Flexible positioning adjustment compensation amount optimization design method and system for aviation composite thin-wall curved surface structure

Abstract

The invention provides an aviation composite thin-wall curved surface structure flexible positioning adjustment compensation amount optimization design method and system, and the method comprises the steps: S1, obtaining the actual manufacturing and assembly state information of an aviation composite thin-wall curved surface part, and expressing the distribution state and geometric continuity of part errors; s2, establishing an assembly deviation calculation model considering the flexible positioning adjustment compensation amount, and predicting a curved surface appearance assembly precision value; s3, establishing a curved surface profile assembly precision prediction model; s4, acquiring an optimal positioning compensation amount set at a plurality of positioning execution tail ends on the tool; and S5, expressing and displaying the correlation and the distribution statistics rule among the positioning errors at the multi-positioning execution tail end of the tool. According to the method, the actual manufacturing / assembling state of the large-scale low-precision composite curved surface structural part is considered, an optimization method driven by force position actual measurement data is adopted, personalized calculation and active adaptive shape adjustment of the flexible assembling and adjusting compensation amount of the tool are achieved, and a foundation is laid for high-precision guarantee of assembling quality and low-stress assembling.

Description

Technical field

[0001] The present invention relates to the technical field of aviation structure assembly positioning quality control, and in particular to a method and system for optimizing the design of flexible positioning, assembly and compensation of aviation composite thin-walled curved surface structures. [Background Technology]

[0002] To mitigate the inconsistencies inherent in traditional analog assembly methods, flexible assembly tooling using digital data transfer has been proposed and implemented in production. By adjusting its structure, it can adapt to varying assembly environments and assemble products of varying shapes and sizes. Flexible tooling offers a degree of variability and versatility. As crucial for ensuring accurate aircraft assembly, the positioning accuracy of each tooling terminal directly determines assembly accuracy.

[0003] The positioning of thin-walled, curved aerospace composite structural parts generally requires the participation of multiple positioning execution terminals. Furthermore, in the actual assembly process, even if each positioning execution terminal is within the allowable positioning error range, the product's assembly accuracy cannot be minimized. This is because the final assembly accuracy of an aerospace product is directly related to two factors: the actual positioning accuracy of the tooling and the actual manufacturing or assembly accuracy of the product positioned by the tooling. In particular, thin-walled aerospace composite structural parts have low manufacturing accuracy (typically 8%-10% of the workpiece's thickness dimension) and are large in size. During the assembly process, they are susceptible to assembly deformation due to loads such as gravity, clamping force, and fastening connection force. Ensuring the assembly quality of thin-walled, curved structures is a major engineering challenge that needs to be addressed urgently.

[0004] During the assembly process, the sampling point errors at each positioning end-point of a flexible tooling fixture do not vary independently and randomly, but rather exhibit certain correlations and regularities, due to the influence of the continuous geometric deformation of product parts and the elastic coupling of adjacent contact points within the parts. Traditional methods use the error range of each positioning end-point to express positioning errors, assuming that the positioning errors at each end-point follow a uniform distribution pattern and are independent of each other within the tolerance band. This indifferent positioning accuracy design approach cannot effectively adapt to the continuity between the deformation / error state of the flexible parts being positioned and the actual surface. Furthermore, it can easily lead to geometric deviations and excessive internal stress in the complex assembly process of positioning, clamping, connection, removal, and rebound. Therefore, it is necessary to consider the positioning and adjustment compensation of the flexible assembly fixture as a factor affecting the final assembly accuracy based on the actual manufacturing error state of the parts to be assembled, based on the actual measurement data of the flexible parts. With the goal of minimizing assembly accuracy, an optimization design method driven by measured force and position data is used to optimize the actual positioning position of each positioning end-point. By obtaining the personalized positioning accuracy values ​​of each positioning execution end of the tooling, adaptive adjustment and active compensation control of the tooling during the assembly process can be achieved. Afterwards, a suitable three-dimensional mathematical model is used to express the correlation of positioning errors at each positioning execution end, and the statistical laws of error state distribution are obtained. This can lay the foundation for high-precision assembly quality assurance and low-stress assembly, and avoid geometric interference problems between parts during on-site assembly.

[0005] Therefore, it is necessary to study an optimization design method and system for the flexible positioning and adjustment compensation of aviation composite thin-walled curved surface structures to address the shortcomings of existing technologies and to solve or alleviate one or more of the above problems. [Summary of the invention]

[0006] In view of this, the present invention provides a method and system for optimizing the design of flexible positioning, assembly and compensation for thin-walled curved surface structures of aviation composite materials. First, digital measurement is used to obtain the actual manufacturing error state of the curved structural parts to be assembled, and mathematical processing is used to express the distribution state and correlation of the errors. Second, based on the actual error state of the workpiece and the load conditions during the assembly process, the positioning and adjustment compensation of the flexible assembly tool is used as a separate variable. The super-element stiffness theory is applied to establish a calculation model for the external assembly deviation of the curved structural parts and predict the assembly deviation value. Third, based on the assembly error theoretical calculation data and the flexible positioning compensation as input, a data-driven method based on a neural network is used to establish a prediction model for the external assembly accuracy of the real curved surface, achieving rapid calculation of assembly accuracy. Then, with the goal of minimizing the assembly accuracy of key feature points on the curved surface shape, an optimization function for flexible positioning and adjustment compensation is established. After using intelligent solution methods to obtain the optimal positioning compensation set at multiple positioning execution ends on the tool, the effectiveness of the solution is verified by field measurement and further feedback correction is made. Finally, the correlation and distribution statistics of the positioning errors at multiple positioning execution ends of the tool are expressed, and the distribution results are visualized, laying a foundation for high-precision assembly quality assurance and low-stress assembly.

[0007] In one aspect, the present invention provides a method for optimizing the design of flexible positioning and adjustment compensation for thin-walled curved surface structures of aviation composite materials. The method for calculating and evaluating the cumulative transfer of coordination errors in assembly of aviation thin-walled structures comprises the following steps:

[0008] S1: Measure and obtain actual manufacturing and assembly status data of thin-walled curved surface parts of aviation composite materials, using the mathematical methods of Legendre polynomials and generalized Fourier transform to express the actual distribution state and geometric continuity of part errors;

[0009] S2: Based on the actual distribution state and geometric continuity of the obtained part errors, combined with the positioning load conditions during the assembly process, an assembly deviation calculation model is established that integrates the flexible positioning and adjustment compensation amounts at multiple positioning execution ends to obtain the surface shape assembly accuracy value;

[0010] S3: Based on the surface shape assembly accuracy values ​​obtained by the assembly deviation calculation model and the measured assembly status data information, and with the flexible positioning and adjustment compensation values ​​at multiple positioning execution terminals as input, a data-driven method based on a neural network is used to establish a data-driven surface shape assembly accuracy prediction model;

[0011] S4: With the goal of minimizing the prediction results of the data-driven surface shape assembly accuracy prediction model, an optimization objective function is established and solved using an optimization algorithm. The optimal positioning and adjustment compensation value set at multiple positioning execution ends on the tooling is obtained. The effectiveness of the solution is then compared and analyzed using field measurements. After that, feedback is provided to further correct the optimal positioning and adjustment compensation value set.

[0012] S5: A geometric covariance model is constructed by feedback-correcting the optimal positioning and adjustment compensation value set to express the correlation and distribution statistical law between the positioning errors at the end of the multi-positioning execution of the tooling, and the distribution results are visualized.

[0013] According to the above aspects and any possible implementation, an implementation is further provided, wherein S1 specifically includes:

[0014] S11: Considering the low manufacturing precision and large size of thin-walled composite structural parts in aviation, a laser tracker and laser scanner are combined to measure key feature points on product parts that reflect assembly quality characteristics, as well as the deviation status of positioning measurement points that match the positioning execution end points of the assembly tooling.

[0015] S12: Based on the deviation states of the key feature points and the positioning measurement points that match the positioning execution ends of the assembly tooling, a geometric covariance model of the surface part on a two-dimensional plane is established, and the surface sampling point deviations are fitted using Legendre polynomials;

[0016] S13: Use the least squares method to solve the coefficients of the Legendre polynomials, perform a generalized Fourier transform on the Legendre polynomial series used for fitting, and express the relationship between the errors of different key feature points;

[0017] S14: Based on the geometric covariance model on the two-dimensional level, a geometric covariance model of the surface shape error of the curved workpiece on the three-dimensional level is established to accurately express the distribution state and geometric continuity of the part error.

[0018] According to the above aspects and any possible implementation, an implementation is further provided, wherein S2 specifically includes:

[0019] S21: In the positioning stage of thin-walled curved surface structural parts, an M-2-1 positioning layout method is adopted, where M is the number of positioning points, and the positioning points are divided into deterministic positioning points and over-constrained positioning points, where M>3 and M is a positive integer;

[0020] S22: Based on the actual distribution state and geometric continuity of the obtained part errors and the geometric covariance model of the curved surface parts on the 3D level, the principles of rigid body kinematics are applied to analyze the assembly position and posture changes of thin-walled curved surface structural parts after the positioning stage, and the changes in key feature points on the shape are obtained;

[0021] S23: Based on the changes in assembly position and posture, the positioning compensation at the end of the multi-positioning execution is introduced as the input variable for the tool positioning at the deterministic positioning point and the over-constrained positioning point. Then, based on the positioning load conditions during the assembly process, a clamping force or adsorption force is applied at the positioning point to securely connect the assembly tool and the part.

[0022] S24: Under the action of assembly clamping force or adsorption force, the super-element stiffness matrix theory is applied. On the basis of introducing positioning compensation, a calculation model for the assembly deviation of curved structural parts considering the positioning compensation of flexible tooling is established. The deviation change of key feature points on the workpiece along the workpiece normal direction is calculated, laying the foundation for subsequent optimization of positioning compensation.

[0023] S25: In the subsequent fastening connections of different workpieces and the removal of the assembly, the super-element stiffness matrix theory is applied to obtain the stiffness matrix of the entire assembly. By calculating the deviation changes of key feature points on the workpiece along the normal direction of the workpiece, the numerical value of the assembly accuracy at the key external shape measuring points of the entire assembly under the action of the clamping force is obtained.

[0024] According to the above aspects and any possible implementation, an implementation is further provided, wherein S3 specifically includes:

[0025] S31: Perform homogenization treatment on the composite thin-walled curved surface structure to obtain the overall physical performance parameters and material mechanical constants of the composite structure;

[0026] S32: Based on the overall physical performance parameters and material mechanical constants of the composite structure, within the allowable range of positioning accuracy, the flexible positioning and adjustment compensation parameters at multiple positioning execution terminals are changed to obtain a set of assembly accuracy values ​​at key shape measurement points of the entire assembly under a small sample, and the optimal Latin hypercube sampling method is used for data sampling;

[0027] S33: Using the sampling data obtained using the optimal Latin hypercube sampling method as training data, a mathematical modeling method based on a Bayesian neural network is applied to fit a proxy model, and the proxy model is used as a prediction model with the flexible positioning and adjustment compensation amounts at multiple positioning execution terminals as input and the shape assembly accuracy of the actual surface as output;

[0028] S34: Statistical indicators are used to perform computational verification on the output prediction model. By reducing the deviation between the trained Bayesian neural network prediction model and the true model, the mapping relationship between the compensation amount of the adjustment system and the assembly deviation is accurately represented.

[0029] According to the above aspects and any possible implementation, an implementation is further provided, wherein S4 specifically includes:

[0030] S41: By mapping the flexible positioning and adjustment compensation amounts at multiple positioning execution terminals to the assembly deviations, the assembly deviation values ​​of the key feature points on the curved surface are coordinated and balanced. While ensuring that the assembly deviation values ​​at the key feature points meet the requirements, the shape assembly deviation is minimized. The problem of solving the flexible positioning and adjustment compensation amounts at multiple positioning execution terminals is converted into a multi-objective optimization problem.

[0031] S42: Establish a multi-objective optimization model using the positioning and adjustment compensation amounts at multiple positioning execution terminals on the flexible tooling as the variables to be optimized, minimizing the shape assembly deviation output of each key point as the optimization goal, and using the positioning accuracy range of the assembly tooling and the shape assembly deviation requirements specified in the assembly process document as constraints;

[0032] S43: The NSGA-II algorithm is used to analyze and solve the optimal positioning compensation. Using the Matlab software platform, the optimal solution that minimizes the assembly deviation of each key feature point on the surface shape is obtained, and the optimal positioning and adjustment compensation at multiple positioning execution ends is obtained.

[0033] S44: Inputting the optimal positioning and adjustment compensation values ​​at the multiple positioning execution terminals into the surface shape assembly accuracy prediction model to quickly calculate the assembly deviation value of each key feature point on the surface shape under the positioning compensation solution;

[0034] S45: Use the actual measurement method in S11 to obtain the actual assembly deviation value of each key feature point on the surface shape, conduct comparative analysis and verification on the effectiveness of the optimal positioning and adjustment compensation amount solved in S43, and judge whether the actual assembly deviation value meets the accuracy requirements specified in the assembly process document. If not, adjust the calculation parameters in the Bayesian neural network prediction model, recalculate the optimal positioning compensation amount at multiple positioning execution ends until the assembly requirements are met, and obtain the optimal positioning and adjustment compensation amount set corrected by feedback.

[0035] According to the above aspects and any possible implementation, an implementation is further provided, wherein S5 specifically includes:

[0036] S51: Based on the optimal positioning compensation amount corrected by feedback while meeting the assembly accuracy requirements, statistical methods are used to derive the variance of the positioning accuracy values ​​of each end according to the tolerance band of the positioning accuracy of each end, and the variance is expressed as a diagonal matrix;

[0037] S52: Based on the measurement data of the sampling points of the positioning accuracy of each positioning terminal, the surface sampling point deviation is fitted using Legendre polynomials, and the Legendre polynomial series used for fitting is generalized Fourier transformed to construct a transformation matrix between different key feature points to represent the geometric correlation between the positioning accuracy sampling points of each positioning execution terminal;

[0038] S53: Based on the variance value and transformation matrix, a geometric covariance model is constructed to express the correlation and spatial distribution statistical laws between the positioning errors at the multiple positioning execution ends of the tooling, and the distribution results are used to fit the three-dimensional spatial surface to intuitively and visually display the actual error distribution of the positioning positions at multiple positioning execution ends.

[0039] According to the above aspects and any possible implementation, an implementation is further provided, wherein the method of fitting the surface sampling point deviation using Legendre polynomials includes:

[0040] On the surface of the part, the arrangement direction of the key feature points in the U direction is selected as the main assembly direction, and the V direction perpendicular to the main direction is selected as the secondary assembly direction;

[0041] The sampling point deviations on the same sampling line in the main and secondary directions are respectively used as the basis for two-dimensional curve fitting, and geometric covariance models on two-dimensional planes are respectively established in the main and secondary directions.

[0042] According to the above aspects and any possible implementation, an implementation is further provided, wherein the surface sampling point deviations in the main direction or the secondary direction are fitted using Legendre polynomials, and a generalized Fourier transform is performed on the Legendre polynomial series used for fitting, including:

[0043] The surface sampling point deviation is expressed using Legendre polynomial series of five orders from 0 to 4.

[0044] r(x i ) * =a s0 F s0 (x i )+a s1 F s1 (x i )+a s2 F s2 (x i )+a s3 F s3 (x i )+a s4F s4 (x i )

[0045]

[0046] Where: N is the number of sampling points on the sampling line of the main direction of the assembly, l is the order of the Legendre polynomial, a sl is the coefficient of the l-th order Legendre polynomial on the s-th sampling line in the main direction, which is solved by the least squares fitting method; Pl(x i ),Pl(x j ) are the l-order Legendre polynomial functions at x i 、x j The value at r is the value of x on the sth sampling line in the main direction j The deviation value of the sampling point at , R is the l-order Fourier Legendre polynomial function on the s-th sampling line at x i The value at ; r* is the fitting value of the jth sampling point on the sth sampling line in the main direction of the assembly;

[0047] Use the least squares method to solve a sl Coefficients, for approximating the profile deviation curve with the fitting polynomial curve;

[0048] Perform a generalized Fourier transform on the Legendre polynomial series and construct the transformation matrix S, where s rl is the coefficient of the l-th order Legendre polynomial on the r-th sampling line in the main direction, expressing the relationship between different key feature points.

[0049] According to the above aspects and any possible implementation, an implementation is further provided, wherein a geometric covariance model of the surface shape error of the curved workpiece on a three-dimensional level is established based on the two-dimensional geometric covariance model, including:

[0050] Arrange the sampling points in order of the main and secondary directions;

[0051] Applying the rounding and remaindering methods to convert the serial number of the sampling point into the coordinates of the sampling point within the sampling area;

[0052] Express the relationship between the deviation of the I-th sampling point with coordinate value (i, j) and the deviation of the J-th sampling point with coordinate value (i0, j0);

[0053] Using statistical methods, the variance of the accuracy values ​​of each key sampling point on the part is derived based on the tolerance band of each sampling point and expressed as a diagonal matrix Λ.

[0054] According to the above aspects and any possible implementation, a system for optimizing the design of flexible positioning, adjustment and compensation for thin-walled curved surface structures of aviation composite materials is further provided. The system for optimizing the design of flexible positioning, adjustment and compensation for thin-walled curved surface structures of aviation composite materials comprises:

[0055] Part error distribution state modeling module, used to obtain the actual manufacturing and assembly state information of thin-walled curved surface parts of aviation composite materials, and use mathematical methods to express the distribution state and geometric continuity of part errors;

[0056] The module for calculating the theoretical calculation of the shape assembly accuracy is used to obtain the actual error distribution state through the distribution state of the part error and the geometric continuity. Based on the actual error distribution state of the part, an assembly deviation calculation model is established that takes into account the flexible positioning and adjustment compensation amount to predict the surface shape assembly accuracy value.

[0057] The shape assembly accuracy data-driven calculation module is used to predict the surface shape assembly accuracy value based on the assembly deviation calculation model. It uses a data-driven method based on a neural network to establish a surface shape assembly accuracy prediction model.

[0058] The tooling positioning error distribution state modeling module is used for the flexible positioning and adjustment compensation optimization module, which is used to minimize the predicted results of the surface shape assembly accuracy prediction model, apply the optimization algorithm to solve, and obtain the optimal positioning and adjustment compensation value set at multiple positioning execution ends on the tooling.

[0059] Compared with the prior art, the present invention can achieve the following technical effects:

[0060] 1) Establish a computational model for the assembly deviation transmission mechanism that integrates the actual force-position-measured state of the assembly process. Specifically, a distribution model of the actual manufacturing error state of the curved surface structural parts to be assembled is established. In combination with the load conditions during the assembly process, the positioning and adjustment compensation of the flexible assembly tooling is used as a separate variable to achieve accurate prediction and computational modeling of the external assembly deviation of the curved surface structural parts.

[0061] 2) Establish a data-driven real surface shape assembly accuracy prediction model. Based on the assembly error theory calculation data and the flexible positioning compensation amount as input, a neural network-based data-driven method is used to establish a real surface shape assembly accuracy prediction model to achieve rapid calculation of assembly accuracy.

[0062] 3) Establish an intelligent optimization solution for flexible positioning and adjustment compensation and a correction mechanism based on measured data feedback. Aiming to minimize the assembly accuracy of key feature points on the curved surface, an optimization function for flexible positioning and adjustment compensation is established. Intelligent solution methods are used to obtain the optimal positioning compensation set at multiple positioning execution terminals on the tooling. Field measurements are then used to verify the effectiveness of the solution and provide further feedback and corrections.

[0063] 4) The correlation between the key feature points of the curved workpiece and the error distribution at each positioning execution end on the tooling was established, and a mathematical expression method for the error distribution state was proposed, laying the front-end analysis foundation for achieving high-quality, precise assembly and low-stress assembly of high-end products.

[0064] Of course, any product implementing the present invention does not necessarily need to achieve all of the above-mentioned technical effects at the same time.

Brief Description of the Drawings

[0065] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0066] Figure 1 This is a technical step diagram for optimizing the design of flexible positioning, assembly and compensation for thin-walled curved surface structures of aviation composite materials according to the present invention;

[0067] Figure 2 This is a description of the structure and adjustable functions of the typical box segment assembly tooling of the present invention;

[0068] Figure 3 This is a diagram of the overall assembly structure of a certain type of composite material wing box provided by one embodiment of the present invention. [Specific implementation method]

[0069] In order to better understand the technical solution of the present invention, the embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0070] It should be understood that the embodiments described are only a portion of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative work are within the scope of protection of the present invention.

[0071] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The singular forms "a", "an", "the" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise.

[0072] The present invention provides a method for optimizing the design of the flexible positioning, adjustment and compensation amount of an aviation composite thin-walled curved surface structure. The method comprises:

[0073] S1: Measure and obtain actual manufacturing and assembly status data of thin-walled curved surface parts of aviation composite materials, using the mathematical methods of Legendre polynomials and generalized Fourier transform to express the actual distribution state and geometric continuity of part errors;

[0074] S2: Based on the actual distribution state and geometric continuity of the obtained part errors, combined with the positioning load conditions during the assembly process, an assembly deviation calculation model is established that integrates the flexible positioning and adjustment compensation amounts at multiple positioning execution ends to obtain the surface shape assembly accuracy value;

[0075] S3: Based on the surface shape assembly accuracy values ​​obtained by the assembly deviation calculation model and the measured assembly status data information, and with the flexible positioning and adjustment compensation values ​​at multiple positioning execution terminals as input, a data-driven method based on a neural network is used to establish a data-driven surface shape assembly accuracy prediction model;

[0076] S4: With the goal of minimizing the prediction results of the data-driven surface shape assembly accuracy prediction model, an optimization objective function is established and solved using an optimization algorithm. The optimal positioning and adjustment compensation value set at multiple positioning execution ends on the tooling is obtained. The effectiveness of the solution is then compared and analyzed using field measurements. After that, feedback is provided to further correct the optimal positioning and adjustment compensation value set.

[0077] S5: A geometric covariance model is constructed by feedback-correcting the optimal positioning and adjustment compensation value set to express the correlation and distribution statistical law between the positioning errors at the end of the multi-positioning execution of the tooling, and the distribution results are visualized.

[0078] Furthermore, the S1 specifically includes:

[0079] S11: Considering the low manufacturing precision and large size of thin-walled composite structural parts in aviation, a laser tracker and laser scanner are combined to measure key feature points on product parts that reflect assembly quality characteristics, as well as the deviation status of positioning measurement points that match the positioning execution end points of the assembly tooling.

[0080] S12: Based on the deviation states of the key feature points and the positioning measurement points that match the positioning execution ends of the assembly tooling, a geometric covariance model of the surface part on a two-dimensional plane is established, and the surface sampling point deviations are fitted using Legendre polynomials;

[0081] S13: Use the least squares method to solve the coefficients of the Legendre polynomials, perform a generalized Fourier transform on the Legendre polynomial series used for fitting, and express the relationship between the errors of different key feature points;

[0082] S14: Based on the geometric covariance model on the two-dimensional level, a geometric covariance model of the surface shape error of the curved workpiece on the three-dimensional level is established to accurately express the distribution state and geometric continuity of the part error.

[0083] Furthermore, the S2 specifically includes:

[0084] S21: In the positioning stage of thin-walled curved surface structural parts, an M-2-1 positioning layout method is adopted, where M is the number of positioning points, and the positioning points are divided into deterministic positioning points and over-constrained positioning points, where M>3 and M is a positive integer;

[0085] S22: Based on the actual distribution state and geometric continuity of the obtained part errors and the geometric covariance model of the curved surface parts on the 3D level, the principles of rigid body kinematics are applied to analyze the assembly position and posture changes of thin-walled curved surface structural parts after the positioning stage, and the changes in key feature points on the shape are obtained;

[0086] S23: Based on the changes in assembly position and posture, the positioning compensation at the end of the multi-positioning execution is introduced as the input variable for the tool positioning at the deterministic positioning point and the over-constrained positioning point. Then, based on the positioning load conditions during the assembly process, a clamping force or adsorption force is applied at the positioning point to securely connect the assembly tool and the part.

[0087] S24: Under the action of assembly clamping force or adsorption force, the super-element stiffness matrix theory is applied. On the basis of introducing positioning compensation, a calculation model for the assembly deviation of curved structural parts considering the positioning compensation of flexible tooling is established. The deviation change of key feature points on the workpiece along the workpiece normal direction is calculated, laying the foundation for subsequent optimization of positioning compensation.

[0088] S25: In the subsequent fastening connections of different workpieces and the removal of the assembly, the super-element stiffness matrix theory is applied to obtain the stiffness matrix of the entire assembly. By calculating the deviation changes of key feature points on the workpiece along the normal direction of the workpiece, the numerical value of the assembly accuracy at the key external shape measuring points of the entire assembly under the action of the clamping force is obtained.

[0089] Furthermore, the S3 specifically includes:

[0090] S31: Perform homogenization treatment on the composite thin-walled curved surface structure to obtain the overall physical performance parameters and material mechanical constants of the composite structure;

[0091] S32: Based on the overall physical performance parameters and material mechanical constants of the composite structure, within the allowable range of positioning accuracy, the flexible positioning and adjustment compensation parameters at multiple positioning execution terminals are changed to obtain a set of assembly accuracy values ​​at key shape measurement points of the entire assembly under a small sample, and the optimal Latin hypercube sampling method is used for data sampling;

[0092] S33: Using the sampling data obtained using the optimal Latin hypercube sampling method as training data, a mathematical modeling method based on a Bayesian neural network is applied to fit a proxy model, and the proxy model is used as a prediction model with the flexible positioning and adjustment compensation amounts at multiple positioning execution terminals as input and the shape assembly accuracy of the actual surface as output;

[0093] S34: Statistical indicators are used to perform computational verification on the output prediction model. By reducing the deviation between the trained Bayesian neural network prediction model and the true model, the mapping relationship between the compensation amount of the adjustment system and the assembly deviation is accurately represented.

[0094] Furthermore, the S4 specifically includes:

[0095] S41: By mapping the flexible positioning and adjustment compensation amounts at multiple positioning execution terminals to the assembly deviations, the assembly deviation values ​​of the key feature points on the curved surface are coordinated and balanced. While ensuring that the assembly deviation values ​​at the key feature points meet the requirements, the shape assembly deviation is minimized. The problem of solving the flexible positioning and adjustment compensation amounts at multiple positioning execution terminals is converted into a multi-objective optimization problem.

[0096] S42: Establish a multi-objective optimization model using the positioning and adjustment compensation amounts at multiple positioning execution terminals on the flexible tooling as the variables to be optimized, minimizing the shape assembly deviation output of each key point as the optimization goal, and using the positioning accuracy range of the assembly tooling and the shape assembly deviation requirements specified in the assembly process document as constraints;

[0097] S43: The NSGA-II algorithm is used to analyze and solve the optimal positioning compensation. Using the Matlab software platform, the optimal solution that minimizes the assembly deviation of each key feature point on the surface shape is obtained, and the optimal positioning and adjustment compensation at multiple positioning execution ends is obtained.

[0098] S44: Inputting the optimal positioning and adjustment compensation values ​​at the multiple positioning execution terminals into the surface shape assembly accuracy prediction model to quickly calculate the assembly deviation value of each key feature point on the surface shape under the positioning compensation solution;

[0099] S45: Use the actual measurement method in S11 to obtain the actual assembly deviation value of each key feature point on the surface shape, conduct comparative analysis and verification on the effectiveness of the optimal positioning and adjustment compensation amount solved in S43, and judge whether the actual assembly deviation value meets the accuracy requirements specified in the assembly process document. If not, adjust the calculation parameters in the Bayesian neural network prediction model, recalculate the optimal positioning compensation amount at multiple positioning execution ends until the assembly requirements are met, and obtain the optimal positioning and adjustment compensation amount set corrected by feedback.

[0100] Furthermore, the S5 specifically includes:

[0101] S51: Based on the optimal positioning compensation amount corrected by feedback while meeting the assembly accuracy requirements, statistical methods are used to derive the variance of the positioning accuracy values ​​of each end according to the tolerance band of the positioning accuracy of each end, and the variance is expressed as a diagonal matrix;

[0102] S52: Based on the measurement data of the sampling points of the positioning accuracy of each positioning terminal, the surface sampling point deviation is fitted using Legendre polynomials, and the Legendre polynomial series used for fitting is generalized Fourier transformed to construct a transformation matrix between different key feature points to represent the geometric correlation between the positioning accuracy sampling points of each positioning execution terminal;

[0103] S53: Based on the variance value and transformation matrix, a geometric covariance model is constructed to express the correlation and spatial distribution statistical laws between the positioning errors at the multiple positioning execution ends of the tooling, and the distribution results are used to fit the three-dimensional spatial surface to intuitively and visually display the actual error distribution of the positioning positions at multiple positioning execution ends.

[0104] The present invention also provides a system for optimizing the design of flexible positioning, adjustment and compensation for thin-walled curved surface structures of aviation composite materials. The system comprises:

[0105] Part error distribution state modeling module, used to obtain the actual manufacturing and assembly state information of thin-walled curved surface parts of aviation composite materials, and use mathematical methods to express the distribution state and geometric continuity of part errors;

[0106] The module for calculating the theoretical calculation of the shape assembly accuracy is used to obtain the actual error distribution state through the distribution state of the part error and the geometric continuity. Based on the actual error distribution state of the part, an assembly deviation calculation model is established that takes into account the flexible positioning and adjustment compensation amount to predict the surface shape assembly accuracy value.

[0107] The shape assembly accuracy data-driven calculation module is used to predict the surface shape assembly accuracy value based on the assembly deviation calculation model. It uses a data-driven method based on a neural network to establish a surface shape assembly accuracy prediction model.

[0108] The tooling positioning error distribution state modeling module is used for the flexible positioning and adjustment compensation optimization module, which is used to minimize the predicted results of the surface shape assembly accuracy prediction model, apply the optimization algorithm to solve, and obtain the optimal positioning and adjustment compensation value set at multiple positioning execution ends on the tooling.

[0109] Example 1:

[0110] An optimization design method for the flexible positioning and adjustment compensation of thin-walled curved surface structures of aviation composite materials, such as Figure 1 As shown, the method for optimizing the design of flexible positioning and adjustment compensation for thin-walled curved surface structures of aviation composite materials includes the following steps:

[0111] S1: Measure and obtain actual manufacturing and assembly status data of thin-walled curved surface parts of aviation composite materials, using the mathematical methods of Legendre polynomials and generalized Fourier transform to express the actual distribution state and geometric continuity of part errors;

[0112] S2: Based on the actual distribution state and geometric continuity of the obtained part errors, combined with the positioning load conditions during the assembly process, an assembly deviation calculation model is established that integrates the flexible positioning and adjustment compensation amounts at multiple positioning execution ends to obtain the surface shape assembly accuracy value;

[0113] S3: Based on the surface shape assembly accuracy values ​​obtained by the assembly deviation calculation model and the measured assembly status data information, and with the flexible positioning and adjustment compensation values ​​at multiple positioning execution terminals as input, a data-driven method based on a neural network is used to establish a data-driven surface shape assembly accuracy prediction model;

[0114] S4: With the goal of minimizing the prediction results of the data-driven surface shape assembly accuracy prediction model, an optimization objective function is established and solved using an optimization algorithm. The optimal positioning and adjustment compensation value set at multiple positioning execution ends on the tooling is obtained. The effectiveness of the solution is then compared and analyzed using field measurements. After that, feedback is provided to further correct the optimal positioning and adjustment compensation value set.

[0115] S5: A geometric covariance model is constructed by feedback-correcting the optimal positioning and adjustment compensation value set to express the correlation and distribution statistical law between the positioning errors at the end of the multi-positioning execution of the tooling, and the distribution results are visualized.

[0116] Said S1 specifically includes:

[0117] S11: Considering the low manufacturing precision and large size of thin-walled structural parts made of aviation composite materials, a laser tracker and a laser scanner are used to measure the m key feature points on the thin-walled skin part P8 that can reflect the assembly quality characteristics, as well as the m key feature points related to the positioning execution ends of the assembly tooling ( Figure 2 Deviation status of the positioning measurement points coordinated with the 6 clamping units in the

[0118] S12: Based on the deviation states of the key feature points and the positioning measurement points that match the positioning execution ends of the assembly tooling, a geometric covariance model of the surface part on a two-dimensional plane is established, and the surface sampling point deviations are fitted using Legendre polynomials;

[0119] S13: Use the least squares method to solve the coefficients of Legendre polynomials of various orders, perform generalized Fourier transform on the Legendre polynomial series used for fitting, and express the relationship between different key feature points;

[0120] S14: Based on the two-dimensional geometric covariance model, a three-dimensional geometric covariance model of the surface shape error of the curved workpiece is established to accurately express the distribution state and geometric continuity of the part error.

[0121] When establishing a two-dimensional geometric covariance model for a curved surface part in S12, the arrangement direction of key feature points in the U direction on the surface of the part is selected as the primary assembly direction, and the V direction perpendicular to the primary direction is selected as the secondary assembly direction. Deviations of sampling points on the same sampling line in the primary and secondary directions are used as the basis for two-dimensional curve fitting, and two-dimensional geometric covariance models are established in each of the primary and secondary directions.

[0122] The surface sampling point deviations fitted by Legendre polynomials in S12 are expressed using Legendre polynomials of five orders from 0 to 4, specifically:

[0123] r(x i ) * =a s0 F s0 (x i )+a s1 F s1 (x i )+a s2 F s2 (x i )+a s3 F s3 (x i )+a s4 F s4 (x i )

[0124]

[0125] In the above formula, N is the number of sampling points on the sampling line of the main direction of the assembly, l is the order of the Legendre polynomial, a sl is the coefficient of the l-th order Legendre polynomial on the s-th sampling line in the main direction, which is solved by the least squares fitting method; Pl(x i ),Pl(x j ) are the l-order Legendre polynomial functions at x i 、x j The value at r is the value of x on the sth sampling line in the main direction j The deviation value of the sampling point at , R is the l-order Fourier Legendre polynomial function on the s-th sampling line at x iThe value at ; r* is the fitting value of the jth sampling point on the sth sampling line in the main direction of the assembly;

[0126] In S13, the least square method is used to solve the coefficients s of the Fourier Legendre polynomials of various orders. rl :

[0127]

[0128] The above equation is the objective function of the least squares method. The smaller its value, the closer the fitted polynomial curve is to the profile deviation curve. When the objective function takes its minimum value, the set of coefficients obtained is the required Legendre polynomial coefficients.

[0129] In S13, after the generalized Fourier transform, the position deviation of the main direction sampling point and the position deviation of other sampling points on the sampling line in the main direction establish a mutual relationship. The transformation matrix S can be expressed as:

[0130]

[0131] Where s rl are the coefficients of the Legendre polynomial of order l on the r-th sampling line in the main direction.

[0132] Furthermore, the modeling method for the correlation between the deviations of the surface sampling points in the secondary assembly direction of the workpiece is the same as that in the primary assembly direction.

[0133] When establishing the geometric covariance model of the surface deviation of the flexible part on the three-dimensional plane in S14, the sampling points are arranged in order according to the sampling order of the main and secondary directions, and the serial numbers of the sampling points are converted into the coordinates of the sampling points in the sampling area by rounding and taking the remainder.

[0134] Furthermore, the relationship between the deviation of the I-th sampling point with coordinate value (i, j) and the deviation of the J-th sampling point with coordinate value (i0, j0) can be expressed as:

[0135]

[0136] Furthermore, statistical methods are used to derive the variance of the precision values ​​of each key sampling point on the part based on the tolerance band of each sampling point, and the variance is represented by a diagonal matrix Λ.

[0137] Furthermore, the geometric covariance model of the surface shape error of the curved workpiece on the three-dimensional level in S14 is expressed as:

[0138] ∑=S IJ ·Λ·S IJ T

[0139] The S2 specifically includes:

[0140] S21: During the positioning phase of the thin-walled skin part P8, a 6-2-1 positioning layout was adopted, dividing the positioning points into two types: deterministic positioning points and over-constrained positioning points. The number of deterministic positioning points was 3, and the number of over-constrained positioning points was also 3.

[0141] S22: Applying the principles of rigid body kinematics, by establishing a geometric relationship between the assembly tool positioning point deviation δR and the positioning deviation δq'0 of the key feature points of the part, the changes in the assembly position and posture of the thin-walled curved surface structural part after the positioning process are analyzed, and the changes in the nine key feature points on the shape are obtained;

[0142] S23: Based on the geometric position and posture variation errors of the assembly, a positioning compensation δR' is introduced as an input variable for the tool positioning at the deterministic positioning points and the over-constrained positioning points. After that, a clamping force or adsorption force F is applied at all the positioning points to securely connect the assembly tool and the part together.

[0143] S24: Under the action of assembly clamping force or adsorption force, the super-element stiffness matrix theory is applied to establish a calculation model for the assembly deviation of curved structural parts taking into account the positioning compensation of flexible tooling. The deviation change V of key feature points on the workpiece along the workpiece normal direction is calculated, laying the foundation for subsequent optimization of positioning compensation.

[0144] S25: In the subsequent fastening connection of different workpieces and assembly removal, the super element stiffness matrix theory is applied to obtain the stiffness matrix of the entire assembly, and the numerical value of the assembly accuracy at the key shape measurement points of the entire assembly under the reverse action of the clamping force is calculated.

[0145] The nine key feature points of the workpiece in S22 are evenly distributed on the surface of the workpiece, that is, there are three measuring points in the U direction and three measuring points in the V direction, and the three measuring points divide the workpiece into four equal parts.

[0146] The force and deformation calculation model of the curved surface structural component introduced with the flexible fixture positioning compensation in S24 is:

[0147] F=Kg[δR+δR'V]

[0148] Where K is the super element stiffness matrix of the part or assembly extracted with the help of finite element simulation analysis software. During the extraction process, the workpiece positioning points that match the various positioning models of the tooling are used as key nodes.

[0149] The S3 specifically includes:

[0150] S31: Homogenize thin-walled composite curved surface components. Based on the continuum equilibrium equation, displacement boundary conditions, stress-strain constitutive equation, and geometric equations, the mathematical relationship between strain and stress is expressed in combination with the derivative rule. The equivalent stiffness matrix of the homogenized material is derived. The inverse of this matrix yields the equivalent macroscopic material parameters, namely the elastic modulus E in the fiber direction, the direction perpendicular to the fiber, and the ply thickness direction, Poisson's ratio v, and shear modulus G. Based on the obtained overall physical performance parameters of the composite structure and the material mechanical constants, the assembly accuracy values ​​at key shape measurement points of the entire assembly are calculated, allowing for rapid and accurate shape assembly accuracy values.

[0151] S32: Based on the overall physical performance parameters and material mechanical constants of the composite structure, combined with changes in the assembly process input parameters, a small sample size prediction data set is obtained. The optimal Latin hypercube sampling method is then used to optimize the sampling method using the maximum and minimum distance criteria to ensure a more uniform distribution of the sample points within the sampling space, achieving a good data sampling effect using the smallest possible data sample size.

[0152] S33: Using the sampled data as training data, a mathematical modeling method based on a Bayesian neural network is applied to fit a proxy model, and a prediction model is established with the flexible positioning and adjustment compensation amount as input and the actual surface shape assembly accuracy as output;

[0153] S34: The correlation coefficient R and mean square error MSE indicators in statistics are used to calculate and verify the accuracy of the Bayesian neural network prediction model, reduce the deviation between the trained Bayesian neural network prediction model and the true model, and accurately represent the mapping relationship between the compensation amount of the assembly system and the assembly deviation.

[0154] In the process of generating the training data set for the proxy model in S32, the optimal Latin hypercube test method is used to fill the sample space of the flexible tooling positioning compensation amount, and 800 sample data are generated within the flexible adjustable range of the assembly tooling. Figure 2 There are 6 tool positioning ends on the assembly tool, so a positioning compensation design matrix of 800*6 can be generated. Each row in the matrix represents a positioning accuracy solution for a set of flexible tooling.

[0155] Furthermore, after obtaining the design matrix of the flexible tooling positioning accuracy scheme, the design matrix is ​​used as input and substituted into the assembly deviation mechanical model established in S2 to obtain the shape assembly deviation value on the skin part. This sample data can then be used to train the assembly deviation prediction model of key feature points.

[0156] The parameter training process of the mathematical modeling method based on the Bayesian neural network in S33 specifically includes the following steps: (1) defining a functional model; (2) defining a random model with weights as random variables; (3) obtaining the posterior distribution of parameters; and (4) calculating the margin to quantify the uncertainty of the model.

[0157] The calculation of the correlation coefficient R of the Bayesian neural network model in S34 is:

[0158]

[0159] The calculation of the mean square error R of the Bayesian neural network model in S34 is:

[0160]

[0161] Where y i is the actual model response value; is the approximate model response value; is the actual model response mean.

[0162] The S4 specifically includes:

[0163] S41: By mapping the compensation amount of the adjustment system to the assembly deviation, the assembly deviation values ​​of key feature points on the surface shape are coordinated and balanced. While ensuring that the assembly deviation values ​​of each point meet the requirements, the shape assembly deviation is minimized as much as possible, and the problem of solving the optimal positioning compensation amount is transformed into a multi-objective optimization problem.

[0164] S42: Taking the positioning and adjustment compensation amounts at the six positioning execution ends on the flexible tooling as the variables to be optimized, minimizing the shape assembly deviation output of each key point as the optimization goal, and the positioning accuracy range of the assembly tooling ±0.10mm and the shape assembly deviation requirements (-0.6mm, +0.8mm) specified in the assembly process document as constraints, a multi-objective optimization model for the adjustment compensation amounts at the multiple positioning execution ends of the assembly tooling is established;

[0165] S43: The NSGA-II algorithm is used to analyze the optimal positioning compensation of flexible tooling. Using the Matlab software platform, the optimal solution is obtained to minimize the overall assembly deviation of key feature points on the surface shape. The optimal positioning and adjustment compensation values ​​at multiple positioning execution terminals are obtained.

[0166] S44: Inputting the solved optimal positioning compensation amount into the surface shape assembly accuracy Bayesian neural network prediction model established in S3, and quickly calculating the assembly deviation value of each key feature point on the surface shape under the current positioning compensation scheme;

[0167] S45: Use the actual measurement method in S11 to obtain the actual assembly deviation value of each key feature point on the surface shape, verify the effectiveness of the optimal positioning and adjustment compensation value solved in S43, and judge whether the actual assembly deviation value meets the accuracy requirements specified in the assembly process document. If not, adjust the calculation parameters in the Bayesian neural network prediction model and recalculate the optimal positioning compensation value until the assembly requirements are met, and obtain the optimal positioning and adjustment compensation value set after feedback correction.

[0168] The NSGA-II intelligent optimization algorithm in S43 generates new populations through an elite strategy and a competitive mechanism between parent and child generations, retaining and inheriting outstanding individuals within the population. By introducing the concept of crowding, it ensures a uniform distribution of solutions in the solution space, maintaining the diversity of the population.

[0169] The parameters of the NSGA-II intelligent optimization algorithm in S43 are set as follows: population size is 50, genetic generations is 70, crossover probability is 0.9, crossover distribution index is 10, mutation probability is 0.1, and mutation distribution index is 20.

[0170] exist Figure 2 and Figure 3 In the assembly scenario, the calculation result of the optimal flexible positioning and adjustment compensation amount in S45 is:

[0171] Positioning compensation Anchor point 1 Anchor point 2 Anchor point 3 Anchor point 4 Anchor point 5 Anchor point 6 Numerical size 0.042 0.021 0.003 -0.009 -0.037 -0.388

[0172] exist Figure 2 and Figure 3 In the assembly scenario, the optimal shape assembly deviation measurement results of the 9 key assembly feature points in S45 are (the unit of value is mm):

[0173]

[0174]

[0175] The S5 specifically includes:

[0176] S51: Based on the optimal positioning compensation amount under the assembly accuracy requirement, the variance value of the positioning accuracy of each end is derived according to the tolerance band of the positioning accuracy of each end by using statistical methods, and is represented by a diagonal matrix Λ';

[0177] S52: According to the calculation method in S2, based on the measurement data of the positioning accuracy sampling points of each positioning execution terminal, the surface sampling point deviations are fitted using Legendre polynomials, and the Legendre polynomial series used for fitting is generalized Fourier transform to construct a transformation matrix between different key feature points to represent the geometric correlation between the positioning accuracy sampling points of each positioning execution terminal;

[0178] S53: According to the calculation method in S2, a geometric covariance model is constructed based on the variance value and the change matrix to express the correlation and spatial distribution statistical laws between the positioning errors at the multi-positioning execution end of the tooling, and the distribution results are used to fit the three-dimensional spatial surface to intuitively and visually display the actual error distribution of each positioning end.

[0179] The present invention also provides a system for optimizing the design of flexible positioning, adjustment and compensation for thin-walled curved surface structures of aviation composite materials, characterized in that the system comprises:

[0180] The part error distribution modeling module is used to express the distribution state and geometric continuity of part errors based on the measured manufacturing error information of thin-walled curved surface structural parts, serving as the part input basis for the prediction of surface shape assembly accuracy considering the actual manufacturing status;

[0181] The module for theoretical calculation of external assembly accuracy is used to establish an assembly deviation calculation model that takes into account the flexible positioning and adjustment compensation according to the load conditions during the assembly process, and to predict the numerical value of external assembly accuracy.

[0182] The shape assembly accuracy data-driven calculation module is used to establish a shape assembly accuracy prediction model for real surfaces using a data-driven method based on a neural network;

[0183] Flexible positioning and adjustment compensation optimization module, used to build and solve multi-objective optimization models, and intelligently obtain the optimal positioning compensation at multiple positioning execution ends on the tooling;

[0184] Tooling positioning error distribution state modeling module: used to express the correlation and distribution statistical laws between the optimal positioning errors at the end of tooling multi-positioning execution.

[0185] The present invention provides a method for optimizing the design of flexible positioning, assembly and compensation of thin-walled curved surface structures of aviation composite materials. Figure 2 The following is a flow chart of the specific implementation of this solution in this embodiment. The overall idea adopted by the present invention to solve its technical problems is:

[0186] First, digital measurement methods are used to obtain the actual manufacturing error state of the curved surface structural parts to be assembled, and mathematical processing methods are used to express the distribution state and correlation of the errors.

[0187] Secondly, based on the actual error state of the workpiece and the load conditions during the assembly process, the positioning and adjustment compensation of the flexible assembly tooling is used as a separate variable. The super-element stiffness theory is applied to establish a calculation model for the external assembly deviation of curved structural parts and predict the assembly deviation value.

[0188] Thirdly, based on the theoretical calculation data of assembly error and taking the flexible positioning compensation as input, a data-driven method based on neural network is adopted to establish a shape assembly accuracy prediction model for real curved surfaces, thus realizing the rapid calculation of assembly accuracy.

[0189] Next, with the goal of minimizing assembly accuracy at key feature points on the curved surface, an optimization function for flexible positioning and adjustment compensation was established. Intelligent solution methods were used to obtain the optimal set of positioning compensation values ​​at multiple positioning execution terminals on the tooling. Field measurements were then used to verify the effectiveness of the solution and provide further feedback and corrections.

[0190] Finally, the correlation and distribution statistics between the positioning errors at the end of the multi-positioning execution of the tooling are expressed, and the distribution results are visualized, laying a basic analytical foundation for high-precision assembly quality assurance and low-stress assembly.

[0191] In summary, compared with the existing technology, the present invention realizes accurate prediction and calculation modeling of the assembly deviation of the curved structural parts by establishing an assembly deviation transmission mechanism calculation model that integrates the real force-position-measured state of the assembly process; realizes rapid calculation of assembly accuracy by establishing a data-driven real curved surface shape assembly accuracy prediction model; establishes an intelligent optimization solution for flexible positioning and adjustment compensation and a measured data feedback correction mechanism, and applies the means of measurement to verify the effectiveness of the solution and further correct it; establishes a correlation between the key feature points of the curved workpiece and the error distribution at each positioning execution end on the tooling, and proposes a mathematical expression method for the error distribution state, laying a front-end analysis foundation for achieving high-quality, precise assembly and low-stress assembly of high-end products.

Claims

1. A method for optimizing the design of flexible positioning and adjustment compensation for thin-walled curved surface structures of aviation composite materials, characterized in that: The calculation and evaluation method for cumulative transmission of coordination errors in assembly of aviation thin-walled structures comprises the following steps: S1: Measure and obtain actual manufacturing and assembly status data of thin-walled curved surface parts of aviation composite materials, using the mathematical methods of Legendre polynomials and generalized Fourier transform to express the actual distribution state and geometric continuity of part errors; S2: Based on the actual distribution state and geometric continuity of the obtained part errors, combined with the positioning load conditions during the assembly process, an assembly deviation calculation model is established that integrates the flexible positioning and adjustment compensation amounts at multiple positioning execution ends to obtain the surface shape assembly accuracy value; S3: Based on the surface shape assembly accuracy values ​​obtained by the assembly deviation calculation model and the measured assembly status data information, and with the flexible positioning and adjustment compensation values ​​at multiple positioning execution terminals as input, a data-driven method based on a neural network is used to establish a data-driven surface shape assembly accuracy prediction model; S4: With the goal of minimizing the prediction results of the data-driven surface shape assembly accuracy prediction model, an optimization objective function is established and solved using an optimization algorithm. The optimal positioning and adjustment compensation value set at multiple positioning execution ends on the tooling is obtained. The effectiveness of the solution is then compared and analyzed using field measurements. After that, feedback is provided to further correct the optimal positioning and adjustment compensation value set. S5: A geometric covariance model is constructed by feedback-correcting the optimal positioning and adjustment compensation value set to express the correlation and distribution statistical law between the positioning errors at the end of the multi-positioning execution of the tooling, and the distribution results are visualized.

2. The method for optimizing the design of flexible positioning and adjustment compensation for thin-walled curved surface structures of aviation composite materials according to claim 1, characterized in that: Said S1 specifically includes: S11: Considering the low manufacturing precision and large size of thin-walled composite structural parts in aviation, a laser tracker and laser scanner are combined to measure key feature points on product parts that reflect assembly quality characteristics, as well as the deviation status of positioning measurement points that match the positioning execution end points of the assembly tooling. S12: Based on the deviation states of the key feature points and the positioning measurement points that match the positioning execution ends of the assembly tooling, a geometric covariance model of the surface part on a two-dimensional plane is established, and the surface sampling point deviations are fitted using Legendre polynomials; S13: Use the least squares method to solve the coefficients of the Legendre polynomials, perform a generalized Fourier transform on the Legendre polynomial series used for fitting, and express the relationship between the errors of different key feature points; S14: Based on the geometric covariance model on the two-dimensional level, a geometric covariance model of the surface shape error of the curved workpiece on the three-dimensional level is established to accurately express the distribution state and geometric continuity of the part error.

3. The method for optimizing the design of flexible positioning and adjustment compensation for thin-walled curved surface structures of aviation composite materials according to claim 2, characterized in that: The S2 specifically includes: S21: In the positioning stage of thin-walled curved surface structural parts, an M-2-1 positioning layout method is adopted, where M is the number of positioning points, and the positioning points are divided into deterministic positioning points and over-constrained positioning points, where M>3 and M is a positive integer; S22: Based on the actual distribution state and geometric continuity of the obtained part errors and the geometric covariance model of the curved surface parts on the 3D level, the principles of rigid body kinematics are applied to analyze the assembly position and posture changes of thin-walled curved surface structural parts after the positioning stage, and the changes in key feature points on the shape are obtained; S23: Based on the changes in assembly position and posture, the positioning compensation at the end of the multi-positioning execution is introduced as the input variable for the tool positioning at the deterministic positioning point and the over-constrained positioning point. Then, based on the positioning load conditions during the assembly process, a clamping force or adsorption force is applied at the positioning point to securely connect the assembly tool and the part. S24: Under the action of assembly clamping force or adsorption force, the super-element stiffness matrix theory is applied. On the basis of introducing positioning compensation, a calculation model for the assembly deviation of curved structural parts considering the positioning compensation of flexible tooling is established. The deviation change of key feature points on the workpiece along the workpiece normal direction is calculated, laying the foundation for subsequent optimization of positioning compensation. S25: In the subsequent fastening connections of different workpieces and the removal of the assembly, the super-element stiffness matrix theory is applied to obtain the stiffness matrix of the entire assembly. By calculating the deviation changes of key feature points on the workpiece along the normal direction of the workpiece, the numerical value of the assembly accuracy at the key external shape measuring points of the entire assembly under the action of the clamping force is obtained.

4. The method for optimizing the design of flexible positioning and adjustment compensation for thin-walled curved surface structures of aviation composite materials according to claim 3, characterized in that: The S3 specifically includes: S31: Perform homogenization treatment on the composite thin-walled curved surface structure to obtain the overall physical performance parameters and material mechanical constants of the composite structure; S32: Based on the overall physical performance parameters and material mechanical constants of the composite structure, within the allowable range of positioning accuracy, the flexible positioning and adjustment compensation parameters at multiple positioning execution terminals are changed to obtain a set of assembly accuracy values ​​at key shape measurement points of the entire assembly under a small sample, and the optimal Latin hypercube sampling method is used for data sampling; S33: Using the sampling data obtained using the optimal Latin hypercube sampling method as training data, a mathematical modeling method based on a Bayesian neural network is applied to fit a proxy model, and the proxy model is used as a prediction model with the flexible positioning and adjustment compensation amounts at multiple positioning execution terminals as input and the shape assembly accuracy of the actual surface as output; S34: Statistical indicators are used to perform computational verification on the output prediction model. By reducing the deviation between the trained Bayesian neural network prediction model and the true model, the mapping relationship between the compensation amount of the adjustment system and the assembly deviation is accurately represented.

5. The method for optimizing the design of flexible positioning and adjustment compensation for thin-walled curved surface structures of aviation composite materials according to claim 4, characterized in that: The S4 specifically includes: S41: By mapping the flexible positioning and adjustment compensation amounts at multiple positioning execution terminals to the assembly deviations, the assembly deviation values ​​of the key feature points on the curved surface are coordinated and balanced. While ensuring that the assembly deviation values ​​at the key feature points meet the requirements, the shape assembly deviation is minimized. The problem of solving the flexible positioning and adjustment compensation amounts at multiple positioning execution terminals is converted into a multi-objective optimization problem. S42: Establish a multi-objective optimization model using the positioning and adjustment compensation amounts at multiple positioning execution terminals on the flexible tooling as the variables to be optimized, minimizing the shape assembly deviation output of each key point as the optimization goal, and using the positioning accuracy range of the assembly tooling and the shape assembly deviation requirements specified in the assembly process document as constraints; S43: The NSGA-II algorithm is used to analyze and solve the optimal positioning compensation. Using the Matlab software platform, the optimal solution that minimizes the assembly deviation of each key feature point on the surface shape is obtained, and the optimal positioning and adjustment compensation at multiple positioning execution ends is obtained. S44: Inputting the optimal positioning and adjustment compensation values ​​at the multiple positioning execution terminals into the surface shape assembly accuracy prediction model to quickly calculate the assembly deviation value of each key feature point on the surface shape under the positioning compensation solution; S45: Use the actual measurement method in S11 to obtain the actual assembly deviation value of each key feature point on the surface shape, conduct comparative analysis and verification on the effectiveness of the optimal positioning and adjustment compensation amount solved in S43, and judge whether the actual assembly deviation value meets the accuracy requirements specified in the assembly process document. If not, adjust the calculation parameters in the Bayesian neural network prediction model, recalculate the optimal positioning compensation amount at multiple positioning execution ends until the assembly requirements are met, and obtain the optimal positioning and adjustment compensation amount set corrected by feedback.

6. The method for optimizing the design of flexible positioning and adjustment compensation for thin-walled curved surface structures of aviation composite materials according to claim 5, characterized in that: The S5 specifically includes: S51: Based on the optimal positioning compensation amount corrected by feedback while meeting the assembly accuracy requirements, statistical methods are used to derive the variance of the positioning accuracy values ​​of each end according to the tolerance band of the positioning accuracy of each end, and the variance is expressed as a diagonal matrix; S52: Based on the measurement data of the sampling points of the positioning accuracy of each positioning terminal, the surface sampling point deviation is fitted using Legendre polynomials, and the Legendre polynomial series used for fitting is generalized Fourier transformed to construct a transformation matrix between different key feature points to represent the geometric correlation between the positioning accuracy sampling points of each positioning execution terminal; S53: Based on the variance value and transformation matrix, a geometric covariance model is constructed to express the correlation and spatial distribution statistical laws between the positioning errors at the multiple positioning execution ends of the tooling, and the distribution results are used to fit the three-dimensional spatial surface to intuitively and visually display the actual error distribution of the positioning positions at multiple positioning execution ends.

7. The method for optimizing the design of flexible positioning and adjustment compensation for thin-walled curved surface structures of aviation composite materials according to claim 6, characterized in that: The method of fitting the surface sampling point deviation using Legendre polynomials includes: The arrangement direction of the key feature points in the U direction on the surface of the part is selected as the main assembly direction, and the V direction perpendicular to the main direction is selected as the secondary assembly direction; The sampling point deviations on the same sampling line in the main and secondary directions are respectively used as the basis for two-dimensional curve fitting, and geometric covariance models on two-dimensional planes are respectively established in the main and secondary directions.

8. The method for optimizing the design of flexible positioning and adjustment compensation for thin-walled curved surface structures of aviation composite materials according to claim 6, characterized in that: The step of fitting the surface sampling point deviations in the main direction or the sub-direction using Legendre polynomials and performing a generalized Fourier transform on the Legendre polynomial series used for fitting includes: The surface sampling point deviation is expressed using Legendre polynomial series of five orders from 0 to 4. r(x i ) * =a s0 F s0 (x i )+a s1 F s1 (x i )+a s2 F s2 (x i )+a s3 F s3 (x i )+a s4 F s4 (x i ) Where: N is the number of sampling points on the sampling line of the main assembly direction, l is the order of Legendre polynomial, a sl is the coefficient of the l-th order Legendre polynomial on the s-th sampling line in the main direction, which is solved by the least squares fitting method; Pl(x i ),Pl(x j ) are the l-order Legendre polynomial functions at x i 、x j The value at r is the value of x on the sth sampling line in the main direction j The deviation value of the sampling point at , R is the l-th order Fourier Legendre polynomial function on the s-th sampling line at x i The value at ; r* is the fitting value of the jth sampling point on the sth sampling line in the main direction of the assembly; Use the least squares method to solve a sl Coefficients, for approximating the profile deviation curve with the fitting polynomial curve; Perform a generalized Fourier transform on the Legendre polynomial series and construct the transformation matrix S, where s rl is the coefficient of the l-th order Legendre polynomial on the r-th sampling line in the main direction, expressing the relationship between different key feature points.

9. The method for optimizing the design of flexible positioning and adjustment compensation for thin-walled curved surface structures of aviation composite materials according to claim 6, characterized in that: The method of establishing a geometric covariance model of the surface shape error of a curved workpiece on a three-dimensional level based on the two-dimensional geometric covariance model includes: Arrange the sampling points in order of the main and secondary directions; Applying the rounding and remaindering methods to convert the serial number of the sampling point into the coordinates of the sampling point within the sampling area; Express the relationship between the deviation of the I-th sampling point with coordinate value (i, j) and the deviation of the J-th sampling point with coordinate value (i0, j0); Using statistical methods, the variance of the accuracy values ​​of each key sampling point on the part is derived based on the tolerance band of each sampling point and expressed as a diagonal matrix Λ.

10. A system for optimizing the design of flexible positioning, adjustment and compensation for thin-walled curved surface structures of aviation composite materials, characterized in that: The flexible positioning, assembly and compensation amount optimization design system for aviation composite thin-walled curved surface structures includes: Part error distribution state modeling module, used to obtain the actual manufacturing and assembly state information of thin-walled curved surface parts of aviation composite materials, and use mathematical methods to express the distribution state and geometric continuity of part errors; The module for calculating the theoretical calculation of the shape assembly accuracy is used to obtain the actual error distribution state through the distribution state of the part error and the geometric continuity. Based on the actual error distribution state of the part, an assembly deviation calculation model is established that takes into account the flexible positioning and adjustment compensation amount to predict the surface shape assembly accuracy value. The shape assembly accuracy data-driven calculation module is used to predict the surface shape assembly accuracy value based on the assembly deviation calculation model. It uses a data-driven method based on a neural network to establish a surface shape assembly accuracy prediction model. The tooling positioning error distribution state modeling module is used for the flexible positioning and adjustment compensation optimization module, which is used to minimize the predicted results of the surface shape assembly accuracy prediction model, apply the optimization algorithm to solve, and obtain the optimal positioning and adjustment compensation value set at multiple positioning execution ends on the tooling.

Images