: Based on the ideal surface model, the actual surface node coordinates X I , Y I , Z I The coordinates of the contour nodes corresponding to the theoretical surface are X R , Y R , Z R The coordinate deviation is normalized to [0,255] and mapped to R, G, and B values ​​respectively to obtain the RGB error image: ； ； ； In the formula, , , are the values ​​of red, green and blue mapped to the i-th point respectively; , , are the coordinate deviations of the i-th point in the X, Y, and Z directions respectively; , , They are the minimum values ​​of coordinate deviation in the X, Y, and Z directions respectively; , , They are the maximum values ​​of coordinate deviation in the X, Y, and Z directions respectively.

8. The method for predicting deformation during the assembly process of a curved conformal antenna according to claim 1, characterized in that: Importing the real-time detected surface data into the assembly deformation prediction model to predict the assembly deformation image under different process parameters specifically includes: The surface data of the current curved conformal antenna is acquired in real time by using a point cloud data acquisition device, and is registered with an ideal curved surface model to obtain simulated surface data including manufacturing errors, and the simulated surface data is reduced to a two-dimensional image as one of the inputs of the assembly deformation prediction model; Taking minimization of the post-assembly deformation error as the optimization goal, the process parameter range is defined, the assembly deformation prediction model is called, the objective function value is calculated, the process parameters are updated, and the iterative optimization is repeated until the convergence conditions are met.

9. The method for predicting deformation during assembly of a curved conformal antenna according to claim 7, characterized in that: The assembly deformation prediction model includes a generator and a discriminator; the generator is based on The random RGB error image of resolution and the process parameter vector are input, and the predicted deformation field RGB image is output. The discriminator inputs the real sample or the generated sample, and outputs the probability that the sample is true. Input real data and generated data, update the discriminator parameters, and maximize the discrimination accuracy; deceive the discriminator by generating data, update the generator parameters, and minimize the discriminator's probability of discriminating the generated samples; After each predetermined number of training of the discriminator, the generator is trained once; the generator and the discriminator are trained iteratively alternately to input a random manufacturing error image and a process parameter vector and output an assembly deformation image; Multi-constraint optimization is performed on the screw tightening sequence and tightening torque to ensure that the gap between the assembly deformation image and the ideal surface model after assembly is minimized.

10. An electronic device, characterized in that: The electronic device comprises: a processor and a memory storing computer program instructions; when the processor executes the computer program instructions, the method for predicting deformation during the assembly process of a curved conformal antenna according to any one of claims 1 to 9 is implemented.

Images