Furniture design parameterization modeling method and system

Abstract

The present application relates to the technical field of furniture digital design and intelligent manufacturing, and particularly relates to a furniture design parameterized modeling method and system. After obtaining the user design requirements, the method calls the basic structure template from the preset process knowledge base and determines the target machining feature set to generate an initial structure model; according to the assembly constraint rules, the machining parameter group and the historical manufacturing data, a process constraint field is constructed, the response coefficient matrix of the design variables of each geometric part to the process constraint characteristic value of the associated machining feature is extracted, the process projection aggregation is carried out along the assembly direction to generate the process offset vector and the process response matrix; according to the process offset vector and the process response matrix, the structure domain compensation amount and the process domain compensation amount are respectively calculated, the geometric skeleton data and the machining parameter group are respectively adjusted, and a parameterized driving model is generated. The present application realizes the quantitative evaluation and adaptive correction of the manufacturing feasibility in the design stage.

Description

Technical Field

[0001] This invention relates to the field of digital design and intelligent manufacturing technology for furniture, and in particular to a parametric modeling method and system for furniture design. Background Technology

[0002] With the deep integration of digital design and manufacturing, parametric modeling technology has become a core tool in furniture product development. Traditional parametric modeling methods mainly rely on the built-in parametric engines of 3D CAD software, achieving model variation design through dimension-driven and geometric constraints. In recent years, some companies have begun to incorporate manufacturing knowledge into parametric modeling, attempting to link information such as processing parameters, material properties, and equipment capabilities with geometric models in the form of external databases, aiming to consider manufacturing feasibility during the design phase. Furthermore, process optimization methods based on historical manufacturing data have seen initial applications in the machining field, providing process parameter references for new products through statistical analysis of past processing cases.

[0003] However, in existing parametric modeling methods, the adjustment of design parameters and the assessment of manufacturing feasibility are two separate decision-making stages. After designers complete the modification of the geometric model, process engineers analyze its manufacturability offline, resulting in long design iteration cycles and delayed scheme verification. A more fundamental flaw lies in the lack of a technical means to simultaneously perceive the sensitivity of design variables to process constraints and quantify the direction and degree of deviation of the current design state from the optimal manufacturing path. Therefore, it is impossible to simultaneously calculate the required linkage compensation amounts for the geometric structural domain and the processing technology domain during the design parameter adjustment process to achieve autonomous correction of the design scheme. This leads to the parametric modeling process always relying on human experience in a "trial and error-verification-modification" cycle, making it difficult to proactively avoid manufacturing risks during the design phase. Furthermore, existing conventional furniture modeling methods lack a complete twin modeling carrier, making it impossible to build a digital twin that maps one-to-one with physical furniture components, and difficult to simultaneously carry multi-dimensional data on geometric structure, processing technology, and on-site manufacturing feedback.

[0004] Therefore, this invention proposes a parametric modeling method and system for furniture design. Summary of the Invention

[0005] This invention provides a parametric modeling method and system for furniture design. By constructing a process constraint field and generating a process offset vector and a process response matrix, it realizes autonomous diagnosis of manufacturing deviation direction and linkage decision-making of compensation amounts between the structural domain and the process domain during the design stage.

[0006] This invention provides a parametric modeling method for furniture design, comprising: Obtain design requirements information input by the user, which includes appearance parameters and functional descriptions. The basic structure template matching the appearance parameters is called from the preset process knowledge base, and the target processing feature set is determined from the preset process knowledge base according to the functional description information. The processing features in the target processing feature set are associated with the geometric parts of the basic structure template to generate an initial structural model. The basic structure template contains geometric skeleton data and assembly constraint rules. Based on the assembly constraint rules, the processing parameter set of the target processing feature set, and the historical manufacturing data in the preset process knowledge base, the process constraint field of the initial structural model is constructed. Each spatial coordinate point of the process constraint field corresponds to a process constraint feature value, which includes the processing difficulty coefficient, deformation risk coefficient, and tolerance achievement coefficient. In the process constraint field, the response coefficient matrix of the design variables of each geometric part to the process constraint feature value of the associated processing feature is extracted. The response coefficient matrix is ​​aggregated by process projection along the assembly direction of the initial structural model to generate the process offset vector and process response matrix of the initial structural model. The process offset vector is used to characterize the offset direction and degree of the current design state from the optimal manufacturing path, and the process response matrix is ​​used to characterize the response sensitivity of the offset direction to each dimension of the design variable. The structural domain compensation and the process domain compensation are calculated based on the process offset vector and the process response matrix, respectively. The geometric skeleton data is adjusted based on the structural domain compensation, and the processing parameter set is adjusted based on the process domain compensation to generate a parameterized driving model.

[0007] Furthermore, based on assembly constraint rules, the processing parameter set of the target processing feature set, and historical manufacturing data in the preset process knowledge base, a process constraint field for the initial structural model is constructed, including: The assembly topology features of the initial structural model are determined according to the assembly constraint rules. The current processing environment parameters are determined according to the processing parameter set of the target processing feature set. Multiple historical manufacturing data samples with topological homeomorphism to the geometric skeleton data of the initial structural model and whose historical processing parameter sets match the current processing environment parameters are obtained from the preset process knowledge base. Each historical manufacturing data sample contains a historical geometric coordinate set, a historical processing parameter set, and a corresponding historical manufacturing quality label. For each historical manufacturing data sample, the historical geometric coordinate set is mapped to the coordinate system of the geometric skeleton data of the initial structural model to obtain multiple registered historical process feature points. Each historical process feature point contains spatial coordinate values, corresponding historical processing parameter values, and corresponding historical manufacturing quality labels such as processing difficulty label value, deformation risk label value, and tolerance achievement label value. Using the spatial coordinates of each historical process feature point as the independent variable, and the corresponding processing difficulty label value, deformation risk label value, and tolerance achievement label value as the dependent variable, we construct the initial interpolation function for processing difficulty, the initial interpolation function for deformation risk, and the initial interpolation function for tolerance achievement, respectively. Calculate the first consistency coefficient between the local gradient vector of the initial interpolation function for processing difficulty and the local gradient vector of the initial interpolation function for deformation risk at each historical process feature point, and the second consistency coefficient between the local gradient vector of the initial interpolation function for deformation risk and the local gradient vector of the initial interpolation function for tolerance achievement. When the first consistency coefficient or the second consistency coefficient is less than the preset consistency threshold, a virtual feature point is inserted at the corresponding historical process feature point. The spatial coordinates of the virtual feature point are the coordinates after applying a preset offset to the spatial coordinates of the corresponding historical process feature point. When the first consistency coefficient is less than the preset consistency threshold, the dependent variable of the virtual feature point is the sum of the dependent variable of the corresponding historical process feature point and the preset adjustment amount. When the second consistency coefficient is less than the preset consistency threshold, the dependent variable of the virtual feature point is the difference between the dependent variable of the corresponding historical process feature point and the preset adjustment amount. Based on all historical process feature points and all virtual feature points, the initial interpolation function for processing difficulty, the initial interpolation function for deformation risk, and the initial interpolation function for tolerance achievement are refitted using the radial basis function interpolation method until the first consistency coefficient and the second consistency coefficient at all historical process feature points are not less than the preset consistency threshold. Then, the three refitted interpolation functions are combined into a process constraint field.

[0008] Furthermore, the response coefficient matrix of the design variables for each geometric part to the process constraint eigenvalues ​​of the associated machining features is extracted from the process constraint field, including: Determine the design variable space for each geometric part in the initial structural model, and determine the distribution of process constraint characteristic values ​​of the associated processing features of each geometric part in the process constraint field; Taking the current coordinate point in the design variable space of each geometric part as the starting node, and the adjacent geometric parts defined in the assembly constraint rules of the basic structure template as the propagation nodes, the perturbation propagation path of the design variable perturbation of the geometric part in the process constraint field is constructed step by step along the assembly direction of the assembly constraint rules. Each perturbation propagation path passes through the process constraint feature value distribution of the associated processing feature of at least one intermediate geometric part. The cumulative effect of the change in process constraint characteristic value caused by a unit disturbance of the design variable on each disturbance propagation path is calculated along the path, and the ratio of the cumulative effect to the path length is used as the path response coefficient of the disturbance propagation path. Calculate the deviation ratio between the path response coefficient of each disturbance propagation path and the historical response sensitivity of the intermediate geometric parts traversed by the disturbance propagation path. After normalizing all deviation ratios, calculate the complement value as the propagation validity of each disturbance propagation path. The complement value is 1 minus the normalized deviation ratio. Multiple disturbance propagation paths with a propagation validity greater than a preset validity threshold are selected. At the starting point of each selected disturbance propagation path, a unit disturbance is applied along the dimension of size, dimension of angle, and dimension of position in the design variable space. The cumulative response vector of the unit disturbance along the entire disturbance propagation path is calculated. The cumulative response vectors of all selected disturbance propagation paths are weighted and aggregated according to the propagation validity of each path to generate a response coefficient matrix of the geometric design variables to the process constraint eigenvalues ​​of the associated processing features.

[0009] Furthermore, the response coefficient matrix is ​​aggregated by process projection along the assembly direction of the initial structural model to generate the process offset vector and process response matrix of the initial structural model, including: Determine the local assembly direction vector for each geometric part in the initial structural model. The local assembly direction vector is the unit vector of the geometric part in the assembly mating direction defined in the assembly constraint rules. The response coefficient matrix of each geometric part is projected and aggregated with the corresponding local assembly direction vector to obtain the local process offset component of each geometric part. The projection and aggregation operation is as follows: the local assembly direction vector is transformed to the design variable space through the Jacobian matrix of the geometric skeleton to obtain the design variable direction vector. The dot product of each column vector of the response coefficient matrix and the design variable direction vector is calculated to obtain the directional derivative of the corresponding process constraint eigenvalue, which is used as the corresponding component of the local process offset component of the current geometric part. Among them, the row vector of the response coefficient matrix corresponds to the response of each design variable dimension of the current geometric part to the process constraint eigenvalue, and the design variable direction vector represents the degree of influence of the assembly direction on each design variable dimension. The local process offset components of all geometric parts are weighted and aggregated to obtain the process offset vector of the initial structural model. Calculate the Jacobian matrix of the local process offset component of each geometric part relative to the geometric skeleton data of the current geometric part, and combine the Jacobian matrices of all geometric parts into a process response matrix.

[0010] Furthermore, the structural domain compensation and the process domain compensation are calculated based on the process offset vector and the process response matrix, respectively, including: Determine the unit direction vector of the process offset vector; Based on the eigenvalues ​​and column vectors of the process response matrix, determine the projection coefficients of the unit direction vector in each design variable dimension; Based on the projection coefficient and the modulus of the process offset vector, calculate the structural domain compensation for each geometric part and the process domain compensation for each processing feature.

[0011] Furthermore, after generating the parameterized driving model, the following is also included: Obtain the change sequence of the process offset vector and process response matrix of the parameterized driving model within a preset number of iterations, and calculate the directional offset and modulus attenuation of the process offset vector and the change of the condition number of the process response matrix based on the change sequence. When the directional offset exceeds a preset offset threshold, the modulus attenuation exceeds a preset attenuation threshold, or the change in the condition number exceeds a preset condition number change threshold, an early warning is triggered. When a condition is triggered, the calculation and adjustment of the structural domain compensation and process domain compensation of the parameterized driven model are repeated until the recalculated directional offset, modulus attenuation, and change in the condition number are less than or equal to the corresponding preset thresholds, and the iterative optimization is completed.

[0012] Furthermore, the construction methods for the pre-defined process knowledge base include: We acquired multiple closed-loop cases of furniture design and manufacturing. Each closed-loop case includes a set of design parameters, structural data, a set of processing parameters, and corresponding manufacturing feedback data. For each closed-loop case, a theoretical structural model is constructed based on the design parameter set and structural data. As the digital twin of the physical entity of the closed-loop case, a process constraint field associated with the theoretical structural model is constructed on the geometric shape of the digital twin based on the processing parameter set and manufacturing feedback data. In the process constraint field, the corresponding response coefficient matrix is ​​extracted according to the method of screening the disturbance propagation path and calculating the propagation validity. The response coefficient matrix is ​​then projected and aggregated along the assembly direction to generate the corresponding process offset vector and process response matrix. At the same time, the propagation validity is recorded as a path confidence label. The process fingerprint data of each closed-loop case is formed by combining the unit direction vector and magnitude of the process offset vector, the eigenvalue sequence and eigenvector matrix of the process response matrix, and the corresponding path confidence label. The process fingerprint data is associated with and stored with the corresponding closed-loop cases to form a pre-defined process knowledge base.

[0013] Furthermore, after generating the parameterized driving model, the following is also included: Calculate the current process fingerprint data of the parameterized driving model and extract the current path confidence label of the disturbance propagation path of the current parameterized driving model; The current process fingerprint data is compared with the historical process fingerprint data stored in the preset process knowledge base. The comprehensive similarity matching includes the unit direction vector similarity matching and modulus similarity matching of the process offset vector, the cosine similarity matching of the eigenvalue sequence of the process response matrix and the F norm similarity matching of the eigenvector matrix, as well as the label similarity matching between the current path confidence label and the historical path confidence label. When historical process fingerprint data with a comprehensive similarity greater than a preset similarity threshold is matched, the structural data and processing parameter set in the closed-loop case associated with the historical process fingerprint data are retrieved from the preset process knowledge base. Based on the structural data and the set of processing parameters, the parameterized driving model is optimized a second time.

[0014] Furthermore, after generating the parameterized driving model, the following is also included: Based on the overall size data of the parameterized driving model and the target machining feature set, the blanking data and machining path data of multiple machining elements are generated; Based on the assembly constraint rules, generate the assembly sequence and assembly positioning data for each machined component; The material cutting data, processing path data, assembly sequence and assembly positioning data are combined into a production guidance data package, and the production guidance data package is output to the CNC machining equipment. Receive actual machining deviation data from CNC machining equipment, update the process constraint field based on the actual machining deviation data, recalculate the updated process offset vector and process response matrix, and store the updated process offset vector and process response matrix in the preset process knowledge base.

[0015] This invention provides a parametric modeling system for furniture design, comprising: The requirement elicitation module is used to obtain design requirement information input by the user. The design requirement information includes appearance parameters and functional description information. The template and feature processing module is used to call the basic structure template that matches the appearance parameters from the preset process knowledge base, and determine the target processing feature set from the preset process knowledge base according to the functional description information. The processing features in the target processing feature set are associated with the geometric parts of the basic structure template to generate an initial structural model. The basic structure template contains geometric skeleton data and assembly constraint rules. The process constraint field construction module is used to construct the process constraint field of the initial structural model based on the assembly constraint rules, the processing parameter set of the target processing feature set, and the historical manufacturing data in the preset process knowledge base. Each spatial coordinate point of the process constraint field corresponds to a process constraint feature value, which includes the processing difficulty coefficient, deformation risk coefficient, and tolerance achievement coefficient. The process projection aggregation module is used to extract the response coefficient matrix of the design variables of each geometric part to the process constraint feature values ​​of the associated processing features in the process constraint field. The response coefficient matrix is ​​then projected and aggregated along the assembly direction of the initial structural model to generate the process offset vector and process response matrix of the initial structural model. The process offset vector is used to characterize the offset direction and degree of the current design state from the optimal manufacturing path, and the process response matrix is ​​used to characterize the response sensitivity of the offset direction to each dimension of the design variables. The compensation and generation module is used to calculate the structural domain compensation amount and the process domain compensation amount based on the process offset vector and the process response matrix, respectively. It adjusts the geometric skeleton data based on the structural domain compensation amount and adjusts the processing parameter group based on the process domain compensation amount to generate a parameterized driving model.

[0016] The beneficial effects of this invention compared to existing technologies are as follows: By constructing a continuous process constraint field with spatial coordinates as independent variables, discrete historical manufacturing experience is transformed into a computable distribution of process constraint characteristic values. Based on this, the response coefficient matrix of design variables to process constraints is extracted, and process projection aggregation is performed along the assembly direction to generate a process offset vector that simultaneously characterizes the offset direction and degree, as well as a process response matrix that characterizes the sensitivity of this offset direction to the response dimensions of each design variable. This achieves the technical effect of simultaneously quantifying and evaluating the degree of deviation of the current scheme from the optimal manufacturing path and automatically calculating the linkage compensation between the structural domain and the process domain during the adjustment of design parameters. This method transforms the evaluation and correction of manufacturing feasibility from the traditional serial "design-verification-modification" mode to real-time closed-loop control within the parameter-driven model, significantly shortening the design iteration cycle and improving the first-time manufacturing success rate of the design scheme. Furthermore, this invention fully implements twin modeling logic, constructing a one-to-one digital twin for each furniture entity, and opening up a unified channel for carrying geometric, process, and on-site processing data.

[0017] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in this application.

[0018] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0019] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of the parametric modeling method for furniture design in an embodiment of the present invention; Figure 2 This is a flowchart illustrating the process constraint field construction in an embodiment of the present invention. Detailed Implementation

[0020] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0021] refer to Figure 1 This invention provides an embodiment of a parametric modeling method for furniture design, comprising: Obtain design requirements information input by the user, which includes appearance parameters and functional descriptions. The basic structure template matching the appearance parameters is called from the preset process knowledge base, and the target processing feature set is determined from the preset process knowledge base according to the functional description information. The processing features in the target processing feature set are associated with the geometric parts of the basic structure template to generate an initial structural model. The basic structure template contains geometric skeleton data and assembly constraint rules. Based on the assembly constraint rules, the processing parameter set of the target processing feature set, and the historical manufacturing data in the preset process knowledge base, the process constraint field of the initial structural model is constructed. Each spatial coordinate point of the process constraint field corresponds to a process constraint feature value, which includes the processing difficulty coefficient, deformation risk coefficient, and tolerance achievement coefficient. In the process constraint field, the response coefficient matrix of the design variables of each geometric part to the process constraint feature value of the associated processing feature is extracted. The response coefficient matrix is ​​aggregated by process projection along the assembly direction of the initial structural model to generate the process offset vector and process response matrix of the initial structural model. The process offset vector is used to characterize the offset direction and degree of the current design state from the optimal manufacturing path, and the process response matrix is ​​used to characterize the response sensitivity of the offset direction to each dimension of the design variable. The structural domain compensation and the process domain compensation are calculated based on the process offset vector and the process response matrix, respectively. The geometric skeleton data is adjusted based on the structural domain compensation, and the processing parameter set is adjusted based on the process domain compensation to generate a parameterized driving model.

[0022] In this embodiment, appearance parameters refer to quantifiable indicators input by the user to describe the external form of the furniture product, including the overall length, overall width, overall height, panel thickness, panel curvature radius, chamfer size, joint gap value, and relative positional dimensions between components.

[0023] In this embodiment, the functional description information refers to the information input by the user to describe the intended use and usage scenarios of the furniture product, including furniture type, load-bearing weight requirements, ambient humidity range, ambient temperature range, expected service life, whether dynamic loads are involved, whether outdoor use is involved, and whether repeated disassembly and assembly are involved.

[0024] In this embodiment, the preset process knowledge base is a structured dataset that is pre-built and stored in a non-volatile storage medium, used to provide basic structure templates, target processing feature sets, processing parameter sets, historical manufacturing data, and structural data and processing parameter sets.

[0025] In this embodiment, calling the basic structure template that matches the appearance parameters from the preset process knowledge base means that the system uses the appearance parameters input by the user as search conditions, traverses all stored basic structure templates in the preset process knowledge base, determines the basic structure template that falls within the range of each appearance parameter input by the user as the matching basic structure template, and calls it to the current modeling session.

[0026] In this embodiment, determining the target processing feature set from the preset process knowledge base based on the functional description information means that the system calculates the semantic similarity between the functional description information input by the user and the functional description tags of each preset processing feature set in the preset process knowledge base, and determines the preset processing feature set with the highest semantic similarity as the target processing feature set.

[0027] In this embodiment, the target processing feature set refers to a set of multiple processing features obtained by matching from a preset process knowledge base based on functional description information. The processing features include mortise and tenon connection features, screw installation features, hinge installation features, slide rail installation features, shelf support features, back panel slot features, edge banding features, and surface coating features.

[0028] In this embodiment, associating the processing features in the target processing feature set with the geometric parts of the basic structure template to generate the initial structural model means that the system matches the corresponding target geometric parts from multiple geometric part identifiers of the basic structure template according to the feature type of each processing feature, establishes a mapping relationship between the processing features and the target geometric parts, and positions each processing feature on the corresponding geometric part.

[0029] In this embodiment, the geometric skeleton data refers to the parameterized dataset in the basic structure template used to define the overall structural framework and spatial positional relationships of the furniture product. It includes the identifiers of multiple geometric parts, the coordinate range of each geometric part in three-dimensional space, the reference point coordinates of each geometric part, the local coordinate system direction of each geometric part, and the topological connection relationships between each geometric part.

[0030] In this embodiment, assembly constraint rules refer to the set of parameterized rules defined in the basic structure template to constrain the relative positions and mating relationships between various geometric parts, including coplanar constraint rules, coaxial constraint rules, parallel constraint rules, perpendicular constraint rules, distance constraint rules, angle constraint rules, and alignment constraint rules.

[0031] In this embodiment, the machining parameter group of the target machining feature set refers to the set of process parameters corresponding to each machining feature in the target machining feature set, including tool type parameters, feed rate parameters, spindle speed parameters, depth of cut parameters, machining allowance parameters, cooling method parameters, and machining sequence parameters.

[0032] In this embodiment, the historical manufacturing data in the preset process knowledge base refers to the actual manufacturing process record data stored in the preset process knowledge base, which comes from past furniture design and manufacturing closed-loop cases. It includes actual processing parameter values, actual processing time, actual material removal amount, actual surface roughness, actual dimensional tolerance, actual processing defect type and defect distribution location, and actual processing pass rate.

[0033] In this embodiment, the processing difficulty coefficient is a quantitative indicator ranging from 0 to 1, with a larger value indicating higher processing difficulty; the deformation risk coefficient is a quantitative indicator ranging from 0 to 1, with a larger value indicating higher deformation risk; and the tolerance achievement coefficient is a quantitative indicator ranging from 0 to 1, with a larger value indicating a higher probability of achieving the tolerance requirements. These three coefficients together constitute the process constraint characteristic value at each spatial coordinate point in the process constraint field.

[0034] In this embodiment, the design variables for the geometric parts refer to the variable geometric parameters of each geometric part, including dimensional parameters, angular parameters, and positional parameters. Dimensional parameters include length, width, height, thickness, diameter, and radius; angular parameters include tilt angle, rotation angle, and pitch angle; positional parameters include the X, Y, and Z coordinate values ​​of the reference point in the global coordinate system.

[0035] In this embodiment, associated processing features refer to processing features in the target processing feature set that are assigned to a certain geometric part through a mapping relationship. Each geometric part is associated with at least one processing feature, and each processing feature is associated with at least one geometric part.

[0036] In this embodiment, the assembly direction of the initial structural model refers to the spatial direction vector used by each geometric part when moving from the initial position to the final assembly position as defined in the assembly constraint rules. It includes the X-axis direction component, the Y-axis direction component and the Z-axis direction component, and the module length is 1.

[0037] In this embodiment, the process offset vector characterizes the direction and degree of deviation of the current design state from the optimal manufacturing path. It is a three-dimensional spatial vector, with the direction indicating the direction to be adjusted and the mold length indicating the distance of deviation. A mold length of 0 indicates that the optimal manufacturing path has been reached, while a mold length greater than 0 requires compensation adjustment based on the direction.

[0038] In this embodiment, the current design state refers to the combination of design variables of all geometric parts in the initial structural model at the current moment, as well as the geometric shape, spatial position, topological relationship and assembly relationship of each geometric part determined by the combination of values.

[0039] In this embodiment, the optimal manufacturing path refers to the set of design variable values ​​that enable all processing features of the initial structural model to complete manufacturing with the highest processing efficiency, lowest processing defect rate, and best processing quality under the current manufacturing environment conditions. At this time, the directional derivatives of each process feature along the assembly direction are all zero, that is, the process constraint feature values ​​reach extreme values ​​in the assembly direction, and the corresponding process offset vector is a zero vector, which is implicitly defined by the distribution law of the process constraint field.

[0040] In this embodiment, the process response matrix is ​​a 3x3 matrix formed by weighting the Jacobian matrix of each geometric part according to the structural importance weight. The rows of the Jacobian matrix of each geometric part correspond to the local process offset component of the current geometric part, and the columns correspond to the design variables of the current geometric part. Each element represents the change in the local process offset component of the current geometric part when the design variables of the current geometric part change by one unit.

[0041] In this embodiment, the design variable dimension refers to the three independent directions of change in the design variable space, including the dimension of size variable, the dimension of angle variable, and the dimension of position variable.

[0042] In this embodiment, the structural domain compensation amount refers to the value that needs to be adjusted for the geometric dimensions and spatial positions of the geometric parts in the geometric skeleton data. The structural domain compensation amount includes the size variable compensation value, angle variable compensation value, and position variable compensation value of the geometric parts. The structural domain compensation amount is calculated based on the product of the projection coefficient of the unit direction vector of the process offset vector in the size variable dimension, angle variable dimension, and position variable dimension and the modulus of the process offset vector.

[0043] In this embodiment, the process domain compensation amount refers to the numerical value that needs to be adjusted for the machining parameter values ​​of the machining features in the target machining feature set. The process domain compensation amount includes the tool type parameter compensation value, feed rate parameter compensation value, spindle speed parameter compensation value, depth of cut parameter compensation value, and machining allowance parameter compensation value of the machining features. The process domain compensation amount is calculated based on the changes in size, angle, and position variables in the structural domain compensation amount of the geometric part, combined with the preset mapping relationship between design variables and machining parameters.

[0044] In this embodiment, adjusting the geometric skeleton data according to the structural domain compensation amount and adjusting the processing parameter group according to the process domain compensation amount to generate a parameterized driving model means that the structural domain compensation amount is superimposed on the current design variable value of the corresponding geometric part in the geometric skeleton data to obtain the adjusted geometric skeleton data, and the process domain compensation amount is superimposed on the current parameter value of the corresponding processing feature in the processing parameter group to obtain the adjusted processing parameter group. The adjusted geometric skeleton data and the adjusted processing parameter group are associated and integrated to generate a parameterized driving model in which the geometric shape and process parameters are matched.

[0045] In this embodiment, the parameterized driven model refers to a three-dimensional digital model that is a geometric skeleton data and a set of machining parameters that are coordinated and matched after dual adjustment of structural domain compensation and process domain compensation. It includes the geometric dimensions, spatial position, topological connection relationship of all geometric parts after adjustment, as well as the machining features associated with each geometric part and their corresponding adjusted machining parameter values. It is the final output result of the parameterized modeling method.

[0046] refer to Figure 2 In another embodiment of the present invention, a process constraint field for the initial structural model is constructed based on assembly constraint rules, the processing parameter set of the target processing feature set, and historical manufacturing data in a preset process knowledge base, including: The assembly topology features of the initial structural model are determined according to the assembly constraint rules. The current processing environment parameters are determined according to the processing parameter set of the target processing feature set. Multiple historical manufacturing data samples with topological homeomorphism to the geometric skeleton data of the initial structural model and whose historical processing parameter sets match the current processing environment parameters are obtained from the preset process knowledge base. Each historical manufacturing data sample contains a historical geometric coordinate set, a historical processing parameter set, and a corresponding historical manufacturing quality label. For each historical manufacturing data sample, the historical geometric coordinate set is mapped to the coordinate system of the geometric skeleton data of the initial structural model to obtain multiple registered historical process feature points. Each historical process feature point contains spatial coordinate values, corresponding historical processing parameter values, and corresponding historical manufacturing quality labels such as processing difficulty label value, deformation risk label value, and tolerance achievement label value. Using the spatial coordinates of each historical process feature point as the independent variable, and the corresponding processing difficulty label value, deformation risk label value, and tolerance achievement label value as the dependent variable, we construct the initial interpolation function for processing difficulty, the initial interpolation function for deformation risk, and the initial interpolation function for tolerance achievement, respectively. Calculate the first consistency coefficient between the local gradient vector of the initial interpolation function for processing difficulty and the local gradient vector of the initial interpolation function for deformation risk at each historical process feature point, and the second consistency coefficient between the local gradient vector of the initial interpolation function for deformation risk and the local gradient vector of the initial interpolation function for tolerance achievement. When the first consistency coefficient or the second consistency coefficient is less than the preset consistency threshold, a virtual feature point is inserted at the corresponding historical process feature point. The spatial coordinates of the virtual feature point are the coordinates after applying a preset offset to the spatial coordinates of the corresponding historical process feature point. When the first consistency coefficient is less than the preset consistency threshold, the dependent variable of the virtual feature point is the sum of the dependent variable of the corresponding historical process feature point and the preset adjustment amount. When the second consistency coefficient is less than the preset consistency threshold, the dependent variable of the virtual feature point is the difference between the dependent variable of the corresponding historical process feature point and the preset adjustment amount. Based on all historical process feature points and all virtual feature points, the initial interpolation function for processing difficulty, the initial interpolation function for deformation risk, and the initial interpolation function for tolerance achievement are refitted using the radial basis function interpolation method until the first consistency coefficient and the second consistency coefficient at all historical process feature points are not less than the preset consistency threshold. Then, the three refitted interpolation functions are combined into a process constraint field.

[0047] In this embodiment, determining the assembly topology features of the initial structural model according to the assembly constraint rules means that the system extracts the coplanar constraints, coaxial constraints, parallel constraints, perpendicular constraints, distance constraints, angular constraints, and alignment constraints contained in the assembly constraint rules from the initial structural model, constructs an assembly topology graph with geometric parts as nodes and assembly constraint relationships as edges, and uses the node connection relationships and edge direction attributes of the assembly topology graph as assembly topology features.

[0048] In this embodiment, determining the current machining environment parameters based on the machining parameter group of the target machining feature set means that the system extracts tool type parameters, feed rate parameters, spindle speed parameters, depth of cut parameters, machining allowance parameters, cooling method parameters, and machining sequence parameters from the machining parameter group of the target machining feature set, and converts them into standardized numerical vectors according to preset coding rules, which are then used as the current machining environment parameters.

[0049] In this embodiment, obtaining multiple historical manufacturing data samples from the preset process knowledge base that have a topological homeomorphic relationship with the geometric skeleton data of the initial structural model and whose historical processing parameter sets match the current processing environment parameters means that the system uses the assembly topology features of the initial structural model and the current processing environment parameters as joint search conditions, first filters closed-loop cases that match the assembly topology features in the preset process knowledge base, and then filters a preset number of closed-loop cases whose historical processing parameter sets are closest to the current processing environment parameters, and extracts their historical geometric coordinate sets, historical processing parameter sets, and corresponding historical manufacturing quality labels.

[0050] In this embodiment, the assembly topology features of the initial structural model refer to the node connection relationships and edge direction attributes of the assembly topology graph constructed with geometric parts as nodes and assembly constraints as edges.

[0051] In this embodiment, the current machining environment parameters refer to the standardized numerical vectors obtained by converting tool type parameters, feed rate parameters, spindle speed parameters, depth of cut parameters, machining allowance parameters, cooling method parameters, and machining sequence parameters.

[0052] In this embodiment, topological homeomorphism means that two geometric skeleton data have the same topological connection structure, that is, the number of nodes and edges in the assembly topology graph is the same and the connection relationship between the nodes is consistent, with only the size ratio and spatial position of the geometric parts being different.

[0053] In this embodiment, the historical geometric coordinate set refers to the set of coordinate ranges, reference point coordinates, and local coordinate system directions of all geometric parts included in the geometric skeleton data in the closed-loop case in three-dimensional space; the historical machining parameter set refers to the set of tool type parameters, feed rate parameters, spindle speed parameters, depth of cut parameters, machining allowance parameters, cooling method parameters, and machining sequence parameters corresponding to each machining feature in the closed-loop case; the historical manufacturing quality label refers to the actual manufacturing result evaluation data of each machining feature in the closed-loop case, including machining difficulty label value, deformation risk label value, and tolerance achievement label value.

[0054] In this embodiment, for each historical manufacturing data sample, the historical geometric coordinate set is mapped to the coordinate system of the geometric skeleton data of the initial structural model to obtain multiple registered historical process feature points. This means that the system calculates the rotation matrix and translation vector that minimizes the average distance between the historical geometric coordinate set and the geometric skeleton data of the initial structural model through an iterative nearest point registration method. The historical geometric coordinate set is then transformed to the coordinate system of the geometric skeleton data of the initial structural model using this rotation matrix and translation vector to obtain the registered historical process feature points.

[0055] In this embodiment, the spatial coordinate values ​​refer to the X-axis, Y-axis, and Z-axis coordinate values ​​of the registered historical process feature points in the coordinate system of the geometric skeleton data of the initial structural model; the corresponding historical machining parameter values ​​refer to the actual feed rate, actual spindle speed, and actual depth of cut corresponding to the historical process feature point; the machining difficulty label value is a quantitative score of the actual machining difficulty ranging from 0 to 1; the deformation risk label value is a quantitative score of the actual deformation degree ranging from 0 to 1; and the tolerance achievement label value is a quantitative score of the actual dimensional tolerance achievement rate ranging from 0 to 1.

[0056] In this embodiment, the spatial coordinates of each historical process feature point are used as independent variables, and the corresponding processing difficulty label value, deformation risk label value, and tolerance achievement label value are used as dependent variables to construct initial interpolation functions for processing difficulty, deformation risk, and tolerance achievement, respectively. This means that the system constructs the initial interpolation function for processing difficulty using spatial coordinates as input and processing difficulty label value as output using the inverse distance weighted interpolation method, constructs the initial interpolation function for deformation risk using spatial coordinates as input and deformation risk label value as output using the inverse distance weighted interpolation method, and constructs the initial interpolation function for tolerance achievement using spatial coordinates as input and tolerance achievement label value as output using the inverse distance weighted interpolation method.

[0057] In this embodiment, calculating the first consistency coefficient between the local gradient vector of the initial interpolation function for processing difficulty and the local gradient vector of the initial interpolation function for deformation risk at each historical process feature point means that the system calculates the local gradient vector of processing difficulty as a function of spatial coordinates at that point in the initial interpolation function for processing difficulty, and calculates the local gradient vector of deformation risk as a function of spatial coordinates at that point in the initial interpolation function for deformation risk. The system then calculates the dot product of the two local gradient vectors and divides it by the product of the magnitudes of the two local gradient vectors, and uses the calculation result as the first consistency coefficient.

[0058] In this embodiment, the second consistency coefficient is calculated for the local gradient vector of the initial interpolation function for deformation risk and the local gradient vector of the initial interpolation function for tolerance achievement at each historical process feature point. This means that the system calculates the local gradient vector of deformation risk as a function of spatial coordinates at that point in the initial interpolation function for deformation risk, and calculates the local gradient vector of tolerance achievement as a function of spatial coordinates at that point in the initial interpolation function for tolerance achievement. The dot product of the two local gradient vectors is divided by the product of the magnitudes of the two local gradient vectors, and the calculation result is used as the second consistency coefficient.

[0059] In this embodiment, the local gradient vector of the initial interpolation function for processing difficulty refers to the three-dimensional vector formed by taking the partial derivatives of the initial interpolation function for processing difficulty with respect to the three coordinate axes of the spatial coordinate value at a certain historical process feature point, indicating the direction in which the processing difficulty coefficient changes with the spatial position at that point at the fastest rate.

[0060] In this embodiment, the local gradient vector of the initial interpolation function for deformation risk refers to the three-dimensional vector formed by taking the partial derivatives of the initial interpolation function for deformation risk with respect to the three coordinate axes of the spatial coordinate value at a certain historical process feature point, indicating the direction in which the deformation risk coefficient changes with the spatial position at that point at the fastest rate.

[0061] In this embodiment, the local gradient vector of the initial interpolation function for tolerance achievement refers to the three-dimensional vector formed by taking the partial derivatives of the initial interpolation function for tolerance achievement with respect to the three coordinate axes of the spatial coordinate value at a certain historical process feature point. It indicates the direction in which the tolerance achievement coefficient changes with the spatial position at that point at the fastest rate.

[0062] In this embodiment, the preset consistency threshold is a pre-set numerical standard with a value range of 0 to 1. In this embodiment, the preset consistency threshold is 0.7, which is used to determine whether the change directions of the two interpolation functions are consistent.

[0063] In this embodiment, virtual feature points are inserted at the corresponding historical process feature points. The spatial coordinates of the virtual feature points are the coordinates of the corresponding historical process feature points after applying a preset offset. When the first consistency coefficient is less than the preset consistency threshold, the dependent variable of the virtual feature point is the sum of the dependent variable of the corresponding historical process feature point and the preset adjustment amount. When the second consistency coefficient is less than the preset consistency threshold, the dependent variable of the virtual feature point is the difference between the dependent variable of the corresponding historical process feature point and the preset adjustment amount. Specifically, when the first consistency coefficient is less than the preset consistency threshold, the system adds a preset offset to the X-axis coordinate of the spatial coordinate of the historical process feature point to obtain the spatial coordinate of the virtual feature point, and adds the preset adjustment amount to the dependent variable to obtain the dependent variable of the virtual feature point. When the second consistency coefficient is less than the preset consistency threshold, the dependent variable is subtracted from the preset adjustment amount to obtain the dependent variable of the virtual feature point.

[0064] In this embodiment, the preset offset is a pre-set distance value, and in this embodiment, the preset offset is 2 millimeters.

[0065] In this embodiment, the dependent variable value of the historical process feature point refers to the processing difficulty label value, deformation risk label value, or tolerance achievement label value corresponding to that historical process feature point. The preset adjustment amount is a pre-set value; in this embodiment, the preset adjustment amount is 0.05.

[0066] In this embodiment, based on all historical process feature points and all virtual feature points, the initial interpolation function for processing difficulty, the initial interpolation function for deformation risk, and the initial interpolation function for tolerance achievement are refitted using the radial basis function interpolation method. This process continues until the first and second consistency coefficients at all historical process feature points are not less than a preset consistency threshold. Then, the three refitted interpolation functions are combined into a process constraint field. This means that the system merges the original historical process feature points and virtual feature points into an updated set of feature points, reconstructs the three interpolation functions using the radial basis function interpolation method, and repeats the iterative process of calculating the consistency coefficient and inserting virtual feature points until the first and second consistency coefficients at all historical process feature points are not less than 0.7. Finally, the three interpolation functions that meet the consistency condition are combined into a process constraint field.

[0067] In this embodiment, the radial basis function interpolation method is an interpolation method that constructs radial basis functions with each known point as the center, and calculates the dependent variable value at any query point by weighting and summing the outputs of each radial basis function according to a preset weight coefficient. In this embodiment, the radial basis function is selected as the Gaussian kernel function, and the width parameter of the Gaussian kernel function is taken as twice the average distance between all historical process feature points.

[0068] In this embodiment, the three refitted interpolation functions are combined into a process constraint field. This means that the system combines the processing difficulty interpolation function, the deformation risk interpolation function, and the tolerance achievement interpolation function in parallel to form a three-dimensional composite function field with spatial coordinates as input and three values ​​of processing difficulty coefficient, deformation risk coefficient, and tolerance achievement coefficient as output. The domain covers all spatial coordinate points in the geometric skeleton data of the initial structural model.

[0069] In another embodiment of the present invention, the response coefficient matrix of the design variables of each geometric part to the process constraint feature values ​​of the associated processing features is extracted from the process constraint field, including: Determine the design variable space for each geometric part in the initial structural model, and determine the distribution of process constraint characteristic values ​​of the associated processing features of each geometric part in the process constraint field; Taking the current coordinate point in the design variable space of each geometric part as the starting node, and the adjacent geometric parts defined in the assembly constraint rules of the basic structure template as the propagation nodes, the perturbation propagation path of the design variable perturbation of the geometric part in the process constraint field is constructed step by step along the assembly direction of the assembly constraint rules. Each perturbation propagation path passes through the process constraint feature value distribution of the associated processing feature of at least one intermediate geometric part. The cumulative effect of the change in process constraint characteristic value caused by a unit disturbance of the design variable on each disturbance propagation path is calculated along the path, and the ratio of the cumulative effect to the path length is used as the path response coefficient of the disturbance propagation path. Calculate the deviation ratio between the path response coefficient of each disturbance propagation path and the historical response sensitivity of the intermediate geometric parts traversed by the disturbance propagation path. After normalizing all deviation ratios, calculate the complement value as the propagation validity of each disturbance propagation path. The complement value is 1 minus the normalized deviation ratio. Multiple disturbance propagation paths with a propagation validity greater than a preset validity threshold are selected. At the starting point of each selected disturbance propagation path, a unit spatial disturbance is applied along the X, Y, and Z axes of the spatial coordinate dimension. The spatial disturbance is converted into a design variable change through the Jacobian matrix of the geometric skeleton. The cumulative spatial response vector of the design variable change along the entire disturbance propagation path is calculated. The spatial coordinate dimension refers to the X, Y, and Z axes in the global coordinate system. A mapping relationship is established between the design variable space and the spatial coordinate dimension through the Jacobian matrix of the geometric skeleton.

[0070] The cumulative response vectors of all selected disturbance propagation paths are weighted and aggregated according to the propagation validity of each path to generate a response coefficient matrix of the geometric design variables to the process constraint eigenvalues ​​of the associated processing features.

[0071] In this embodiment, determining the design variable space for each geometric part in the initial structural model means that the system extracts the type and value range of the variable geometric parameters of each geometric part from the geometric skeleton data of the initial structural model, and organizes the dimension of size variable, the dimension of angle variable, and the dimension of position variable into a three-dimensional space as the design variable space for that geometric part.

[0072] In this embodiment, the associated processing features of the geometric part refer to the processing features in the target processing feature set that are assigned to a certain geometric part through a mapping relationship, including mortise and tenon connection features, screw installation features, hinge installation features, slide rail installation features, shelf support features, back panel slot features, edge banding features, and surface coating features.

[0073] In this embodiment, determining the distribution of process constraint feature values ​​of the associated processing features of each geometric part in the process constraint field means that the system obtains the spatial coordinate coverage range of each geometric part in the initial structural model, inputs all spatial coordinate points within the range into the process constraint field, reads the processing difficulty coefficient, deformation risk coefficient and tolerance achievement coefficient corresponding to each spatial coordinate point, and organizes them into three distribution fields.

[0074] In this embodiment, the assembly direction of the assembly constraint rule refers to the spatial direction vector used by the geometric part when moving from the initial position to the final assembly position as defined in the assembly constraint rule. It includes X-axis, Y-axis and Z-axis directional components and has a magnitude of 1. It is determined by a combination of coplanar constraints, coaxial constraints and alignment constraints.

[0075] In this embodiment, the design variable disturbance of the geometric part refers to the small change applied to a certain design variable in the design variable space of the geometric part, and the direction is along one of the dimensions of the size variable, the angle variable, or the position variable.

[0076] In this embodiment, the current coordinate point in the design variable space of each geometric part is taken as the starting node, and the adjacent geometric parts defined in the assembly constraint rules of the basic structure template are taken as the propagation nodes. The perturbation propagation path of the design variable disturbance of the geometric part in the process constraint field is constructed step by step along the assembly direction of the assembly constraint rules. This means that the system takes the current geometric part as the starting node, searches for adjacent geometric parts as propagation nodes step by step along the assembly direction, combines the starting node, the propagation nodes at each level, and the connection sequence between each node into a perturbation propagation path, and performs the above construction process for each geometric part.

[0077] In this embodiment, the process constraint feature value distribution of the associated processing features of each disturbance propagation path through at least one intermediate geometric part means that when the system constructs the disturbance propagation path, it skips geometric parts that are not associated with any processing features, ensuring that each intermediate geometric part on the path has a queryable process constraint feature value distribution.

[0078] In this embodiment, the intermediate geometric part refers to all geometric parts located between the starting node and the target node in the disturbance propagation path.

[0079] In this embodiment, the cumulative effect of the change in process constraint characteristic value caused by a unit disturbance of the design variable on each disturbance propagation path is calculated. The ratio of the cumulative effect to the path length is used as the path response coefficient of the disturbance propagation path. This means that after the system applies a unit disturbance at the starting node, it is passed to each level of propagation node. The cumulative effect is obtained by summing the changes in process constraint characteristic values ​​of all nodes on the path, and the path response coefficient is obtained by dividing the cumulative effect by the path length.

[0080] In this embodiment, the unit perturbation of the design variable refers to a small change with an amplitude of 1 applied to the design variable. The unit spatial perturbation of the spatial coordinate dimension has an amplitude of 1 mm in the X-axis, Y-axis and Z-axis directions.

[0081] In this embodiment, the cumulative effect of the change in process constraint feature value on the path refers to the sum of the changes in the process constraint feature value of the associated processing feature of each propagation node relative to the state before the disturbance, which includes three independent components: the change in processing difficulty coefficient, the change in deformation risk coefficient, and the change in tolerance achievement coefficient.

[0082] In this embodiment, the path length refers to the number of propagation steps traversed by the disturbance propagation path, which is the number of intermediate geometric parts included between the starting node and the target node plus 1.

[0083] In this embodiment, calculating the deviation ratio between the path response coefficient of each disturbance propagation path and the historical response sensitivity value of the intermediate geometric parts traversed by the disturbance propagation path refers to the system obtaining response sensitivity data recorded in historical closed-loop cases with the same geometric part identifier as the current intermediate geometric part from the preset process knowledge base, calculating the absolute value of the difference between the path response coefficient and the historical response sensitivity value of each intermediate geometric part, dividing it by the historical response sensitivity value, and then arithmetically averaging the deviation ratios of all intermediate geometric parts to obtain the final deviation ratio of the disturbance propagation path.

[0084] In this embodiment, the historical value of the response sensitivity of the intermediate geometric part refers to the ratio of the change in the process constraint characteristic value to the unit disturbance amplitude when the design variable is disturbed in the historical closed-loop cases stored in the preset process knowledge base that have the same geometric part identifier as the current intermediate geometric part. It includes three independent components: the corresponding processing difficulty coefficient, the deformation risk coefficient, and the tolerance achievement coefficient.

[0085] In this embodiment, calculating the propagation validity of each disturbance propagation path refers to the system determining the maximum and minimum values ​​of the deviation ratios of all disturbance propagation paths, subtracting the minimum value from the deviation ratio of each disturbance propagation path, and then dividing by the difference between the maximum and minimum values ​​to obtain the normalized deviation ratio. The propagation validity of the disturbance propagation path is then taken as 1 minus the normalized deviation ratio, i.e., the propagation validity equals 1 minus the normalized deviation ratio. The larger the normalized deviation ratio, the greater the deviation between the path response coefficient of the disturbance propagation path and the historical value of the response sensitivity, and the lower the propagation validity. The smaller the normalized deviation ratio, the smaller the deviation between the path response coefficient of the disturbance propagation path and the historical value of the response sensitivity, and the higher the propagation validity.

[0086] In this embodiment, the preset validity threshold is a pre-set numerical standard. In this embodiment, the preset validity threshold is 0.6, which is used to filter out disturbance propagation paths with a validity greater than 0.6 as effective propagation paths.

[0087] In this embodiment, at the starting point of each selected disturbance propagation path, a unit disturbance is applied along the dimension of size, dimension of angle, and dimension of position in the design variable space. The cumulative response vector of the unit disturbance along the entire disturbance propagation path is calculated. This means that for each disturbance propagation path with a propagation validity greater than 0.6, the system applies a unit disturbance along the three dimensions, calculates the change in the process constraint feature value of the associated processing feature of each intermediate geometric part step by step according to the propagation order, and accumulates them to obtain three cumulative response vectors along the three dimensions of the disturbance propagation path. Each cumulative response vector contains three components: the cumulative change in processing difficulty coefficient, the cumulative change in deformation risk coefficient, and the cumulative change in tolerance achievement coefficient.

[0088] In this embodiment, the design variable space refers to a three-dimensional space consisting of dimension variables, angle variables, and position variables.

[0089] In this embodiment, the dimension of size variable is the direction of change of linear dimension parameters such as length, width, height, thickness, diameter, and radius in the design variable space; the dimension of angle variable is the direction of change of angle parameters such as tilt angle, rotation angle, and pitch angle in the design variable space; and the dimension of position variable is the direction of change of position parameters such as the X, Y, and Z coordinate values ​​of the reference point in the global coordinate system in the design variable space.

[0090] In this embodiment, a unit disturbance refers to a small change with an amplitude of 1 applied to the design variable. The amplitude of a unit spatial disturbance in the spatial coordinate dimension is 1 millimeter in the X-axis, Y-axis and Z-axis directions.

[0091] In this embodiment, the cumulative response vectors of all screened disturbance propagation paths are weighted and aggregated according to the propagation validity of each path to generate a response coefficient matrix of the geometric design variables to the process constraint feature values ​​of the associated processing features. This means that the system uses the propagation validity as the weight to perform a weighted summation of the three components of the cumulative response vectors of all screened paths in three dimensions, obtaining three comprehensive response vectors along the three dimensions. The three comprehensive response vectors are arranged into 3 rows according to the design variable dimensions, and the process constraint feature value components in each comprehensive response vector are arranged into 3 columns, combined into a 3x3 square matrix as the response coefficient matrix. The rows of the response coefficient matrix correspond to the design variable dimensions (size variable dimension, angle variable dimension, position variable dimension), and the columns correspond to the process constraint feature value dimensions (processing difficulty coefficient, deformation risk coefficient, tolerance achievement coefficient). Each element represents the amount of change in the corresponding process constraint feature value caused by a unit change in the corresponding design variable.

[0092] In another embodiment of the present invention, the response coefficient matrix is ​​aggregated by process projection along the assembly direction of the initial structural model to generate the process offset vector and process response matrix of the initial structural model, including: Determine the local assembly direction vector for each geometric part in the initial structural model. The local assembly direction vector is the unit vector of the geometric part in the assembly mating direction defined in the assembly constraint rules. The response coefficient matrix of each geometric part is projected and aggregated with the corresponding local assembly direction vector to obtain the local process offset component of each geometric part. The projection and aggregation operation is as follows: the local assembly direction vector is transformed to the design variable space through the Jacobian matrix of the geometric skeleton to obtain the design variable direction vector. The dot product of each column vector of the response coefficient matrix and the design variable direction vector is calculated to obtain the directional derivative of the corresponding process constraint eigenvalue, which is used as the corresponding component of the local process offset component of the geometric part. Among them, the row vector of the response coefficient matrix corresponds to the response of each design variable dimension of the geometric part to the process constraint eigenvalue, and the design variable direction vector represents the degree of influence of the assembly direction on each design variable dimension. The local process offset components of all geometric parts are weighted and aggregated to obtain the process offset vector of the initial structural model. Calculate the Jacobian matrix of the local process offset component of each geometric part relative to the geometric skeleton data of the current geometric part, and combine the Jacobian matrices of all geometric parts into a process response matrix.

[0093] In this embodiment, determining the local assembly direction vector of each geometric part in the initial structural model means that the system extracts the assembly mating direction of each geometric part in the assembly process from the assembly constraint rules of the basic structural template, normalizes the direction vector into a unit vector with a modulus of 1, and uses it as the local assembly direction vector of the geometric part.

[0094] In this embodiment, the unit vector in the assembly mating direction defined in the assembly constraint rules for the geometric part refers to the direction vector with a magnitude of 1 obtained by dividing the assembly mating direction vector by the magnitude of the vector.

[0095] In this embodiment, the assembly mating direction refers to the spatial direction in which one geometric part moves relative to another geometric part during assembly, as specified in the assembly constraint rules of the basic structure template. It is determined by the normal direction of the mating surface in the coplanar constraint, the direction of the common axis in the coaxial constraint, or the direction of the alignment feature connection line in the alignment constraint.

[0096] In this embodiment, the response coefficient matrix of each geometric part is projected and aggregated with the corresponding local assembly direction vector to obtain the local process offset component of each geometric part. This means that the system transforms the local assembly direction vector to the design variable space through the Jacobian matrix of the geometric skeleton to obtain the design variable direction vector. The dot product of each column vector of the response coefficient matrix and the design variable direction vector is calculated to obtain the local process offset component of the geometric part: the first component of the local process offset component is equal to the dot product of the first column of the response coefficient matrix (corresponding to the processing difficulty coefficient column) and the design variable direction vector; the second component is equal to the dot product of the second column of the response coefficient matrix (corresponding to the deformation risk coefficient column) and the design variable direction vector; and the third component is equal to the dot product of the third column of the response coefficient matrix (corresponding to the tolerance achievement coefficient column) and the design variable direction vector. Among them, the column vector of the response coefficient matrix corresponds to the response of each design variable dimension of the geometric part to the specific process constraint characteristic value. The design variable direction vector is obtained by mapping the spatial coordinate component of the local assembly direction vector to the design variable dimension component by the Jacobian matrix of the geometric skeleton. The result of the dot product operation is the directional derivative of the process constraint characteristic value along the assembly direction, which represents the rate of change of the process constraint characteristic value when moving along the assembly direction. The three process feature directional derivative components are combined into a three-dimensional vector as the local process offset component of the geometric part.

[0097] In this embodiment, the local process offset components of all geometric parts are weighted and aggregated to obtain the process offset vector of the initial structural model. This means that the system assigns weight values ​​according to the structural importance of each geometric part, multiplies the three components of the local process offset component of each geometric part by the corresponding weight value, adds them together by the components, and then divides by the sum of the weight values ​​of all geometric parts to obtain the process offset vector of the initial structural model.

[0098] In this embodiment, calculating the Jacobian matrix of the local process offset component of each geometric part relative to the geometric skeleton data of that geometric part refers to the system using the design variable value of that geometric part as the independent variable and the local process offset component of that geometric part as the dependent variable, calculating the partial derivative of each dependent variable with respect to each independent variable, organizing all the partial derivatives into a matrix according to the arrangement of the dependent variable rows and independent variable columns, and using this matrix as the Jacobian matrix of that geometric part; and then weighting the Jacobian matrices of all geometric parts according to the structural importance weights to obtain a 3-row, 3-column matrix, which serves as the process response matrix.

[0099] In another embodiment of the present invention, the structural domain compensation amount and the process domain compensation amount are calculated based on the process offset vector and the process response matrix, respectively, including: Determine the unit direction vector of the process offset vector; Based on the eigenvalues ​​and column vectors of the process response matrix, determine the projection coefficients of the unit direction vector in each design variable dimension; Based on the projection coefficient and the modulus of the process offset vector, calculate the structural domain compensation for each geometric part and the process domain compensation for each processing feature.

[0100] In this embodiment, determining the unit direction vector of the process offset vector means that the system extracts the processing difficulty direction derivative component, deformation risk direction derivative component, and tolerance achievement direction derivative component from the process offset vector, divides the three-dimensional vector by the magnitude of the vector, and obtains a unit direction vector with a magnitude of 1.

[0101] In this embodiment, the eigenvalues ​​and eigenvectors of the process response matrix refer to the system decomposing the process response matrix using the Jacobian eigenvalue decomposition method to obtain all the eigenvalues ​​of the matrix, selecting the top 3 largest eigenvalues ​​and the eigenvectors corresponding to each eigenvalue, with the 3 eigenvalues ​​arranged in descending order.

[0102] In this embodiment, determining the projection coefficients of the unit direction vector in each design variable dimension based on the eigenvalues ​​of the process response matrix means that the system performs a dot product operation between the unit direction vector of the process offset vector and the three column vectors of the process response matrix to obtain three dot product results. The three dot product results are then multiplied by the absolute value of the corresponding eigenvalue and divided by the sum of the absolute values ​​of the three eigenvalues ​​to obtain three normalized projection coefficients, corresponding to the dimension of size variable, the dimension of angle variable, and the dimension of position variable, respectively. The values ​​of all coefficients range from 0 to 1 and their sum equals 1.

[0103] In this embodiment, based on the projection coefficients and the modulus of the process offset vector, the structural domain compensation amount for each geometric part and the process domain compensation amount for each processing feature are calculated. This refers to using the square root of the sum of the squares of the processing difficulty directional derivative components, deformation risk directional derivative components, and tolerance achievement directional derivative components of the process offset vector as the modulus. The modulus is then multiplied by the projection coefficients for the dimension, angle, and position variables to obtain the basic compensation amounts for the dimension, angle, and position variables. The basic compensation amounts for the dimension, angle, and position variables are then multiplied by the feature dimensions of the geometric part to obtain the dimensional compensation amount for the dimension, and multiplied by the feature angle of the geometric part to obtain the dimensional compensation amount for the position variable. Multiply the dimensional compensation by the feature dimension of the geometric part to obtain the dimensional compensation of the position variable. Based on the current coordinates of the geometric part in the design variable space, the dimensional compensation of the dimensional variable is allocated to the dimensional variables of the geometric part, the dimensional compensation of the angular variable is allocated to the angular variables of the geometric part, and the dimensional compensation of the position variable is allocated to the position variables of the geometric part, thus obtaining the structural domain compensation of the geometric part. Based on the changes in dimensional, angular, and position variables in the structural domain compensation of the geometric part, and combined with the preset mapping relationship between the feed rate parameters, spindle speed parameters, and depth of cut parameters in the machining parameter set and the design variables, the feed rate compensation, spindle speed compensation, and depth of cut compensation of the machining feature are calculated, thus obtaining the process domain compensation of the machining feature.

[0104] In another embodiment of the invention, after generating the parameterized driving model, the method further includes: Obtain the change sequence of the process offset vector and process response matrix of the parameterized driving model within a preset number of iterations, and calculate the directional offset and modulus attenuation of the process offset vector and the change of the condition number of the process response matrix based on the change sequence. When the directional offset exceeds a preset offset threshold, the modulus attenuation exceeds a preset attenuation threshold, or the change in the condition number exceeds a preset condition number change threshold, an early warning is triggered. When a condition is triggered, the calculation and adjustment of the structural domain compensation and process domain compensation of the parameterized driven model are repeated until the recalculated directional offset, modulus attenuation, and change in the condition number are less than or equal to the corresponding preset thresholds, and the iterative optimization is completed.

[0105] In this embodiment, the preset number of iterations refers to the number of iteration rounds set in advance. In this embodiment, the preset number of iterations is 10.

[0106] In this embodiment, obtaining the change sequence of the process offset vector and process response matrix of the parameterized driving model within a preset number of iterations means that the system obtains the process offset vector and process response matrix of each iteration according to a preset sampling interval within a preset number of iterations, and arranges them in the iteration order to form a process offset vector sequence and a process response matrix sequence.

[0107] In this embodiment, calculating the directional offset and modulus attenuation of the process offset vector and the change in the condition number of the process response matrix based on the change sequence involves the system extracting the process offset vectors for the starting and ending rounds from the change sequence, using the difference in the angle between the unit direction vectors as the directional offset, and dividing the difference in modulus by the initial modulus as the modulus attenuation; extracting the process response matrices for the starting and ending rounds, and dividing the difference in condition numbers by the initial condition number as the change in condition number. The condition number of the process response matrix is ​​the ratio of the absolute value of the largest eigenvalue to the absolute value of the smallest eigenvalue.

[0108] In this embodiment, the preset offset threshold is 5 degrees.

[0109] In this embodiment, the preset attenuation threshold is 0.1.

[0110] In this embodiment, the preset condition number change threshold is 0.2.

[0111] In this embodiment, the early warning prompt refers to the abnormal status notification signal triggered by the system when the directional offset is greater than a preset offset threshold, the modulus attenuation is greater than a preset attenuation threshold, or the change in the condition number is greater than a preset condition number change threshold. It includes an abnormality type identifier and the round in which the abnormality occurred.

[0112] In this embodiment, when the directional offset is greater than a preset offset threshold, the modulus attenuation is greater than a preset attenuation threshold, or the condition number change is greater than a preset condition number change threshold, an early warning is triggered and closed-loop iterative optimization is started. This means that after the early warning is triggered, the system re-executes the calculation and adjustment operations of the structural domain compensation and the process domain compensation, updates the parameterized driving model, and recalculates the directional offset, modulus attenuation, and condition number change. The iteration terminates when all three indicators fall back to below the corresponding preset threshold.

[0113] In this embodiment, the calculation and adjustment of the structural domain compensation amount and the process domain compensation amount refers to the system recalculating the structural domain compensation amount and the process domain compensation amount based on the current process offset vector and process response matrix, adding the structural domain compensation amount to the design variable values ​​of the geometric skeleton data, adding the process domain compensation amount to the parameter values ​​of the processing parameter group, and generating an updated parameterized driving model.

[0114] In another embodiment of the present invention, the construction method of the preset process knowledge base includes: We acquired multiple closed-loop cases of furniture design and manufacturing. Each closed-loop case includes a set of design parameters, structural data, a set of processing parameters, and corresponding manufacturing feedback data. For each closed-loop case, a theoretical structural model is constructed based on the design parameter set and structural data. As the digital twin of the physical entity of the closed-loop case, a process constraint field associated with the theoretical structural model is constructed on the geometric shape of the digital twin based on the processing parameter set and manufacturing feedback data. In the process constraint field, the corresponding response coefficient matrix is ​​extracted according to the method of screening the disturbance propagation path and calculating the propagation validity. The response coefficient matrix is ​​then projected and aggregated along the assembly direction to generate the corresponding process offset vector and process response matrix. At the same time, the propagation validity is recorded as a path confidence label. The process fingerprint data of each closed-loop case is formed by combining the unit direction vector and magnitude of the process offset vector, the eigenvalue sequence and eigenvector matrix of the process response matrix, and the corresponding path confidence label. The process fingerprint data is associated with and stored with the corresponding closed-loop cases to form a pre-defined process knowledge base.

[0115] In this embodiment, the design parameter set refers to the specific numerical set of appearance parameters and functional description information of the furniture product recorded in the closed-loop case; the structural data refers to the geometric skeleton data, assembly constraint rules, and topological connection relationships between various geometric parts of the furniture product recorded in the closed-loop case; the machining parameter set refers to the tool type parameters, feed rate parameters, spindle speed parameters, depth of cut parameters, machining allowance parameters, cooling method parameters, and machining sequence parameters corresponding to each machining feature of the furniture product recorded in the closed-loop case; and the manufacturing feedback data refers to the actual machining deviation data, actual machining pass rate, actual machining defect type, and defect distribution location obtained after the actual manufacturing of the furniture product is completed, as recorded in the closed-loop case.

[0116] In this embodiment, a theoretical structural model is constructed based on the design parameter set and structural data. The digital twin of the physical entity of the closed-loop case refers to the system using the appearance parameters and functional description information in the design parameter set as driving inputs and the geometric skeleton data and assembly constraint rules in the structural data as the construction framework to generate a three-dimensional digital geometric model corresponding to the closed-loop case. This three-dimensional digital geometric model is the digital twin of the physical entity of the closed-loop case.

[0117] In this embodiment, constructing a process constraint field associated with the theoretical structural model on the geometry of the digital twin based on the processing parameter set and manufacturing feedback data means that the system obtains the position of each spatial coordinate point in the geometry of the digital twin, and uses the processing parameter values ​​corresponding to each processing feature in the processing parameter set and the actual processing deviation data, actual processing pass rate, actual processing defect type and defect distribution location in the manufacturing feedback data as the source of the process constraint feature value at that spatial coordinate point. A continuous function field with spatial coordinate values ​​as independent variables and process constraint feature values ​​as dependent variables is constructed on all spatial coordinate points covered by the geometry of the digital twin.

[0118] In this embodiment, extracting the corresponding response coefficient matrix in the process constraint field according to the method of screening the disturbance propagation path and calculating the propagation validity means that the system extracts the response coefficient matrix corresponding to the closed-loop case in the process constraint field according to the aforementioned method of constructing a disturbance propagation path with the current coordinate point of the design variable space of each geometric part as the starting node, and the adjacent geometric parts as propagation nodes, calculating the path response coefficient, calculating the deviation ratio, determining the propagation validity by normalizing the deviation ratio complement, screening the disturbance propagation path with the propagation validity greater than the preset validity threshold, applying a unit disturbance at the starting point of the screened path to calculate the cumulative response vector, and weighting and aggregating all the screened cumulative response vectors according to the propagation validity to generate the response coefficient matrix.

[0119] In this embodiment, the response coefficient matrix is ​​projected and aggregated along the assembly direction to generate the corresponding process offset vector and process response matrix. Simultaneously, the propagation validity is recorded as a path confidence label. This means that in this closed-loop case, the system generates the process offset vector and process response matrix corresponding to this closed-loop case by following the aforementioned method: determining the local assembly direction vector of each geometric part, performing projection and aggregation operations on the response coefficient matrix and the local assembly direction vector to obtain local process offset components, weighting and aggregating all local process offset components to obtain the process offset vector, and calculating the Jacobian matrix of all local process offset components relative to the geometric skeleton data as the process response matrix. Furthermore, the propagation validity of all selected disturbance propagation paths involved in the response coefficient matrix extraction in this closed-loop case is summarized and recorded as a path confidence label.

[0120] In this embodiment, the process fingerprint data of each closed-loop case is formed by combining the unit direction vector and magnitude of the process offset vector, the eigenvalue sequence and eigenvector matrix of the process response matrix, and the corresponding path confidence label. This means that the system calculates the unit direction vector and magnitude of the process offset vector of the closed-loop case, calculates all the eigenvalues ​​of the process response matrix of the closed-loop case using the Jacobian eigenvalue decomposition method, selects the top three largest eigenvalues ​​and arranges them in descending order to form an eigenvalue sequence, and forms an eigenvector matrix with the eigenvectors corresponding to each eigenvalue. The path confidence label corresponding to the closed-loop case is then combined with the aforementioned unit direction vector, magnitude, eigenvalue sequence, and eigenvector matrix into a data packet, which serves as the process fingerprint data of the closed-loop case.

[0121] In this embodiment, the process fingerprint data is associated with the corresponding closed-loop case to form a preset process knowledge base. This means that the system associates the process fingerprint data of each closed-loop case with the design parameter set, structural data, processing parameter set and manufacturing feedback data of the closed-loop case. Using the closed-loop case identifier as an index, all associated data is stored in a non-volatile storage medium to form a preset process knowledge base.

[0122] In another embodiment of the invention, after generating the parameterized driving model, the method further includes: Calculate the current process fingerprint data of the parameterized driving model and extract the current path confidence label of the disturbance propagation path of the current parameterized driving model; The current process fingerprint data is compared with the historical process fingerprint data stored in the preset process knowledge base. The comprehensive similarity matching includes the unit direction vector similarity matching and modulus similarity matching of the process offset vector, the cosine similarity matching of the eigenvalue sequence of the process response matrix and the F norm similarity matching of the eigenvector matrix, as well as the label similarity matching between the current path confidence label and the historical path confidence label. When historical process fingerprint data with a comprehensive similarity greater than a preset similarity threshold is matched, the structural data and processing parameter set in the closed-loop case associated with the historical process fingerprint data are retrieved from the preset process knowledge base. Based on the structural data and the set of processing parameters, the parameterized driving model is optimized a second time.

[0123] In this embodiment, calculating the current process fingerprint data of the parameterized driving model means that the system uses the same method as when constructing the preset process knowledge base to extract the corresponding process offset vector and process response matrix from the process constraint field of the current parameterized driving model, calculates the unit direction vector and magnitude of the process offset vector and the eigenvalue sequence and eigenvector matrix of the process response matrix, and combines them into the process fingerprint data of the current parameterized driving model.

[0124] In this embodiment, extracting the current path confidence label of the disturbance propagation path of the current parameterized driving model means that during the process of extracting the response coefficient matrix in the process constraint field of the current parameterized driving model according to the disturbance propagation path screening and propagation validity calculation method, the propagation validity of all screened disturbance propagation paths involved in the response coefficient matrix extraction is summarized as the current path confidence label.

[0125] In this embodiment, the historical process fingerprint data stored in the preset process knowledge base refers to the process fingerprint data associated with each closed-loop case in the preset process knowledge base. Each historical process fingerprint data includes the unit direction vector and magnitude of the process offset vector of the corresponding closed-loop case, the eigenvalue sequence and eigenvector matrix of the process response matrix, and the corresponding path confidence label.

[0126] In this embodiment, performing comprehensive similarity matching between the current process fingerprint data and historical process fingerprint data stored in a preset process knowledge base involves the system calculating the similarity between the current process fingerprint data and each historical process fingerprint data in multiple dimensions: calculating the negative of the absolute value of the difference between the unit direction vectors as the unit direction vector similarity; calculating the smaller value divided by the larger value of the ratio of the magnitudes as the magnitude similarity; calculating the cosine similarity of the feature value sequences; calculating the F-norm similarity of the feature vector matrices, i.e., the negative of the F-norm of the difference between the two matrices; and calculating the label similarity between the current path confidence label and the historical path confidence label, i.e., the cosine of the angle between the two label vectors. The system then performs a weighted summation of the above five similarities to obtain the comprehensive similarity.

[0127] In this embodiment, the preset similarity threshold is a pre-set numerical standard with a value range of 0 to 1. In this embodiment, the preset similarity threshold is 0.5. When the overall similarity is greater than 0.5, the match is considered successful.

[0128] In this embodiment, retrieving the structural data and processing parameter set from the closed-loop case associated with historical process fingerprint data from the preset process knowledge base means that the system retrieves the closed-loop case from the preset process knowledge base based on the closed-loop case identifier associated with the successfully matched historical process fingerprint data, and reads the structural data and processing parameter set stored in the closed-loop case.

[0129] In this embodiment, the set of structural data and processing parameters in the closed-loop case refers to the set of structural data and processing parameters corresponding to the closed-loop case that is successfully matched with the process fingerprint data of the current parameterized driving model and stored in the preset process knowledge base.

[0130] In this embodiment, the secondary optimization of the parameterized driven model based on structural data and processing parameter set refers to the system superimposing and averaging the geometric skeleton data in the structural data of historical closed-loop cases with the geometric skeleton data of the current parameterized driven model, and superimposing and averaging the processing parameter set of historical closed-loop cases with the processing parameter set of the current parameterized driven model to obtain the secondary optimized geometric skeleton data and processing parameter set, thereby generating the secondary optimized parameterized driven model.

[0131] In another embodiment of the invention, after generating the parameterized driving model, the method further includes: Based on the overall size data of the parameterized driving model and the target machining feature set, the blanking data and machining path data of multiple machining elements are generated; Based on the assembly constraint rules, generate the assembly sequence and assembly positioning data for each machined component; The material cutting data, processing path data, assembly sequence and assembly positioning data are combined into a production guidance data package, and the production guidance data package is output to the CNC machining equipment. Receive actual machining deviation data from CNC machining equipment, update the process constraint field based on the actual machining deviation data, recalculate the updated process offset vector and process response matrix, and store the updated process offset vector and process response matrix in the preset process knowledge base.

[0132] In this embodiment, the overall size data of the parameterized driving model refers to the overall length, overall width, overall height of all geometric parts in the parameterized driving model in three-dimensional space, as well as the relative positional dimensions between each geometric part; the target machining feature set refers to the set of multiple machining features obtained by matching from the preset process knowledge base according to the functional description information.

[0133] In this embodiment, a processing element refers to a single, independently processable component obtained after the parametric driving model is disassembled according to assembly constraint rules. Each processing element corresponds to a geometric part in the parametric driving model and its associated processing features.

[0134] In this embodiment, the blanking data refers to the outline shape, size markings, texture direction markings, and material type identification of the raw material blank required for each processing element before processing.

[0135] In this embodiment, the machining path data refers to the tool movement trajectory corresponding to the machining features of each machining element in the CNC machining equipment, including the coordinates of the tool entry point, the coordinates of the tool exit point, the sequence of tool path points, the feed rate value, and the spindle speed value.

[0136] In this embodiment, generating blanking data and machining path data for multiple machining elements based on the overall size data of the parametric driving model and the target machining feature set means that the system decomposes the parametric driving model into multiple machining elements according to the assembly constraint rules, generates corresponding blanking data based on the geometric dimensions of each machining element, and generates corresponding machining path data based on the machining features associated with each machining element and the tool type parameters, feed rate parameters, spindle speed parameters, and depth of cut parameters in the machining parameter group.

[0137] In this embodiment, generating the assembly sequence and assembly positioning data for each processing element according to the assembly constraint rules means that the system uses coplanar constraints, coaxial constraints, parallel constraints, perpendicular constraints, distance constraints, angular constraints, and alignment constraints in the assembly constraint rules as constraint conditions to determine the assembly order between each processing element and obtain the assembly sequence; and determines the final spatial position coordinates and spatial attitude angle of each processing element after assembly to obtain the assembly positioning data.

[0138] In this embodiment, combining material cutting data, processing path data, assembly sequence, and assembly positioning data into a production guidance data package and outputting the production guidance data package to the CNC machining equipment means that the system associates and packages the material cutting data, processing path data, assembly sequence, and assembly positioning data of each processing element according to the identification of the processing element, generates a production guidance data package, and sends the production guidance data package to the control terminal of the CNC machining equipment through a wired network or wireless network.

[0139] In this embodiment, receiving actual machining deviation data fed back by the CNC machining equipment refers to the system receiving, through a wired or wireless network, the difference between the actual dimensional measurement value and the theoretical dimensional value of each detection point of each machining element transmitted back by the CNC machining equipment after machining is completed.

[0140] In this embodiment, the actual processing deviation data refers to the difference between the actual size measurement value and the theoretical size value of each detection point of each processing element returned by the CNC machining equipment after processing is completed.

[0141] In this embodiment, updating the process constraint field based on actual processing deviation data means that the system uses the actual processing deviation data as a new historical manufacturing quality label, the geometric skeleton data of the current parameterized driving model as a new historical geometric coordinate set, and the processing parameter set of the current target processing feature set as a new historical processing parameter set, combining them into a new historical manufacturing data sample. The new historical manufacturing data sample is then incorporated into the process constraint field, and the processing difficulty interpolation function, deformation risk interpolation function, and tolerance achievement interpolation function are refitted according to the method used during the initial construction to obtain the updated process constraint field.

[0142] In this embodiment, recalculating the updated process offset vector and process response matrix means that the system re-extracts the response coefficient matrix in the updated process constraint field, performs process projection aggregation on the response coefficient matrix along the assembly direction, and generates the updated process offset vector and process response matrix.

[0143] In this embodiment, storing the updated process offset vector and process response matrix into the preset process knowledge base means that the system uses the updated process offset vector and process response matrix as newly added closed-loop case data, and stores them in the preset process knowledge base in association with the corresponding geometric skeleton data, processing parameter group and actual processing deviation data, using the identifier of the newly added closed-loop case as an index.

[0144] This invention provides an embodiment of a parametric modeling system for furniture design, comprising: The requirement elicitation module is used to obtain design requirement information input by the user. The design requirement information includes appearance parameters and functional description information. The template and feature processing module is used to call the basic structure template that matches the appearance parameters from the preset process knowledge base, and determine the target processing feature set from the preset process knowledge base according to the functional description information. The processing features in the target processing feature set are associated with the geometric parts of the basic structure template to generate an initial structural model. The basic structure template contains geometric skeleton data and assembly constraint rules. The process constraint field construction module is used to construct the process constraint field of the initial structural model based on the assembly constraint rules, the processing parameter set of the target processing feature set, and the historical manufacturing data in the preset process knowledge base. Each spatial coordinate point of the process constraint field corresponds to a process constraint feature value, which includes the processing difficulty coefficient, deformation risk coefficient, and tolerance achievement coefficient. The process projection aggregation module is used to extract the response coefficient matrix of the design variables of each geometric part to the process constraint feature values ​​of the associated processing features in the process constraint field. The response coefficient matrix is ​​then projected and aggregated along the assembly direction of the initial structural model to generate the process offset vector and process response matrix of the initial structural model. The process offset vector is used to characterize the offset direction and degree of the current design state from the optimal manufacturing path, and the process response matrix is ​​used to characterize the response sensitivity of the offset direction to each dimension of the design variables. The compensation and generation module is used to calculate the structural domain compensation amount and the process domain compensation amount based on the process offset vector and the process response matrix, respectively. It adjusts the geometric skeleton data based on the structural domain compensation amount and adjusts the processing parameter group based on the process domain compensation amount to generate a parameterized driving model.

[0145] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of this invention and its equivalents, this invention also intends to include these modifications and variations.

Claims

1. A parametric modeling method for furniture design, characterized in that, include: Obtain design requirements information input by the user, which includes appearance parameters and functional descriptions. The basic structure template matching the appearance parameters is called from the preset process knowledge base, and the target processing feature set is determined from the preset process knowledge base according to the functional description information. The processing features in the target processing feature set are associated with the geometric parts of the basic structure template to generate an initial structural model. The basic structure template contains geometric skeleton data and assembly constraint rules. Based on the assembly constraint rules, the processing parameter set of the target processing feature set, and the historical manufacturing data in the preset process knowledge base, the process constraint field of the initial structural model is constructed. Each spatial coordinate point of the process constraint field corresponds to a process constraint feature value, which includes the processing difficulty coefficient, deformation risk coefficient, and tolerance achievement coefficient. In the process constraint field, the response coefficient matrix of the design variables of each geometric part to the process constraint feature value of the associated processing feature is extracted. The response coefficient matrix is ​​aggregated by process projection along the assembly direction of the initial structural model to generate the process offset vector and process response matrix of the initial structural model. The process offset vector is used to characterize the offset direction and degree of the current design state from the optimal manufacturing path, and the process response matrix is ​​used to characterize the response sensitivity of the offset direction to each dimension of the design variable. The structural domain compensation and the process domain compensation are calculated based on the process offset vector and the process response matrix, respectively. The geometric skeleton data is adjusted based on the structural domain compensation, and the processing parameter set is adjusted based on the process domain compensation to generate a parameterized driving model.

2. The parametric modeling method for furniture design according to claim 1, characterized in that, Based on assembly constraint rules, the processing parameter set of the target processing feature set, and historical manufacturing data in the preset process knowledge base, the process constraint field of the initial structural model is constructed, including: The assembly topology features of the initial structural model are determined according to the assembly constraint rules. The current processing environment parameters are determined according to the processing parameter set of the target processing feature set. Multiple historical manufacturing data samples with topological homeomorphism to the geometric skeleton data of the initial structural model and whose historical processing parameter sets match the current processing environment parameters are obtained from the preset process knowledge base. Each historical manufacturing data sample contains a historical geometric coordinate set, a historical processing parameter set, and a corresponding historical manufacturing quality label. For each historical manufacturing data sample, the historical geometric coordinate set is mapped to the coordinate system of the geometric skeleton data of the initial structural model to obtain multiple registered historical process feature points. Each historical process feature point contains spatial coordinate values, corresponding historical processing parameter values, and corresponding historical manufacturing quality labels such as processing difficulty label value, deformation risk label value, and tolerance achievement label value. Using the spatial coordinates of each historical process feature point as the independent variable, and the corresponding processing difficulty label value, deformation risk label value, and tolerance achievement label value as the dependent variable, we construct the initial interpolation function for processing difficulty, the initial interpolation function for deformation risk, and the initial interpolation function for tolerance achievement, respectively. Calculate the first consistency coefficient between the local gradient vector of the initial interpolation function for processing difficulty and the local gradient vector of the initial interpolation function for deformation risk at each historical process feature point, and the second consistency coefficient between the local gradient vector of the initial interpolation function for deformation risk and the local gradient vector of the initial interpolation function for tolerance achievement. When the first consistency coefficient or the second consistency coefficient is less than the preset consistency threshold, a virtual feature point is inserted at the corresponding historical process feature point. The spatial coordinates of the virtual feature point are the coordinates after applying a preset offset to the spatial coordinates of the corresponding historical process feature point. When the first consistency coefficient is less than the preset consistency threshold, the dependent variable of the virtual feature point is the sum of the dependent variable of the corresponding historical process feature point and the preset adjustment amount. When the second consistency coefficient is less than the preset consistency threshold, the dependent variable of the virtual feature point is the difference between the dependent variable of the corresponding historical process feature point and the preset adjustment amount. Based on all historical process feature points and all virtual feature points, the initial interpolation function for processing difficulty, the initial interpolation function for deformation risk, and the initial interpolation function for tolerance achievement are refitted using the radial basis function interpolation method until the first consistency coefficient and the second consistency coefficient at all historical process feature points are not less than the preset consistency threshold. Then, the three refitted interpolation functions are combined into a process constraint field.

3. The parametric modeling method for furniture design according to claim 1, characterized in that, Extract the response coefficient matrix of the design variables for each geometric part to the process constraint eigenvalues ​​of the associated machining features from the process constraint field, including: Determine the design variable space for each geometric part in the initial structural model, and determine the distribution of process constraint characteristic values ​​of the associated processing features of each geometric part in the process constraint field; Taking the current coordinate point in the design variable space of each geometric part as the starting node, and the adjacent geometric parts defined in the assembly constraint rules of the basic structure template as the propagation nodes, the perturbation propagation path of the design variable perturbation of the geometric part in the process constraint field is constructed step by step along the assembly direction of the assembly constraint rules. Each perturbation propagation path passes through the process constraint feature value distribution of the associated processing feature of at least one intermediate geometric part. The cumulative effect of the change in process constraint characteristic value caused by a unit disturbance of the design variable on each disturbance propagation path is calculated along the path, and the ratio of the cumulative effect to the path length is used as the path response coefficient of the disturbance propagation path. Calculate the deviation ratio between the path response coefficient of each disturbance propagation path and the historical response sensitivity of the intermediate geometric parts traversed by the disturbance propagation path. After normalizing all deviation ratios, calculate the complement value as the propagation validity of each disturbance propagation path. The complement value is 1 minus the normalized deviation ratio. Multiple disturbance propagation paths with a propagation validity greater than a preset validity threshold are selected. At the starting point of each selected disturbance propagation path, a unit disturbance is applied along the dimension of size, dimension of angle, and dimension of position in the design variable space. The cumulative response vector of the unit disturbance along the entire disturbance propagation path is calculated. The cumulative response vectors of all selected disturbance propagation paths are weighted and aggregated according to the propagation validity of each path to generate a response coefficient matrix of the geometric design variables to the process constraint eigenvalues ​​of the associated processing features.

4. The parametric modeling method for furniture design according to claim 1, characterized in that, The response coefficient matrix is ​​aggregated by process projection along the assembly direction of the initial structural model to generate the process offset vector and process response matrix of the initial structural model, including: Determine the local assembly direction vector for each geometric part in the initial structural model. The local assembly direction vector is the unit vector of the geometric part in the assembly mating direction defined in the assembly constraint rules. The response coefficient matrix of each geometric part is projected and aggregated with the corresponding local assembly direction vector to obtain the local process offset component of each geometric part. The projection and aggregation operation is as follows: the local assembly direction vector is transformed to the design variable space through the Jacobian matrix of the geometric skeleton to obtain the design variable direction vector. The dot product of each column vector of the response coefficient matrix and the design variable direction vector is calculated to obtain the directional derivative of the corresponding process constraint eigenvalue, which is used as the corresponding component of the local process offset component of the current geometric part. Among them, the row vector of the response coefficient matrix corresponds to the response of each design variable dimension of the current geometric part to the process constraint eigenvalue, and the design variable direction vector represents the degree of influence of the assembly direction on each design variable dimension. The local process offset components of all geometric parts are weighted and aggregated to obtain the process offset vector of the initial structural model. Calculate the Jacobian matrix of the local process offset component of each geometric part relative to the geometric skeleton data of the current geometric part, and combine the Jacobian matrices of all geometric parts into a process response matrix.

5. The parametric modeling method for furniture design according to claim 1, characterized in that, The structural domain compensation and the process domain compensation are calculated based on the process offset vector and the process response matrix, respectively, including: Determine the unit direction vector of the process offset vector; Based on the eigenvalues ​​and column vectors of the process response matrix, determine the projection coefficients of the unit direction vector in each design variable dimension; Based on the projection coefficient and the modulus of the process offset vector, calculate the structural domain compensation for each geometric part and the process domain compensation for each processing feature.

6. The parametric modeling method for furniture design according to claim 1, characterized in that, After generating the parameterized driven model, the following is also included: Obtain the change sequence of the process offset vector and process response matrix of the parameterized driving model within a preset number of iterations, and calculate the directional offset and modulus attenuation of the process offset vector and the change of the condition number of the process response matrix based on the change sequence. When the directional offset exceeds a preset offset threshold, the modulus attenuation exceeds a preset attenuation threshold, or the change in the condition number exceeds a preset condition number change threshold, an early warning is triggered. When a condition is triggered, the calculation and adjustment of the structural domain compensation and process domain compensation of the parameterized driven model are repeated until the recalculated directional offset, modulus attenuation, and change in the condition number are less than or equal to the corresponding preset thresholds, and the iterative optimization is completed.

7. The parametric modeling method for furniture design according to claim 3, characterized in that, The construction methods for the pre-defined process knowledge base include: We acquired multiple closed-loop cases of furniture design and manufacturing. Each closed-loop case includes a set of design parameters, structural data, a set of processing parameters, and corresponding manufacturing feedback data. For each closed-loop case, a theoretical structural model is constructed based on the design parameter set and structural data. As the digital twin of the physical entity of the closed-loop case, a process constraint field associated with the theoretical structural model is constructed on the geometric shape of the digital twin based on the processing parameter set and manufacturing feedback data. In the process constraint field, the corresponding response coefficient matrix is ​​extracted according to the method of screening the disturbance propagation path and calculating the propagation validity. The response coefficient matrix is ​​then projected and aggregated along the assembly direction to generate the corresponding process offset vector and process response matrix. At the same time, the propagation validity is recorded as a path confidence label. The process fingerprint data of each closed-loop case is formed by combining the unit direction vector and magnitude of the process offset vector, the eigenvalue sequence and eigenvector matrix of the process response matrix, and the corresponding path confidence label. The process fingerprint data is associated with and stored with the corresponding closed-loop cases to form a pre-defined process knowledge base.

8. The parametric modeling method for furniture design according to claim 7, characterized in that, After generating the parameterized driven model, the following is also included: Calculate the current process fingerprint data of the parameterized driving model and extract the current path confidence label of the disturbance propagation path of the current parameterized driving model; The current process fingerprint data is compared with the historical process fingerprint data stored in the preset process knowledge base. The comprehensive similarity matching includes the unit direction vector similarity matching and modulus similarity matching of the process offset vector, the cosine similarity matching of the eigenvalue sequence of the process response matrix and the F norm similarity matching of the eigenvector matrix, as well as the label similarity matching between the current path confidence label and the historical path confidence label. When historical process fingerprint data with a comprehensive similarity greater than a preset similarity threshold is matched, the structural data and processing parameter set in the closed-loop case associated with the historical process fingerprint data are retrieved from the preset process knowledge base. Based on the structural data and the set of processing parameters, the parameterized driving model is optimized a second time.

9. The parametric modeling method for furniture design according to claim 1, characterized in that, After generating the parameterized driven model, the following is also included: Based on the overall size data of the parameterized driving model and the target machining feature set, the blanking data and machining path data of multiple machining elements are generated; Based on the assembly constraint rules, generate the assembly sequence and assembly positioning data for each machined component; The material cutting data, processing path data, assembly sequence and assembly positioning data are combined into a production guidance data package, and the production guidance data package is output to the CNC machining equipment. Receive actual machining deviation data from CNC machining equipment, update the process constraint field based on the actual machining deviation data, recalculate the updated process offset vector and process response matrix, and store the updated process offset vector and process response matrix in the preset process knowledge base.

10. A parametric modeling system for furniture design, characterized in that, include: The requirement elicitation module is used to obtain design requirement information input by the user. The design requirement information includes appearance parameters and functional description information. The template and feature processing module is used to call the basic structure template that matches the appearance parameters from the preset process knowledge base, and determine the target processing feature set from the preset process knowledge base according to the functional description information. The processing features in the target processing feature set are associated with the geometric parts of the basic structure template to generate an initial structural model. The basic structure template contains geometric skeleton data and assembly constraint rules. The process constraint field construction module is used to construct the process constraint field of the initial structural model based on the assembly constraint rules, the processing parameter set of the target processing feature set, and the historical manufacturing data in the preset process knowledge base. Each spatial coordinate point of the process constraint field corresponds to a process constraint feature value, which includes the processing difficulty coefficient, deformation risk coefficient, and tolerance achievement coefficient. The process projection aggregation module is used to extract the response coefficient matrix of the design variables of each geometric part to the process constraint feature values ​​of the associated processing features in the process constraint field. The response coefficient matrix is ​​then projected and aggregated along the assembly direction of the initial structural model to generate the process offset vector and process response matrix of the initial structural model. The process offset vector is used to characterize the offset direction and degree of the current design state from the optimal manufacturing path, and the process response matrix is ​​used to characterize the response sensitivity of the offset direction to each dimension of the design variables. The compensation and generation module is used to calculate the structural domain compensation amount and the process domain compensation amount based on the process offset vector and the process response matrix, respectively. It adjusts the geometric skeleton data based on the structural domain compensation amount and adjusts the processing parameter set based on the process domain compensation amount to generate a parameterized driving model.

Images