A product assembly accuracy prediction method and system driven by mechanism and data fusion

Abstract

The present invention provides a product assembly accuracy prediction method and system driven by mechanism and data fusion, including: S1: construction of a geometric error spinor model of characteristic changes of multiple error sources; S2: construction of an assembly error transmission mechanism model of multiple assemblies; S3: construction of an assembly error transmission correction model considering assembly deformation errors and measured errors; S4: construction of an assembly accuracy prediction data model based on support vector regression; S5: construction of a product assembly accuracy prediction model integrating mechanism and data; S6: verification of the effectiveness of a hybrid-driven assembly accuracy prediction model. The present invention comprehensively considers the deformation effects of tooling, part contact and springback and the load on the assembly, as well as the construction of a product assembly accuracy prediction model integrating mechanism and data, which can effectively improve the prediction of product assembly accuracy and has good interpretability, can significantly reduce repair costs and time, and is conducive to the rapid and accurate assembly of complex products on site.

Description

[Technical field]

[0001] The present invention relates to the field of mechanical product digital assembly coordination process technology in the field of mechanical manufacturing, and in particular to a product assembly accuracy prediction method and system driven by mechanism and data fusion. [Background technology]

[0002] As the final link in the manufacturing of modern high-end products, assembly is of great significance in improving product production efficiency and ensuring service performance. As an important manifestation of the production quality of aerospace products, the steady improvement and stable guarantee of assembly accuracy can effectively support the high performance and service requirements of the product throughout its life cycle, and avoid problems such as reduced production efficiency and performance loss due to assembly tolerances and excessive stress. However, in the process of analyzing and improving product assembly accuracy, there are many error links that affect assembly accuracy, and the error transmission mechanism is relatively complex. It is difficult to construct an error transmission model to accurately predict product assembly accuracy, which brings certain difficulties to improving product assembly accuracy. Therefore, how to model and analyze the assembly error transmission process and quickly and accurately construct an error transmission model is of great guiding significance for improving product assembly accuracy.

[0003] The accuracy of the error transfer model will greatly affect the subsequent assembly accuracy prediction and optimization. The construction methods mainly include error transfer mechanism model, data-driven assembly error prediction model and mechanism-data hybrid driven assembly error prediction model. Domestic and foreign researchers have conducted a lot of research on the construction of error transfer models in the above three aspects. At present, the relevant methods for the construction of mechanism models and data-driven assembly prediction models are relatively mature, but there is a lack of mechanism-data fusion model building methods. The method that only considers the mechanism model can complete the modeling of the system under ideal conditions. The model has good interpretability, but the prediction results of the model are less accurate; while the prediction results of the model based on data are more accurate, but it is difficult to explore the mechanism effect inside the system, and its interpretability is poor.

[0004] Therefore, it is necessary to study a product assembly accuracy prediction method and system driven by mechanism and data fusion, to model the fusion mechanism model and the data model, and to form an error transfer modeling method driven by mechanism analysis and data fusion, to improve the prediction of high product assembly accuracy, to address the shortcomings of existing technologies, and to solve or alleviate one or more of the above problems. [Summary of the invention]

[0005] In view of this, the present invention provides a product assembly accuracy prediction method and system driven by the fusion of mechanism and data, which comprehensively considers the deformation effects of tooling, part contact and springback and the load on the assembly, and constructs a product assembly accuracy prediction model that integrates mechanism and data. It can effectively improve the prediction of product assembly accuracy and has good interpretability, which can significantly reduce repair costs and time.

[0006] On the one hand, the present invention provides a product assembly accuracy prediction method driven by mechanism and data fusion, the assembly accuracy prediction method is used to predict the assembly accuracy of composite material wing box assembly parts during the production process, and the assembly prediction method comprises the following steps:

[0007] S1: According to the structural characteristics and assembly process of the parts to be produced, a geometric error screw model with characteristic changes of multiple basic error sources is constructed;

[0008] S2: According to the matching relationship between the parts to be assembled and the positioning of the tooling, the assembly error transmission mechanism model of multiple assemblies is constructed;

[0009] S3: Based on the geometric error screw model, assembly deformation error and measured error assembly, the error transfer mechanism model is corrected to obtain the assembly error transfer correction model;

[0010] S4: acquiring parameter data based on the assembly error transfer correction model, constructing an assembly accuracy prediction data model based on support vector regression, optimizing the assembly accuracy prediction data model through the parameter data, and obtaining an assembly accuracy prediction data optimization model;

[0011] S5: Construct a product assembly accuracy prediction model through the assembly error transmission mechanism model, the assembly accuracy prediction data optimization model and the measured assembly accuracy data;

[0012] S6: Train the product assembly accuracy prediction model through measured assembly accuracy data and verify its effectiveness.

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

[0014] S11: Analyze the assembly product structure and assembly performance requirements, obtain specific product assembly accuracy requirements, and determine the various error sources existing in the product assembly process based on the assembly accuracy requirements, and convert the impact of the error sources on the assembly into corresponding feature changes;

[0015] S12: According to the key measurement points and geometric feature information of each error source, the position change data of the geometric features of the assembly are obtained, and the kinematic theory and small displacement screw method are used to convert the position change data into an error screw model in matrix format, so as to realize the construction of geometric error screw models for the characteristic changes of various basic error sources.

[0016] According to the above aspects and any possible implementation, an implementation is further provided, wherein the error sources in S11 include but are not limited to initial manufacturing deviation of parts, deformation deviation of thin-walled parts caused by the clamping force of assembly tooling, deformation deviation caused during assembly connection, and rebound deformation deviation caused during the removal process, wherein;

[0017] The deformation deviation of thin-walled parts caused by the clamping force of assembly tooling is specifically as follows: the deformation deviation generation mechanism of thin-walled parts caused by the positioning and clamping of assembly tooling is analyzed, the deformation of the parts is solved, and the contact force of the parts is calculated using the minimum total residual energy as the objective function under the condition of satisfying the constraints of the tooling positioning position and the unilateral contact constraints between the workpiece and the fixture, and the deformation of the parts is obtained by combining the stiffness matrix of the parts;

[0018] The deformation deviation caused by thin-walled parts in the assembly and connection process is specifically as follows: the mechanism of part contact deformation deviation caused by thin-walled parts in the assembly and connection process is analyzed, and the deviation of the wall plate along the X direction and the Y direction is obtained by establishing the relationship between the bolt connection force load and the deflection. The deformation error distribution is superimposed with the ideal design surface shape error to obtain the change of the actual shape of the part after contact deformation;

[0019] The springback deformation deviation caused by the removal process is specifically as follows: the mechanism of the springback deformation deviation of the parts caused by the removal of the assembly after assembly is completed is analyzed, and the relationship between the springback force and the clamping force of the tooling and the assembly deformation caused by the elastic force is analyzed by obtaining the assembly springback force of the product and the stiffness matrix of the entire assembly, and the deformation deviation caused by the springback is calculated and solved.

[0020] According to the above aspects and any possible implementation manner, an implementation manner is further provided, wherein S2 specifically includes:

[0021] S21: Consider the matching relationship of the characteristic surfaces of each component and the tooling positioning method according to the assembly sequence, and establish multiple sets of local coordinate systems and posture / tolerance constraints;

[0022] S22: The posture error conversion matrix is ​​obtained by combining the Jacobian matrix with the small displacement screw model. This matrix can describe the transmission relationship between the positioning tooling, the inside of the part and the mating surface of each error link in the product assembly process, thereby realizing the construction of the assembly error mechanism analysis model.

[0023] According to the above aspects and any possible implementation, an implementation is further provided, wherein S3 specifically includes:

[0024] S31: real-time acquisition of test data obtained by sensors pre-arranged on the assembly structure, and real-time and accurate acquisition of the deformation of the wing box during the assembly process by a laser displacement measuring instrument;

[0025] S32: Collect data through a laser tracker, extract the original node coordinates of key functional features and the deformation of each node along the three directions of the global coordinate system, superimpose the node deformation with the original coordinates along the corresponding direction, and obtain the actual geometric surface point set considering the deformation deviation. Use the least squares method to fit the actual measured geometric surface data to obtain the real fitting surface.

[0026] S33: Describe the influence of deformation of functional features of parts on tolerance in mathematical form, use the surface fitting method to describe the functional features of products, and on the basis of the original Jacobian rotation error transfer model, comprehensively consider the influence of component deformation during assembly, correct the assembly error transfer mechanism model, and obtain the assembly error transfer mechanism correction model.

[0027] According to the above aspects and any possible implementation manner, an implementation manner is further provided, wherein S4 specifically includes:

[0028] S41: According to the structure of the error transmission mechanism correction model, determine the data model acquisition requirements, and collect data. The data collection content is a data set related to assembly accuracy, including but not limited to parameter information and workpiece quality information in the assembly process;

[0029] S42: preprocessing the data collected by the data model to remove outliers, and then using a support vector regression method to train the assembly accuracy prediction data model. During the training process, a suitable fitting function and kernel function are selected to fit the data so that the assembly accuracy can be effectively predicted;

[0030] S43: The assembly accuracy prediction data model calculates the final product assembly accuracy deviation value for prediction by using the support vector regression method, and uses this deviation value to compensate the mechanism calculation result to obtain the actual calculation accuracy of the assembly.

[0031] As described above and any possible implementation method, a further implementation method is provided, in which the assembly accuracy prediction data model calculates the final product assembly accuracy deviation value for prediction by adopting the support vector regression method, and uses this deviation value to compensate the mechanism calculation result to obtain the actual calculation accuracy of the assembly.

[0032] According to the aspects described above and any possible implementation manner, an implementation manner is further provided, wherein the data collection content is a data set related to assembly accuracy, including but not limited to parameter information and workpiece quality information during the assembly process.

[0033] According to the above aspects and any possible implementation manner, an implementation manner is further provided, wherein S5 specifically includes:

[0034] S51: reading various parameters of product assembly, and inputting various assembly parameters into the product assembly error transmission mechanism model to obtain the theoretical calculation value of the mechanism model, and at the same time, subtracting the actual measured data such as assembly accuracy generated at the measurement collection site from the theoretical calculation value of the mechanism model to obtain the calculated deviation value of the mechanism model assembly error;

[0035] S52: Based on the assembly accuracy measured data, assembly process data, related simulation data and error data as training samples, a support vector regression-based assembly accuracy prediction data model is constructed, and the theoretical calculation value is used as the training input of the model to obtain the prediction value of the mechanism calculation deviation through the data model;

[0036] S53: Add the deviation prediction value calculated by the assembly accuracy prediction data model and the calculated deviation value of the mechanism model assembly error calculated by the assembly error transfer mechanism model to obtain the compensated assembly accuracy calculation value, and obtain the product assembly accuracy prediction model that integrates the mechanism and data.

[0037] According to the above aspects and any possible implementation manner, an implementation manner is further provided, wherein S6 specifically includes:

[0038] S61: Based on the engineering digital model provided by the design department, read the assembly process plan of the product, determine the assembly sequence of parts, the clamping plan and the matching relationship between parts, and collect the real physical assembly process data and the assembly parameters of the wing box assembly site, and organize the obtained data to obtain sample data for model verification;

[0039] S62: Using various assembly errors and measured values ​​as input, and the deviation between the theoretical calculated error value and the measured value as output, the product assembly accuracy prediction model integrating mechanism and data is trained based on the Matlab platform;

[0040] S63: Based on the training results of the product assembly accuracy prediction model that integrates the mechanism and data, verify the effectiveness of the accuracy prediction results of the test model.

[0041] According to the aspects and any possible implementation methods described above, an implementation method is further provided, wherein the content of verifying the effectiveness in S63 includes but is not limited to the mean absolute error MAE value, the mean square error MSE value, the root mean square error RMSE value and the absolute coefficient R2 value of the prediction model.

[0042] According to the above aspects and any possible implementation, a product assembly accuracy prediction system driven by mechanism and data fusion is further provided, wherein the assembly accuracy prediction system is used to predict the assembly accuracy of composite wing box assembly parts during the production process, and the assembly accuracy prediction system comprises:

[0043] A geometric error screw model building module is used to build a geometric error screw model with characteristic changes of multiple basic error sources according to the structural characteristics and assembly process of the parts to be produced;

[0044] Assembly error transmission mechanism model construction module: used to construct the assembly error transmission mechanism model of multiple assemblies based on the matching relationship between the parts to be assembled and the positioning of the tooling;

[0045] The assembly error transmission mechanism model correction module is used to correct the error transmission mechanism model based on the geometric error screw model, the assembly deformation error and the measured error assembly to obtain the assembly error transmission correction model;

[0046] An assembly accuracy prediction data optimization model construction optimization module is used to obtain parameter data based on the assembly error transfer correction model, construct an assembly accuracy prediction data model based on support vector regression, and optimize the assembly accuracy prediction data model through parameter data to obtain an assembly accuracy prediction data optimization model;

[0047] A product assembly accuracy prediction model fusion construction module is used to construct a product assembly accuracy prediction model through an assembly error transmission mechanism model, an assembly accuracy prediction data optimization model, and measured assembly accuracy data;

[0048] The product assembly accuracy prediction model training and verification module is used to train the product assembly accuracy prediction model through measured assembly accuracy data and verify its effectiveness.

[0049] Compared with the prior art, the present invention can achieve the following technical effects:

[0050] 1) Considering the deformation caused by the contact between parts and the deformation error factors caused by the contact between parts and tooling, the error transfer model is corrected by the shape tolerance after deformation, and the assembly error transfer mechanism model is accurately constructed;

[0051] 2) It can read the field measured data set containing various assembly parameters and assembly accuracy, organically integrate the mechanism model with the data, and construct a data-mechanism fusion mechanism calculation error prediction model, which quickly and effectively improves the assembly accuracy prediction results, and is conducive to the rapid and accurate assembly of complex products on site.

[0052] 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

[0053] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. 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 creative work.

[0054] Figure 1 It is a schematic diagram of the mechanism of the deformation error of parts caused by the positioning and clamping of the assembly tooling proposed by the present invention;

[0055] Figure 2 It is a schematic diagram of the deformation deviation caused by the thin-walled parts proposed by the present invention during the assembly and connection process;

[0056] Figure 3 It is a schematic diagram of the springback deformation caused by the thin-walled parts proposed by the present invention during the unloading process;

[0057] Figure 4 It is a schematic diagram of the overall structure and key assembly features of the wing box proposed by the present invention;

[0058] Figure 5 It is a schematic diagram of the assembly error transmission relationship at the straight seam of the assembly proposed by the present invention (including gap and step difference);

[0059] Figure 6 It is a schematic diagram of the idea of ​​constructing a product assembly accuracy prediction model that integrates the mechanism and data proposed by the present invention;

[0060] Figure 7 It is a schematic diagram of the straight seam gap calculation process under the mechanism-data hybrid drive proposed by the present invention;

[0061] Figure 8 It is a schematic diagram of input data and corresponding output data in the training sample set proposed by the present invention;

[0062] Fig. 9 It is a schematic diagram of the accuracy comparison of alternative prediction models proposed by the present invention;

[0063] Fig.10 The invention proposes a product assembly accuracy prediction method and system flow chart driven by mechanism and data fusion. [Specific implementation method]

[0064] 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.

[0065] It should be clear that the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0066] 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", "said" 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 other meanings.

[0067] The present invention provides a product assembly accuracy prediction method driven by mechanism and data fusion, the assembly accuracy prediction method is used to predict the assembly accuracy of composite material wing box assembly parts during the production process, and the assembly prediction method comprises the following steps:

[0068] S1: According to the structural characteristics and assembly process of the parts to be produced, a geometric error screw model with characteristic changes of multiple basic error sources is constructed;

[0069] S2: According to the matching relationship between the parts to be assembled and the positioning of the tooling, the assembly error transmission mechanism model of multiple assemblies is constructed;

[0070] S3: Based on the geometric error screw model, assembly deformation error and measured error assembly, the error transfer mechanism model is corrected to obtain the assembly error transfer correction model;

[0071] S4: acquiring parameter data based on the assembly error transfer correction model, constructing an assembly accuracy prediction data model based on support vector regression, optimizing the assembly accuracy prediction data model through the parameter data, and obtaining an assembly accuracy prediction data optimization model;

[0072] S5: Construct a product assembly accuracy prediction model through the assembly error transmission mechanism model, the assembly accuracy prediction data optimization model and the measured assembly accuracy data;

[0073] S6: Train the product assembly accuracy prediction model through measured assembly accuracy data and verify its effectiveness.

[0074] The S1 specifically includes:

[0075] S11: Analyze the assembly product structure and assembly performance requirements, obtain specific product assembly accuracy requirements, and determine the various error sources existing in the product assembly process based on the assembly accuracy requirements, and convert the impact of the error sources on the assembly into corresponding feature changes;

[0076] S12: According to the key measurement points and geometric feature information of each error source, the position change data of the geometric features of the assembly are obtained, and the kinematic theory and small displacement screw method are used to convert the position change data into an error screw model in matrix format, so as to realize the construction of geometric error screw models for the characteristic changes of various basic error sources.

[0077] The error sources in S11 include but are not limited to initial manufacturing deviation of parts, deformation deviation of thin-walled parts caused by the clamping force of assembly tooling, deformation deviation caused during assembly connection, and springback deformation deviation caused during the unloading process, wherein;

[0078] The deformation deviation of thin-walled parts caused by the clamping force of assembly tooling is specifically as follows: the deformation deviation generation mechanism of thin-walled parts caused by the positioning and clamping of assembly tooling is analyzed, the deformation of the parts is solved, and the contact force of the parts is calculated using the minimum total residual energy as the objective function under the condition of satisfying the constraints of the tooling positioning position and the unilateral contact constraints between the workpiece and the fixture, and the deformation of the parts is obtained by combining the stiffness matrix of the parts;

[0079] The deformation deviation caused by thin-walled parts in the assembly and connection process is specifically as follows: the mechanism of part contact deformation deviation caused by thin-walled parts in the assembly and connection process is analyzed, and the deviation of the wall plate along the X direction and the Y direction is obtained by establishing the relationship between the bolt connection force load and the deflection. The deformation error distribution is superimposed with the ideal design surface shape error to obtain the change of the actual shape of the part after contact deformation;

[0080] The springback deformation deviation caused by the removal process is specifically as follows: the mechanism of the springback deformation deviation of the parts caused by the removal of the assembly after assembly is completed is analyzed, and the relationship between the springback force and the clamping force of the tooling and the assembly deformation caused by the elastic force is analyzed by obtaining the assembly springback force of the product and the stiffness matrix of the entire assembly, and the deformation deviation caused by the springback is calculated and solved.

[0081] The S2 specifically includes:

[0082] S21: Consider the matching relationship of the characteristic surfaces of each component and the tooling positioning method according to the assembly sequence, and establish multiple sets of local coordinate systems and posture / tolerance constraints;

[0083] S22: The posture error conversion matrix is ​​obtained by combining the Jacobian matrix with the small displacement screw model. This matrix can describe the transmission relationship between the positioning tooling, the inside of the part and the mating surface of each error link in the product assembly process, thereby realizing the construction of the assembly error mechanism analysis model.

[0084] The S3 specifically includes:

[0085] S31: real-time acquisition of test data obtained by sensors pre-arranged on the assembly structure, and real-time and accurate acquisition of the deformation of the wing box during the assembly process by a laser displacement measuring instrument;

[0086] S32: Collect data through a laser tracker, extract the original node coordinates of key functional features and the deformation of each node along the three directions of the global coordinate system, superimpose the node deformation with the original coordinates along the corresponding direction, and obtain the actual geometric surface point set considering the deformation deviation. Use the least squares method to fit the actual measured geometric surface data to obtain the real fitting surface.

[0087] S33: Describe the influence of deformation of functional features of parts on tolerance in mathematical form, use the surface fitting method to describe the functional features of products, and on the basis of the original Jacobian rotation error transfer model, comprehensively consider the influence of component deformation during assembly, correct the assembly error transfer mechanism model, and obtain the assembly error transfer mechanism correction model.

[0088] The S4 specifically includes:

[0089] S41: Modify the model structure according to the error transmission mechanism, determine the data model acquisition requirements, and collect data.

[0090] S42: pre-processing the data collected by the data model to remove outliers, and then using the support vector regression method to train the assembly accuracy prediction data model. During the training process, by selecting appropriate fitting functions and kernel functions to fit the data, it can effectively predict the assembly accuracy.

[0091] The assembly accuracy prediction data model uses the support vector regression method to calculate the final product assembly accuracy deviation value for prediction, and uses this deviation value to compensate the mechanism calculation result to obtain the actual calculation accuracy of the assembly.

[0092] The data collection content is a data set related to assembly accuracy, including but not limited to parameter information and workpiece quality information during the assembly process.

[0093] The S5 specifically includes:

[0094] S51: reading various parameters of product assembly, and inputting various assembly parameters into the product assembly error transmission mechanism model to obtain the theoretical calculation value of the mechanism model, and at the same time, subtracting the actual measured data such as assembly accuracy generated at the measurement collection site from the theoretical calculation value of the mechanism model to obtain the calculated deviation value of the mechanism model assembly error;

[0095] S52: Based on the assembly accuracy measured data, assembly process data, related simulation data and error data as training samples, a support vector regression-based assembly accuracy prediction data model is constructed, and the theoretical calculation value is used as the training input of the model to obtain the prediction value of the mechanism calculation deviation through the data model;

[0096] S53: Add the deviation prediction value calculated by the assembly accuracy prediction data model and the calculated deviation value of the mechanism model assembly error calculated by the assembly error transfer mechanism model to obtain the compensated assembly accuracy calculation value, and obtain the product assembly accuracy prediction model that integrates the mechanism and data.

[0097] The S6 specifically includes:

[0098] S61: Based on the engineering digital model provided by the design department, read the assembly process plan of the product, determine the assembly sequence of parts, the clamping plan and the matching relationship between parts, and collect the real physical assembly process data and the assembly parameters of the wing box assembly site, and organize the obtained data to obtain sample data for model verification;

[0099] S62: Using various assembly errors and measured values ​​as input, and the deviation between the theoretical calculated error value and the measured value as output, the product assembly accuracy prediction model integrating mechanism and data is trained based on the Matlab platform;

[0100] S63: Based on the training results of the product assembly accuracy prediction model that integrates the mechanism and data, verify the effectiveness of the accuracy prediction results of the test model.

[0101] The contents of the verification effectiveness in S63 include but are not limited to the mean absolute error MAE value, mean square error MSE value, root mean square error RMSE value and absolute coefficient R2 value of the prediction model.

[0102] The present invention also provides a product assembly accuracy prediction system driven by mechanism and data fusion, the assembly accuracy prediction system is used to predict the assembly accuracy of composite material wing box assembly parts during the production process, and the assembly accuracy prediction system includes:

[0103] A geometric error screw model building module is used to build a geometric error screw model with characteristic changes of multiple basic error sources according to the structural characteristics and assembly process of the parts to be produced;

[0104] Assembly error transmission mechanism model construction module: used to construct the assembly error transmission mechanism model of multiple assemblies based on the matching relationship between the parts to be assembled and the positioning of the tooling;

[0105] The assembly error transmission mechanism model correction module is used to correct the error transmission mechanism model based on the geometric error screw model, the assembly deformation error and the measured error assembly to obtain the assembly error transmission correction model;

[0106] An assembly accuracy prediction data optimization model construction optimization module is used to obtain parameter data based on the assembly error transfer correction model, construct an assembly accuracy prediction data model based on support vector regression, and optimize the assembly accuracy prediction data model through parameter data to obtain an assembly accuracy prediction data optimization model;

[0107] A product assembly accuracy prediction model fusion construction module is used to construct a product assembly accuracy prediction model through an assembly error transmission mechanism model, an assembly accuracy prediction data optimization model, and measured assembly accuracy data;

[0108] The product assembly accuracy prediction model training and verification module is used to train the product assembly accuracy prediction model through measured assembly accuracy data and verify its effectiveness.

[0109] The principle of the present invention is as follows:

[0110] The present invention provides a product assembly accuracy prediction method driven by mechanism and data fusion, which mainly includes three parts: construction of an assembly error mechanism model considering assembly deformation, an assembly accuracy prediction data model based on support vector regression, and construction of a product assembly accuracy prediction model integrating mechanism and data, which can achieve: 1) considering deformation factors such as deformation deviation caused by the clamping force of assembly tooling on thin-walled parts, deformation deviation caused by part assembly connection contact, and springback deformation deviation caused by part removal, adopting a deformed shape tolerance correction error transfer model to accurately complete the construction of an assembly error mechanism transmission mechanism model of multiple assemblies, so that the model is more accurate and has better interpretability; 2) being able to read a field measured data set containing various assembly parameters and assembly accuracy, constructing a data model for product assembly accuracy prediction based on a support vector regression algorithm, so that the prediction result of the model is more accurate; 3) organically integrating the mechanism model with the data model, constructing a data-mechanism fusion mechanism calculation error prediction model, and comprehensively utilizing the advantages of the mechanism model and the data model, which can effectively improve the accuracy of the product assembly accuracy prediction result and has better interpretability. On the one hand, the present invention provides a product assembly accuracy prediction method driven by mechanism and data fusion, the assembly accuracy prediction method includes three parts: construction of an assembly error mechanism model considering assembly deformation, construction of an assembly accuracy prediction data model based on support vector regression, and construction of a product assembly accuracy prediction model fused with mechanism and data, such as Fig.10 As shown, the three-part implementation includes the following steps:

[0111] S1: Construction of geometric error spinor model with characteristic changes of multiple error sources;

[0112] S2: Construction of assembly error transmission mechanism model for multiple assemblies;

[0113] S3: Construction of assembly error transfer correction model considering assembly deformation error and measured error;

[0114] S4: Construction of assembly accuracy prediction data model based on support vector regression;

[0115] S5: Construction of product assembly accuracy prediction model integrating mechanism and data;

[0116] S6: Validation of the hybrid-driven assembly accuracy prediction model.

[0117] In a specific embodiment, the S1 specifically includes:

[0118] S11: Analyze the assembly product structure and assembly performance requirements, obtain specific product assembly accuracy requirements, and clarify the various error sources in the product assembly process based on the assembly accuracy characteristics, mainly including the initial manufacturing deviation of parts, the deformation deviation of thin-walled parts caused by the clamping force of assembly tooling, the deformation deviation caused by the assembly connection process, and the rebound deformation deviation caused by the removal process. Convert the impact of these error sources on the assembly into changes in the corresponding characteristics.

[0119] S12: Analyze the deformation deviation mechanism of thin-walled parts caused by the positioning and clamping of assembly tools, and solve the deformation of the parts. Under the condition of satisfying the constraints of the tooling positioning position and the one-sided contact constraints between the workpiece and the fixture, the contact force of the parts is calculated using the minimum total residual energy as the objective function, and the deformation of the parts is obtained by combining the stiffness matrix of the parts.

[0120] S13: Analyze the mechanism of part contact deformation deviation caused by thin-walled parts during assembly and connection. By establishing the relationship between bolt connection force load and deflection, the deviation of the wall panel along the X and Y directions is obtained. The deformation error distribution is superimposed with the ideal design surface shape error to obtain the change in the actual shape of the part after contact deformation.

[0121] S14: Analyze the mechanism of springback deformation deviation of parts after the assembly is completed and removed from the shelf. By obtaining the assembly springback force of the product and the stiffness matrix of the entire assembly, analyze the relationship between the springback force and the clamping force of the tooling and the assembly deformation caused by the elastic force, and calculate the deformation deviation caused by the springback.

[0122] S15: The position change data of the geometric features of the assembly are obtained based on the key measurement points and geometric feature information of each error source. The kinematic theory and small displacement screw method are used to convert the position change data into an error screw model in a matrix format, so as to realize the construction of a geometric error screw model for the feature changes of various basic error sources.

[0123] In a specific embodiment, the S2 specifically includes:

[0124] S21: Based on the assembly sequence, the matching relationship of the characteristic surfaces of each component and the tooling positioning method are considered, and multiple sets of local coordinate systems and posture / tolerance constraints are established.

[0125] S22: The posture error conversion matrix is ​​obtained by combining the Jacobian matrix with the small displacement screw model. This matrix can describe the transmission relationship between the positioning tooling, the inside of the part and the mating surface of each error link in the product assembly process, thereby realizing the construction of the assembly error mechanism analysis model.

[0126] In a specific embodiment, the S3 specifically includes:

[0127] S31: real-time acquisition of test data obtained by sensors pre-arranged on the assembly structure, and real-time and accurate acquisition of the deformation of the wing box during the assembly process by a laser displacement measuring instrument.

[0128] S32: Collect data through laser tracker, extract the original node coordinates of key functional features and the deformation of each node along the three directions of the global coordinate system. Superimpose the deformation of these nodes with the original coordinates along the corresponding directions to obtain the actual geometric surface point set considering the deformation deviation. Use the least squares method to fit the actual measured geometric surface data to obtain the real fitting surface.

[0129] S33: Describe the influence of deformation of functional features of parts on tolerance in mathematical form, and describe the functional features of products by fitting surface method. Based on the original Jacobian rotation error transfer model, the influence of component deformation during assembly is comprehensively considered, and the above error transfer model is modified to achieve the correction of error transfer model.

[0130] In a specific embodiment, the S4 specifically includes:

[0131] S41: First, according to the structure of the assembly error mechanism model, clarify the data model collection requirements and conduct data collection, mainly collecting data sets related to assembly accuracy, including various parameters in the assembly process, workpiece quality and other information.

[0132] S42: pre-processing the data collected by the data model to remove outliers, and then using the support vector regression method to train the assembly accuracy prediction data model. During the training process, by selecting appropriate fitting functions and kernel functions to fit the data, it can effectively predict the assembly accuracy.

[0133] S43: The assembly accuracy prediction data model calculates the final product assembly accuracy deviation value for prediction by using the support vector regression method, and uses this deviation value to compensate the mechanism calculation result to obtain the actual calculation accuracy of the assembly.

[0134] In a specific embodiment, the S4 specifically includes:

[0135] S51: Read various parameters of product assembly, and input various assembly parameters into the product assembly error transmission mechanism model to obtain the theoretical calculation value of the mechanism model. At the same time, subtract the actual measured data such as assembly accuracy generated at the measurement collection site from the theoretical calculation value of the mechanism model to obtain the calculated deviation value of the mechanism model assembly error.

[0136] S52: Based on the assembly accuracy measured data, assembly process data, related simulation data and error data as training samples, an assembly accuracy prediction data model based on support vector regression is constructed. The theoretical value calculated by the previous assembly error transmission mechanism model is used as the input of the obtained training model, and the predicted value of the mechanism calculation deviation is obtained through the data model after substitution.

[0137] S53: Add the deviation prediction value calculated by the assembly accuracy prediction data model to the theoretical value calculated by the assembly error transmission mechanism model to obtain the compensated assembly accuracy calculation value, and complete the construction of the assembly accuracy prediction model of the mechanism and data.

[0138] In a specific embodiment, the S5 specifically includes:

[0139] S61: Based on the engineering digital model provided by the design department, read the product assembly process plan, determine the assembly sequence of parts, the clamping plan and the matching relationship between parts. Collect the real physical assembly process data and the assembly parameters of the wing box assembly site, and organize the obtained data to obtain sample data for model verification.

[0140] S62: Taking various assembly errors and measured values ​​as input, and the deviation between the theoretical calculated error value and the measured value as output, the assembly accuracy prediction model integrating mechanism and data is trained based on the Matlab platform.

[0141] S63: According to the training results of the assembly accuracy prediction model of the fusion mechanism and data, the effectiveness of the model accuracy prediction results is analyzed, including the mean absolute error MAE value, mean square error MSE value, root mean square error RMSE value, absolute coefficient R2 value, etc. of the prediction model.

[0142] The present invention provides a product assembly accuracy prediction method driven by mechanism and data fusion, which comprises three parts: construction of an assembly error mechanism model considering assembly deformation, construction of an assembly accuracy prediction data model based on support vector regression, and construction of a product assembly accuracy prediction model integrating mechanism and data.

[0143] Embodiment 1:

[0144] In this embodiment, the assembly accuracy prediction of the straight seam gap and step difference of a certain type of composite wing box assembly is taken as an example. Figure 1 As shown in the figure, the wing box is composed of machined ribs, front and rear beams, frame plates, partitions and composite skins. The assembly sequence is from outside to inside, and the wing box frame and composite wall panels are assembled in sequence. At the same time, during the assembly process, it is necessary to control the assembly accuracy of the left straight seam gap and step difference to ensure the assembly quality. Taking the key assembly error loop prediction value of the left straight seam gap and step difference of the product as an example, the implementation steps of the product assembly accuracy prediction method driven by mechanism and data fusion are explained.

[0145] The general idea adopted by the present invention to solve the technical problem is:

[0146] Firstly, according to the structural characteristics and assembly process of the parts to be produced, a geometric error screw model with characteristic changes of multiple basic error sources is constructed; and according to the matching relationship between the assembled parts and the positioning of the tooling, the Jacobian screw matrix is ​​constructed to obtain the assembly error transmission mechanism model of multiple assemblies; secondly, considering the influence of the initial manufacturing error of the parts, the load deformation of the parts and the error of the tooling fixture, the geometric error screw model matrix is ​​corrected based on the deformation of the parts and the tooling and the measured data to obtain the corrected assembly error transmission mechanism model; then, the data sets related to the assembly accuracy are collected and preprocessed, and the support vector regression method is used to train, evaluate and optimize the assembly accuracy prediction data model, and the trained model is used to predict the new data to realize the application of the assembly accuracy prediction data model and obtain the optimal assembly accuracy prediction data model; finally, the field measured data set containing various assembly parameters and assembly accuracy is read, the mechanism model is organically integrated with the data, and a mechanism calculation error prediction model integrating data and mechanism is constructed, which can solve the problems of difficult construction of assembly error transmission and difficult prediction of assembly accuracy and provide an effective solution.

[0147] The present invention provides a product assembly accuracy prediction method driven by mechanism and data fusion, specifically a method which takes a data model as the main body and a mechanism model as the auxiliary, integrates the mechanism model into the feature hierarchy of the data model, and the obtained product assembly accuracy prediction method driven by mechanism and data fusion is used for assembly accuracy prediction in the assembly process of mechanical products. The product assembly accuracy prediction method driven by mechanism and data fusion comprises the following steps:

[0148] S1: Construction of geometric error screw model with characteristic changes of multiple basic error sources

[0149] Specifically, we first need to clarify the various sources of errors that exist in the product assembly process. According to the four main processes of product assembly: positioning, clamping, connection, and springback after unloading, we identify the main sources of product assembly errors, including the initial manufacturing deviation of the product, the deformation deviation of thin-walled parts caused by the clamping force of the assembly tooling, the deformation deviation of parts caused by the connection force, and the springback deformation deviation of parts. Subsequently, the impact of these error sources on assembly is converted into changes in the corresponding features.

[0150] Furthermore, considering that aerospace products contain a large number of thin-walled structures and intersection structure parts, the following three surface types are used to describe the manufacturing errors according to the key geometric features of the above parts: the plane error mainly changes perpendicular to the plane surface, including displacement along the z direction and rotation around the x and y axes; the cylindrical surface change is the displacement along the x and y directions and rotation around the x and y axes; the curved surface change is the displacement along the x, y, and z directions and rotation around the x, y, and z axes. The changes in the above error forms are expressed in the form of a spinor matrix, as shown in Table 1.

[0151]

[0152] Table 1

[0153] Furthermore, the deformation deviation mechanism of thin-walled parts caused by the positioning and clamping of assembly tooling is analyzed, the deformation of the parts is solved, and expressed by the small displacement screw method. In the process of modeling the deformation deviation of thin-walled workpieces caused by the positioning and clamping of assembly tooling, due to factors such as the positioning accuracy of the tooling itself and the deformation of the tooling caused by the force between the tooling and the feature surface of the part, the deformation error of the workpiece caused by the positioning and clamping factors of the assembly tooling is inevitable: the installation and processing errors of the positioning elements of the assembly tooling will cause the position of the tooling positioning end to shift, and the position error of the thin-walled workpiece will also cause the position of the product positioning point to shift. The above two kinds of shifts, together with the multiple clamping forces and contact model of the assembly tooling, cause the position of the thin-walled workpiece to shift, thereby causing the deformation of the positioning elements, the contact surface and the thin-walled workpiece, resulting in part posture deviation. Its generation mechanism is as follows: Figure 1 shown.

[0154] According to the mechanism of part deformation error caused by tooling positioning and clamping, the contact force between the tooling and thin-walled parts is calculated. While satisfying the constraints of the tooling positioning position and the unilateral contact constraints between the workpiece and the fixture (the parts and the tooling positioning execution ends are always in contact, and the normal of the contact force points to the part), the contact force between the tooling and the part is calculated using the minimum total residual energy as the objective function, and the deformation of the part is obtained by combining the stiffness matrix of the part:

[0155]

[0156]

[0157]

[0158]

[0159] In the formula, k w is the stiffness matrix of the workpiece, is the nodal force of the workpiece (including gravity, machining force and corresponding contact force), W e is the machining force torque, To clamp the clamping element in step j, is the contact force in clamping step j. According to the above calculation method, the contact force in clamping step j is calculated by combining the product material properties and the stiffness matrix, and the deformation increment in the jth clamping step is also obtained. Furthermore, the workpiece deformation in the clamping step j can be obtained: for:

[0160]

[0161] Calculate the part deformation After that, the deformation of the part is decomposed into two parts: translation and rotation. After obtaining the deformation of the tooling, its deformation is decomposed into the change of the tooling positioning position and the rotation of the positioning axis. Its displacement facing the positioning error can be expressed as a 6×1 vector, as shown in the following formula.

[0162] δ={δ tu ,δ tv ,δ tw ,δ ra ,δ rβ ,δ rγ} T

[0163] Furthermore, the deformation deviation generation process caused by thin-walled parts during the assembly and connection process is analyzed, the deformation of the parts is solved, and it is expressed by the small displacement screw method. In the on-site assembly and connection process of aviation composite wall panels, bolt clearance fit connection is often used. The resulting connection force will cause stress on the surface of the parts, resulting in deformation deviation on the characteristic surface of the parts. In the calculation process of the deformation deviation caused by the connection force, since the force direction of the two parts is mainly longitudinal due to the tightening of the bolts, only the influence of the longitudinal bolt connection force on the contact deformation error of the two parts is considered. Figure 2 As shown, the solid line portion represents the mating feature surface before the parts are connected, and the dotted line portion represents the actual mating feature surface that is deformed after the connection.

[0164] To simplify the modeling and calculation process, the following assumptions are made: the thickness of the wall panel does not change during the connection process; the normal line of the mid-surface remains straight and perpendicular to the mid-surface after the wall panel is deformed; and there is no movement of the points on the mid-surface parallel to the mid-surface. Then the volume force along the Z direction is 0, that is, F VZ =0, at the same time, assuming that the axial load on the upper surface is q and the axial load on the lower surface is 0, the relationship between the connection force load q and the deflection ω can be obtained as follows:

[0165]

[0166] From the constitutive equation, we can know that the principal stress q of the wall panel in the Z direction can be expressed as:

[0167]

[0168] At the same time, according to the three assumptions mentioned above, we have: the strain of the skin panel in the Z direction ε z is 0, so the deflection of the wall panel is considered to be the deformation error μ in the Z direction. z ; During the connection process, the shear strain of the wall along the ZX direction and the ZY direction is also 0, that is, γ zx =γ zy =0; the displacements in the X and Y directions parallel to the mid-plane are also 0, that is, θ x(x,y) =θ y(x,y) = 0. On this basis, the deformation in the X and Y directions during the connection process is μ x With μ x It can be expressed as:

[0169]

[0170] Substituting the established relationship between the bolt connection force load q and the deflection ω into the above formula, the deviation of the wall panel along the X direction and the Y direction can be obtained. The deformation error distribution is superimposed with the ideal design surface shape error. It represents the actual shape error of the part after contact deformation, and converts the actual shape error into a matrix form to obtain the corrected small screw model D. ε represents the change in the position of A1 before and after the force, ρ represents the change in the posture of A1 before and after the force, and D can be represented by a 6×1 vector, as shown in the following formula.

[0171] D=[ρ' ε'] T =[α' β' γ' u' v' w'] T

[0172] Furthermore, the springback deformation deviation caused by thin-walled parts during the unloading process is analyzed, the deformation of the parts is solved, and it is expressed by the small displacement screw method. Springback refers to the phenomenon that the residual internal stress between the parts will be released due to the unloading of the fixture clamping force when the assembly on the assembly jig is unloaded after the main positioning and connection process is completed, resulting in the change of shape and size caused by elastic recovery, that is, the springback deformation error, such as Figure 3 shown.

[0173] In order to effectively solve the deformation deviation caused by part springback, it is necessary to obtain the assembly springback force F of the product. r The deformation deviation caused by springback is calculated with the stiffness matrix of the entire assembly. In the calculation process of springback deformation, the springback force F r The point of action of is the same as the point of action of the fixture clamping force. At the same time, the magnitude of the rebound force is related to the magnitude of the fixture clamping force, which can be expressed as:

[0174]

[0175] Where Q is the transfer matrix between the rebound force and the clamping force; n is the number of clamping force points on the skin panel.

[0176] After the assembly is removed from the rack, the rebound force acts on the composite wall panel. At this time, the assembly connection operation is completed, and the stiffness matrix of the assembly as a whole no longer changes and becomes a fixed value. The assembly deformation V caused by the rebound force r It can be expressed as:

[0177]

[0178] Where M is the rebound correlation matrix, that is, the rebound stiffness matrix K r The inverse matrix of .

[0179] Introducing the relationship between the clamping equivalent node force and the clamping deformation error, the stiffness matrix of the part in the tooling positioning and clamping stage is substituted into the springback assembly deformation V r The final springback deformation error can be obtained from the calculation formula:

[0180]

[0181] And the final springback assembly deformation V is obtained r Converted into matrix form, the modified small spinor model D can be obtained V ε V Indicates the change in the position of A1 before and after the force is applied, ρ V Indicates the change in the posture of A1 before and after the force is applied, D V It can be represented by a 6×1 vector as shown below.

[0182] D V =[ρ V 'ε V '] T =[α V ' β V ' γ V ' V 'v V 'ω V '] T

[0183] S2: Construction of assembly error transmission mechanism model for multiple assemblies

[0184] Specifically, first, a global coordinate system of the assembly system is created, and then a local coordinate system is created on the mating surface of each part. Each mating surface has factors such as manufacturing errors and assembly deformation errors, and the changes on the mating surface are simulated through the matrix. Starting from the global coordinate system, each mating surface is passed in turn, and the parts are constrained by the mating surfaces. The proposed method is verified by the assembly accuracy of the left straight seam gap and step difference after the product is assembled. Figure 4 shown.

[0185] Furthermore, according to the matching relationship of each part and the direction of the global coordinate system, the local coordinate system of each part and the assembly feature FE of each component are established. n , the deviation between the straight seam features of each composite skin component in the Y direction is taken as the respective gap, and the deviation in the Z direction is taken as the step difference of the straight seam. By solving the deviations of the features FE34 and FE62 in each direction, the straight seam gap FRA and step difference FRB of the assembly are calculated. According to the above assembly process, the structural error transmission relationship diagram is established, as shown in Figure 5 shown.

[0186] Furthermore, the error changes on the mating surfaces will be mutually transmitted and accumulated along the assembly chain dimensions, and finally accumulated to the end of the assembly dimension chain, and form a posture error relative to the global coordinate system, thereby establishing a unified Jacobi-screw model. For the calculation of the straight seam gap, it is identified as the deviation value of the feature F34 and the feature F61 in the Y direction. The sources of the deviation mainly include the positioning deviation of each component, the manufacturing form and position tolerance and the deformation error caused by the force load. According to the deviation FR1 and FR2 of the straight seam feature FE34 of the upper wall panel No. 1 and the straight feature FE2 of the upper wall panel FE62 in the Y direction, the straight seam gap FRA of the assembly can be obtained. Furthermore, based on the error transmission path and the matrix correction method, each deviation FR is calculated.

[0187] S3: Construction of assembly error transmission mechanism model considering initial manufacturing error of parts, part load deformation and fixture error

[0188] Specifically, a laser tracker is used to scan the positioning surfaces of each assembly tooling and the assembly features of each component to measure the displacement and shape change values ​​of each assembly feature, thereby realizing the monitoring and extraction of the deformation data of the wing box tooling and components.

[0189] Furthermore, based on the displacement and shape change values ​​of each assembly feature collected by the laser tracker, the original node coordinates of the key functional features required for tolerance analysis and the deformation of each node along the three directions of the axis of the global coordinate system are extracted. Then, the deformation of the corresponding node is superimposed with the original coordinates along the corresponding direction to obtain the actual geometric surface point set considering the deformation deviation, and the least squares method is used to fit the actually measured geometric surface data to obtain the real fitting surface.

[0190] Furthermore, each feature surface is fitted according to the actual deformation, and the position error and shape tolerance of each feature surface are adjusted based on the surface deformation data obtained by fitting. The feature surface deviation is superimposed with the ideal design surface shape tolerance to obtain the actual shape tolerance Δd and position tolerance O of the part. i' , and then the corrected assembly error can be obtained:

[0191]

[0192]

[0193] S4: Construction of assembly accuracy prediction data model based on support vector regression

[0194] Specifically, the various error sources and error transmission paths in the wing box assembly process clearly need to collect data sets related to assembly accuracy. The collected data mainly include various assembly parameters at the wing box assembly site (tolerances of machined ribs, front and rear beams, frame plates, partitions, composite skins, and positioning accuracy of tooling) and actual assembly data generated during the assembly process (straight seam gaps and step differences). At the same time, it is also necessary for on-site sensors to measure multiple sets of actual values ​​of seam gaps and step differences.

[0195] Further, the above obtained data are sorted to obtain sample data D = {(x i ,z i )}, where x i Represents the tolerance values ​​of each assembly part, the positioning accuracy of the tooling, the assembly error calculated by the mechanism model, the assembly error measured on site, etc. i The deviation value of the mechanism model calculation is obtained by subtracting the assembly error mechanism calculation result from the assembly error measured result. The product's link errors, positioning errors, mechanism calculation errors, and the deviation between the mechanism model calculation and the actual result are taken as characteristic variables (independent variables), and the target variable (dependent variable) is defined as straight seam. 75% of the sample data is used for training the support vector regression model. In order to find an optimal hyperplane in the training data to minimize the prediction error, the fitting function f(x) of the assembly accuracy prediction data model calculation deviation is defined separately:

[0196]

[0197] The kernel function of the fitting function is a radial basis function (RBF) which can form a nonlinear mapping and is easy to implement:

[0198]

[0199] During the training process, the above fitting function and kernel function are used to effectively predict the assembly accuracy.

[0200] Furthermore, the assembly accuracy prediction data model predicts the final product assembly accuracy deviation value by using the support vector regression method, namely:

[0201]

[0202] In the formula, f(x1,…x n ) is the compensation value of the assembly accuracy deviation prediction result, y m is the measured product assembly accuracy, y t (x1,…x n ) is the theoretical assembly deviation calculated by the assembly accuracy prediction data model, is the feature variation error of part feature 1, is the feature variation error of part feature 1. Part feature 1 and part feature 2 together constitute the assembly accuracy error.

[0203] S5: Construction of product assembly accuracy prediction model integrating mechanism and data

[0204] Specifically, the data model is taken as the main body and the mechanism model is taken as the auxiliary. The mechanism model is integrated into the feature level of the data model, thereby realizing the model construction of the fusion of data and mechanism. The idea of ​​constructing the product assembly accuracy prediction model integrating mechanism and data is as follows: Figure 6 First, the actual manufacturing accuracy and deformation deviation values ​​of the product ribs, beams, walls, partitions, upper and lower wall panels are measured and collected. After the wing box product is assembled, the assembly gap and assembly step difference accuracy data of the left straight seam are measured. The above-collected data sets are used as training samples to train the constructed assembly accuracy prediction data model based on support vector regression.

[0205] Furthermore, the design tolerance information of the ribs, beams, walls, partitions, upper and lower wall panels and other components of the wing box product is input into the established assembly error transmission mechanism model, and the straight seam gap error and step difference error of the wing box product are calculated respectively. The straight seam gap and step difference measured accuracy data, assembly process data, simulation data and error data generated during the wing box assembly process are read and substituted into the established assembly accuracy prediction data model to calculate the mechanism calculation deviation prediction values ​​of the straight seam gap and step difference of the product.

[0206] Furthermore, the predicted value of the mechanism calculation deviation of the straight seam gap and step difference of the calculated product is added to the corresponding gap and step difference mechanism calculation value to complete the correction of the theoretical calculation value of the mechanism model, and the compensated wing box straight seam gap and compensated assembly accuracy calculation value are obtained.

[0207] Finally, real physical assembly process data is introduced, and a mechanism calculation deviation prediction model of data-mechanism fusion is obtained through training with multiple sets of sample data. The mechanism calculation model is compensated to improve the calculation accuracy, and the assembly accuracy prediction model of the fusion mechanism and data is completed to form an assembly accuracy prediction method of the hybrid model.

[0208] S6: Validation of the hybrid drive assembly accuracy prediction model

[0209] Specifically, the real physical assembly process data is introduced, and the compensation prediction models of the deviation features FR1 and FR2 are trained respectively through multiple sets of sample data. The model is verified using the assembly parameters and actual assembly data of the wing box assembly site, and the positioning accuracy and tolerance data of each assembly component that constitutes FR are substituted into the assembly mechanism model to calculate the theoretical calculation value ET of the assembly error. At the same time, according to the multiple sets of gap values ​​measured by the on-site sensors, the straight gap calculation process under the mechanism-data hybrid drive is as follows: Figure 7 Afterwards, 70% of the samples in the multiple groups are used as training sample sets to train the prediction model, and the remaining 30% of the samples are used as test sample sets to verify the prediction accuracy of the model.

[0210] Furthermore, the prediction model is constructed based on the Matlab platform, taking various assembly errors and measured values ​​as inputs, and the deviation between the theoretical calculated error value and the measured error value as output. The product's various link errors, positioning errors, mechanism calculation errors ET, and the deviation EE between the mechanism model calculation and the actual result are taken as characteristic variables (independent variables), and the target variables (dependent variables) are defined as the product assembly accuracy. The kernel type, penalty parameters, and kernel function are determined, and the SVM train function is used to train the regression model. After the training is completed, the accuracy of the prediction model is verified based on the test samples, and the model-related prediction error data is calculated, as shown in the following table. The calculation process, model accuracy verification, and related data for the straight-to-seam gap are as follows: Figure 8 , 9 shown.

[0211] Furthermore, after calculation, the average absolute error MAE value of the trained prediction model is 0.052176, the mean square error MSE value is 0.007424, the root mean square error RMSE value is 0.079945, and the absolute coefficient R2 value is 0.9854, indicating that the obtained training model has high accuracy and reliability.

[0212] The above is a detailed introduction to a product assembly accuracy prediction method and system driven by a mechanism and data fusion provided in an embodiment of the present application. The present invention adopts a deformed shape tolerance correction error transfer model by considering the deformation caused by the contact between parts and the deformation error factors caused by the contact between parts and tooling. It not only accurately constructs an assembly error transmission mechanism model, but also can read the field measured data set containing various assembly parameters and assembly accuracy, organically integrate the mechanism model with the data, and then construct a data-mechanism fusion mechanism calculation error prediction model, which quickly and effectively improves the assembly accuracy prediction results, and is conducive to the rapid and accurate assembly of complex products on site. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for those skilled in the art, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

[0213] For example, certain words are used in the specification and claims to refer to specific components. Those skilled in the art should understand that hardware manufacturers may use different nouns to refer to the same component. This specification and claims do not use differences in names as a way to distinguish components, but use differences in the functions of components as the criteria for distinction. As mentioned throughout the specification and claims, "including" and "comprising" are open-ended terms, so they should be interpreted as "including / including but not limited to". "Approximately" means that within an acceptable error range, those skilled in the art can solve the technical problem within a certain error range and basically achieve the technical effect. The subsequent description of the specification is a preferred embodiment of the present application, but the description is for the purpose of illustrating the general principles of the present application, and is not used to limit the scope of the present application. The scope of protection of the present application shall be determined by the definition of the attached claims.

[0214] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a product or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such a product or system. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the product or system including the elements.

[0215] It should be understood that the term "and / or" used in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0216] The above description shows and describes several preferred embodiments of the present application, but as mentioned above, it should be understood that the present application is not limited to the form disclosed herein, and should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications and environments, and can be modified within the scope of the application concept described herein through the above teachings or the technology or knowledge of the relevant field. The changes and modifications made by those skilled in the art do not deviate from the spirit and scope of the present application, and should be within the scope of protection of the claims attached to the present application.

Claims

1. A product assembly accuracy prediction method driven by mechanism and data fusion, the assembly accuracy prediction method is used to predict the assembly accuracy of composite wing box assembly parts during the production process, characterized in that: The assembly accuracy prediction method comprises the following steps: S1: According to the structural characteristics and assembly process of the parts to be produced, a geometric error screw model with characteristic changes of multiple basic error sources is constructed; S2: According to the matching relationship between the parts to be assembled and the positioning of the tooling, the assembly error transmission mechanism model of multiple assemblies is constructed; S3: Based on the geometric error screw model, assembly deformation error and measured assembly error, the assembly error transmission mechanism model is corrected to obtain the assembly error transmission correction model; S4: acquiring parameter data based on the assembly error transfer correction model, constructing an assembly accuracy prediction data model based on support vector regression, optimizing the assembly accuracy prediction data model through the parameter data, and obtaining an assembly accuracy prediction data optimization model; S5: Construct a product assembly accuracy prediction model through the assembly error transmission mechanism model, the assembly accuracy prediction data optimization model and the measured assembly accuracy data; S6: Train the product assembly accuracy prediction model through measured assembly accuracy data and verify its effectiveness.

2. The assembly accuracy prediction method according to claim 1, characterized in that: The S1 specifically includes: S11: Analyze the assembly product structure and assembly performance requirements, obtain specific product assembly accuracy requirements, and determine the various error sources in the product assembly process based on the assembly accuracy requirements, and convert the impact of the error sources on the assembly into corresponding feature changes; S12: The position change data of the geometric features of the assembly are obtained based on the key measurement points and geometric feature information of each error source. The kinematic theory and small displacement screw method are used to convert the position change data into an error screw model in a matrix format, so as to realize the construction of a geometric error screw model for the feature changes of various basic error sources.

3. The assembly accuracy prediction method according to claim 2, characterized in that: The error sources in S11 include but are not limited to initial manufacturing deviation of parts, deformation deviation of thin-walled parts caused by the clamping force of assembly tooling, deformation deviation caused during assembly connection, and springback deformation deviation caused during the unloading process, wherein; The deformation deviation of thin-walled parts caused by the clamping force of assembly tooling is specifically as follows: the deformation deviation generation mechanism of thin-walled parts caused by the positioning and clamping of assembly tooling is analyzed, the deformation of the parts is solved, and the contact force of the parts is calculated using the minimum total residual energy as the objective function under the condition of satisfying the constraints of the tooling positioning position and the unilateral contact constraints between the workpiece and the fixture, and the deformation of the parts is obtained by combining the stiffness matrix of the parts; The deformation deviation caused by thin-walled parts in the assembly and connection process is specifically as follows: the mechanism of part contact deformation deviation caused by thin-walled parts in the assembly and connection process is analyzed, and the deviation of the wall panel along the X direction and the Y direction is obtained by establishing the relationship between the bolt connection force load and the deflection. The deviation of the wall panel along the X direction and the Y direction is superimposed with the ideal design surface shape error to obtain the change of the actual shape of the part after contact deformation; The springback deformation deviation caused by the removal process is specifically as follows: the mechanism of the springback deformation deviation of the parts caused by the removal of the assembly after assembly is completed is analyzed, and the relationship between the springback force and the clamping force of the tooling and the assembly deformation caused by the elastic force is analyzed by obtaining the assembly springback force of the product and the stiffness matrix of the entire assembly, and the deformation deviation caused by the springback is calculated and solved.

4. The assembly accuracy prediction method according to claim 1, characterized in that: The S2 specifically includes: S21: Consider the matching relationship of the characteristic surfaces of each component and the tooling positioning method according to the assembly sequence, and establish multiple sets of local coordinate systems and posture / tolerance constraints; S22: The posture error conversion matrix is ​​obtained by combining the Jacobian matrix with the small displacement screw model. This matrix can describe the transmission relationship between the positioning tooling, the inside of the part and the mating surface of each error link in the product assembly process, thereby realizing the construction of the assembly error transmission mechanism model.

5. The assembly accuracy prediction method according to claim 1, characterized in that: The S3 specifically includes: S31: real-time acquisition of test data obtained by sensors pre-arranged on the assembly structure, and real-time and accurate acquisition of the deformation of the wing box during the assembly process by a laser displacement measuring instrument; S32: Collect data through a laser tracker, extract the original node coordinates of key functional features and the deformation of each node along the three directions of the global coordinate system, superimpose the node deformation with the original coordinates along the corresponding direction, obtain the actual geometric surface point set considering the deformation deviation, and use the least squares method to fit the actually measured geometric surface data to obtain the real fitting surface; S33: Describe the influence of deformation of functional features of parts on tolerance in mathematical form, use the fitting surface method to describe the functional features of products, comprehensively consider the influence of component deformation during assembly, correct the assembly error transmission mechanism model, and obtain the assembly error transmission correction model.

6. The assembly accuracy prediction method according to claim 5, characterized in that: The S4 specifically includes: S41: According to the structure of the assembly error transfer correction model, determine the data model collection requirements and perform data collection. The data collection content is a data set related to assembly accuracy, including but not limited to parameter information and workpiece quality information in the assembly process; S42: preprocessing the data collected by the data model to remove outliers, and then using a support vector regression method to train the assembly accuracy prediction data model. During the training process, a suitable fitting function and kernel function are selected to fit the data so that the assembly accuracy can be effectively predicted; S43: The assembly accuracy prediction data model predicts the final product assembly deviation value by using the support vector regression method, and uses this deviation value to compensate the calculation result of the assembly error transmission mechanism model to obtain the actual calculation accuracy of the assembly.

7. The assembly accuracy prediction method according to claim 6, characterized in that: The S5 specifically includes: S51: reading various parameters of product assembly, and inputting various assembly parameters into the product assembly error transmission mechanism model to obtain the theoretical calculation value of the mechanism model, and at the same time, subtracting the assembly accuracy measured data generated at the measurement collection site from the theoretical calculation value of the mechanism model to obtain the calculated deviation value of the mechanism model assembly error; S52: Based on the assembly accuracy measured data, assembly process data, related simulation data and error data as training samples, an assembly accuracy prediction data model based on support vector regression is constructed, and the theoretical calculation value is used as the training input of the model, and the prediction value of the deviation value calculated by the assembly error transmission mechanism model is obtained through the data model; S53: Add the deviation prediction value calculated by the assembly accuracy prediction data model and the calculated deviation value of the mechanism model assembly error calculated by the assembly error transfer mechanism model to obtain the compensated assembly accuracy calculation value, and obtain the product assembly accuracy prediction model that integrates the mechanism and data.

8. The assembly accuracy prediction method according to claim 1, characterized in that: The S6 specifically includes: S61: Based on the engineering digital model provided by the design department, read the assembly process plan of the product, determine the assembly sequence of parts, the clamping plan and the matching relationship between parts, and collect the real physical assembly process data and the assembly parameters of the wing box assembly site, and organize the obtained data to obtain sample data for model verification; S62: Using various assembly errors and measured values ​​as input, and the deviation between the theoretical calculated error value and the measured value as output, the product assembly accuracy prediction model integrating mechanism and data is trained based on the Matlab platform; S63: Verify the validity of the prediction results of the product assembly accuracy prediction model based on the training results of the product assembly accuracy prediction model that integrates the mechanism and data.

9. The assembly accuracy prediction method according to claim 8, characterized in that: The contents of the verification effectiveness in S63 include but are not limited to the mean absolute error MAE value, mean square error MSE value, root mean square error RMSE value and absolute coefficient R2 value of the prediction model.

10. A product assembly accuracy prediction system driven by mechanism and data fusion, the assembly accuracy prediction system is used to predict the assembly accuracy of composite wing box assembly parts during the production process, characterized in that: The assembly accuracy prediction system comprises: A geometric error screw model building module is used to build a geometric error screw model with characteristic changes of multiple basic error sources according to the structural characteristics and assembly process of the parts to be produced; Assembly error transmission mechanism model construction module: used to construct the assembly error transmission mechanism model of multiple assemblies based on the matching relationship between the parts to be assembled and the positioning of the tooling; An assembly error transmission mechanism model correction module is used to correct the assembly error transmission mechanism model based on the geometric error screw model, the assembly deformation error and the measured assembly error to obtain the assembly error transmission correction model; An assembly accuracy prediction data optimization model construction optimization module is used to obtain parameter data based on the assembly error transfer correction model, construct an assembly accuracy prediction data model based on support vector regression, and optimize the assembly accuracy prediction data model through parameter data to obtain an assembly accuracy prediction data optimization model; A product assembly accuracy prediction model fusion construction module is used to construct a product assembly accuracy prediction model through an assembly error transmission mechanism model, an assembly accuracy prediction data optimization model, and measured assembly accuracy data; The product assembly accuracy prediction model training and verification module is used to train the product assembly accuracy prediction model through measured assembly accuracy data and verify its effectiveness.

Images