Model-based complex product positioning reference forward design method
Through the model-based positioning benchmark forward design method, the product model is analyzed to obtain key reference points, the candidate benchmark point set is screened and the knowledge graph is constructed, which solves the problem of relying on experience in the design of complex product body-in-white and realizes efficient and scientific benchmark design.
Patent Information
- Application Number
- CN202510966321.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-14
- Publication Date
- 2025-10-24
AI Technical Summary
In the body-in-white design of complex products, existing technologies rely on professional experience, resulting in low efficiency and unstable quality of baseline design, making it difficult to achieve high-quality and efficient baseline design.
A model-based positioning benchmark forward design method is adopted. By analyzing the product model to obtain key reference points, spatial relationship rules are used to screen the candidate benchmark point set, and stability evaluation is performed. A multi-level knowledge graph architecture is constructed to achieve automation and intelligence of benchmark design.
It improves the scientificity and consistency of benchmark design, reduces the cost of manual trial and error, and provides an adaptive design framework to adapt to design changes and process upgrades.
Smart Images
Figure CN120832771A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The application relates to a model-based complex product forward design method, in particular to a forward design method for locating a reference of a complex part. BACKGROUND
[0002] A complex product has the characteristics of highly complex function integration, multi-field coupling and correlation, cross-region and remote design, etc. Especially in the system design process of a complex product, complex data interaction is faced.
[0003] Taking vehicle manufacturing as an example, a white body which is completed by welding and has not been painted is a key link in the automobile manufacturing process. The design of the white body is directly related to the structural strength and safety of the vehicle. Taking a white body of a new energy automobile as an example, the white body contains more than 500 parts and more than 130 sub-assemblies, and the assembly of each layer is required to be robust enough. According to statistics, the white body part level reference is 3000+, and the sub-assembly reference is nearly 1100+, and the references among the parts, sub-assemblies and assemblies of each level are related to each other. How to complete the white body reference design with high quality and high efficiency in the design stage is a big pain point in the size development process of the automobile industry.
[0004] Many automobile companies have formed their own reference design specifications and methods through practice to unify the design, process and detection reference and ensure the size stability and assembly accuracy of the parts in the whole life cycle. For example, the German automobile enterprises proposed the RPS reference point system (Referenz Punkt Systematik), the ASKF system (Locate-Sketch-CarBody-Production) of Opel, the MCP (Master Control Points) of Ford, the PLP (Principle Location Point) of Mazda and the LM (Locator Map) of Toyota. Although many schemes and optimization methods have been developed for the body reference design, the application still needs to be highly dependent on professional experience. SUMMARY
[0005] The purpose of the application is to establish a positioning reference forward design method which can be used throughout the design, manufacturing and detection processes and reduce the dependence on human experience, and the method is based on the multi-domain information weak coupling of "demand-function-behavior-architecture-geometry".
[0006] The model-based complex product positioning datum forward design method performs the following operations: characterized by: obtaining a three-dimensional model or a two-dimensional model of a complex product, parsing the product model into a part family composed of multiple parts; parsing each part model to obtain part features, including geometric contour, thickness, and functional elements; representing the functional elements as a set of spatial points with coordinates and vector directions, extracting a set of key reference points from the functional elements; for each key reference point, filtering a set of candidate datum points according to spatial relationship rules; evaluating the stability of the set of candidate datum points, if the stability requirement is met, outputting the set of candidate datum points as a set of datum points, if the stability requirement is not met, issuing an alarm to update the spatial relationship rules, and reextracting key reference points and filtering candidate datum points using the updated spatial relationship rules. The existing stability requirement calculation algorithm (such as the maximum projection surface algorithm) is used to evaluate the stability of the set of candidate datum points. The updated spatial relationship rules can be adjusted by parameters or new rule functions.
[0007] Further, the extraction rules of the key reference points include: for a sagittal plane feature, taking a center point as a key reference point, for a bar feature, taking a point with the maximum or minimum value in the X direction or Y direction or Z direction of the part coordinate system as a key reference point, for a curved surface feature, taking a center point and a set of face contour boundary points of the curved surface feature as key reference points, for a hole feature, taking a center point as a key reference point, and for a line feature, taking two end points as key reference points. The bar feature refers to a closed graph in a digital model, with a length significantly greater than a width. A closed graph with a width within 14mm-35mm and a length greater than 300mm is taken as a bar feature.
[0008] Further, a curved surface with a shortest width >40mm and an area greater than 0.01m² is taken as a curved surface feature.
[0009] Further, the spatial relationship rules select one or more of boundary proportion relationship, distribution distance relationship, coincidence relationship, and direction relationship.
[0010] Further, the boundary proportion relationship is expressed as: , wherein, is the maximum span between datum points in the axial direction; is the total length of the part in the axial direction.
[0011] Further, the distribution distance relationship is expressed as: , wherein, is a functional key reference point; is a datum target point.
[0012] Further, the coincidence relationship is expressed as: the key reference point is the datum point.
[0013] Further, the direction relationship is expressed as: any two positioning holes of the same part form a group, the center points of each positioning hole in each group are obtained respectively, a connecting line between the two center points is drawn to form a center point connecting line, the center point connecting line is projected to the maximum projection plane to obtain a projection connecting line, if at least one of the included angles between the projection connecting line and the X-axis, or the Y-axis, or the Z-axis of the world coordinate system is less than 25°, the group of positioning holes is listed as a candidate positioning hole group.
[0014] Further, the projection connecting line is recommended as a positioning hole group when the included angle with any coordinate axis is close to 0°.
[0015] Further, the key reference point includes a functional key reference point and a geometric key reference point representing a geometric feature; the functional key reference point is represented as , wherein is a spatial coordinate in the body coordinate system, is a vector direction thereof; the functional key reference point is represented as , wherein is a spatial coordinate in the body coordinate system, is a vector direction thereof, is an axial direction;
[0016] The geometric key reference point is represented as , wherein is a spatial coordinate in the global coordinate system, is a vector direction thereof.
[0017] Further, the key reference point set and the reference point set are taken as a reference layer, the part feature point cloud is taken as a point cloud layer, the part feature is taken as a feature layer, and the part model is taken as a part layer; the part layer, the feature layer, the point cloud layer and the reference layer are stored in hierarchy to form a product forward model.
[0018] Further, the product forward model is represented by a knowledge graph architecture, the part layer is the first layer, the feature layer is the second layer, the point cloud layer is the third layer, and the reference layer is the fourth layer in the forward model;
[0019] The part layer of the forward model includes a part family and process connections and process relationships between parts, the feature layer includes part features and a derivation relationship of each part feature from the part, the point cloud layer includes part feature point clouds and a derivation relationship of each point cloud and the part feature, and the reference layer includes a key reference point set and a derivation relationship of each key reference point from the part feature, a reference point set and a derivation relationship of each reference point from the key reference point; the derivation relationship refers to that the next layer is obtained by calculation from the previous layer.
[0020] The advantages of the present application are:
[0021] 1. Key reference points represent all functional elements on a part that require control and measurement. By constructing a parametric representation model of geometric elements, subjective empirical parameters are converted into a data set with computable characteristics, and the benchmark layout is transformed into a resolvable spatial topological relationship. This improves the scientific nature and verifiability of the benchmark design while ensuring design specifications. This data-driven design paradigm not only establishes an objective design evaluation system, but also possesses inherent programmability due to its structured nature, providing a technical path for design automation and intelligentization.
[0022] 2. Use a spatial point set with coordinates and vector directions to represent functional elements, simplifying complex and discrete functional elements, which is more conducive to the calculation of the spatial position relationship between functional elements and reference points, and simplifies the design process.
[0023] 3. The functional elements of complex parts are transformed into functional key reference points and geometric key reference points, achieving parametric representation of part features. Combined with spatial relationship rules, a quantifiable and analyzable benchmark positioning solution is constructed. A multi-level knowledge graph architecture is introduced, integrating multi-dimensional constraints at the part, feature, point cloud, and benchmark levels to construct a machine-parseable engineering semantic network, laying the foundation for intelligent benchmark design. This method can significantly improve the scientificity and consistency of benchmark design, reduce the cost of manual trial and error, and provide an adaptive framework for intelligent benchmark design of complex body parts, with high practical application value. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] Figure 1 It is the traditional body base design process.
[0025] Figure 2 It is the effect of shape, position and spatial changes on the baseline design.
[0026] Figure 3 It indicates the overlapping relationship.
[0027] Figure 4 It is an indication of the direction relationship.
[0028] Figure 5 It is a knowledge graph architecture.
[0029] Figure 6 It is a benchmark design process based on key reference points.
[0030] Figure 7 It is a benchmark knowledge graph based on key reference points. DETAILED DESCRIPTION
[0031] The function of a component is usually embodied by its functional elements, which can be divided into reference elements and measured elements according to different functions. Reference design is based on reference elements, and the reference element refers to an actual element on a part that plays a positioning role, an element used to determine the shape, direction or (and) position of the geometric tolerance band of the measured element, such as a certain plane, hole, etc. on the part. On a white body, the reference element usually refers to a feature with assembly function, such as a welding surface, a bonding surface, an assembly surface, a matching surface, etc. Traditional reference design usually adopts a global optimization design method, which first analyzes part information (such as geometric contour, material thickness, part function), then sets a reference scheme according to experience, and then adds or removes or moves reference points to carry out reference position optimization design under the constraint conditions of stability, inheritance, functionality, etc. According to the positioning principle, iteration is carried out to determine the final reference scheme, such as shown in Figure 2 This method is highly dependent on experience and low in efficiency. For complex curved surface parts or asymmetric geometric bodies, there are often compatibility problems in geometric shape, which makes the universal proportion principle lose engineering feasibility and other problems, resulting in unstable design quality, and the optimization of artificial trial and error is prone to local optimization and difficult to balance multiple target requirements, and the solidification scheme is difficult to adapt to design changes and process upgrading. Therefore, it is necessary to simplify and quantify the traditional reference design method.
[0032] In some embodiments, as shown in Figure 6 A model-based positioning reference forward design method for complex products is provided, which performs the following operations: obtaining a three-dimensional model or a two-dimensional model of a complex product, analyzing the product model into a part family composed of multiple parts; analyzing each part model to obtain part features, including geometric contour, thickness, and functional elements; representing the functional elements with a set of spatial points with coordinates and vector directions, extracting a set of key reference points from the functional elements; for each key reference point, filtering a set of candidate reference points according to spatial relationship rules; evaluating the stability of the set of candidate reference points, if the stability requirement is met, outputting the set of candidate reference points as a set of reference points, if the stability requirement is not met, issuing an alarm to update the spatial relationship rules, and reextracting the key reference points and filtering the set of candidate reference points using the updated spatial relationship rules. The existing stability requirement calculation algorithm (such as the maximum projection surface algorithm) is used to evaluate the stability of the set of candidate reference points. The updated spatial relationship rules can be adjusted by parameters or new rule functions.
[0033] In some embodiments, the extraction rule of the key reference points comprises: taking a center point as the key reference point for a sagittal plane feature, taking a point with the maximum or minimum value in the X direction or Y direction or Z direction of the part coordinate system of the strip-shaped feature as the key reference point, taking the center point and the face contour boundary point set of the curved surface feature as the key reference point, taking the center point as the key reference point for the hole feature, and taking two end points as the key reference points for the line feature. The strip-shaped feature refers to a closed graph in the digital model, and the length is obviously greater than the width. The closed graph with a width within 14mm-35mm and a length greater than 300mm is taken as the strip-shaped feature. The curved surface with a shortest width greater than 40mm and an area greater than 0.01m2 is taken as the curved surface feature.
[0034] In some embodiments, the spatial relationship rule selects one or more of the boundary proportion relationship, the distribution distance relationship, the coincidence relationship, and the direction relationship.
[0035] The boundary proportion relationship is expressed as: , wherein, is the maximum span between the reference points in the axial direction; is the total length of the part in the axial direction.
[0036] The distribution distance relationship is expressed as: , wherein, is the functional key reference point; is the reference target point.
[0037] The coincidence relationship is expressed as: the key reference point is the reference point, as shown in Figure 3 .
[0038] The direction relationship is expressed as: any two positioning holes of the same part are a group, the center points of each positioning hole of each group are obtained, a connecting line between the two center points is formed, a projection connecting line is obtained by projecting the center point connecting line to the maximum projection plane, and if at least one of the included angles of the projection connecting line with the X axis, or the Y axis, or the Z axis of the world coordinate system is less than 25°, the group of positioning holes is listed as the candidate positioning hole group, as shown in Figure 4 .
[0039] The projection connecting line is more recommended as the positioning hole group when the included angle with any coordinate axis is close to 0°.
[0040] In some embodiments, the key reference points include functional key reference points and geometric key reference points representing geometric features; the functional key reference points are represented as , wherein is the spatial coordinate in the body coordinate system, is the vector direction thereof; the functional key reference points are represented as , wherein represents the spatial coordinates in the body coordinate system, represents the vector direction thereof, represents the axial direction;
[0041] The geometric key reference point is represented as wherein is the spatial coordinates in the global coordinate system, is the vector direction thereof.
[0042] In some embodiments, the key reference point set and the datum point set are taken as a datum layer, the part feature point cloud is taken as a point cloud layer, the part feature is taken as a feature layer, and the part model is taken as a part layer; the part layer, the feature layer, the point cloud layer and the datum layer are stored in hierarchy to form a product forward model.
[0043] As shown in Figure 7 , the product forward model is represented by a knowledge graph architecture, the part layer in the forward model is the first layer, the feature layer is the second layer, the point cloud layer is the third layer, and the datum layer is the fourth layer;
[0044] The part layer of the forward model contains the part family and the process connection and process relationship between parts, the feature layer includes the part feature and the derivation relationship of each part feature from the part, the point cloud layer includes the part feature point cloud and the derivation relationship of each point cloud with the part feature, and the datum layer includes the key reference point set and the derivation relationship of each key reference point from the part feature, the derivation relationship of each datum point from the key reference point; the derivation relationship refers to that the next layer is obtained by calculation from the previous layer. The key reference point set is obtained by any one of the embodiments in the present application or a combination of several embodiments.
[0045] The present application has the following advantages:
[0046] 1. The key reference points represent all functional elements that need to be controlled and measured on the part. By constructing a parameterized representation model of geometric elements, subjective experience parameters are converted into a data set with calculable characteristics, and datum arrangement is converted into an analyzable spatial topological relationship, which not only ensures design specifications, but also improves the scientificity and verifiability of datum design. This data-driven design paradigm not only establishes an objective design evaluation system, but also has a natural programmable advantage due to its structured characteristics, providing a technical implementation path for design automation and intelligentization.
[0047] 2. The spatial point set with coordinates and vector direction is used to represent the functional elements, which simplifies the complex discrete functional elements and is more conducive to the position relationship operation of the functional elements and the datum points in space, simplifying the design process.
[0048] 3、The functional elements of complex parts are converted into functional key reference points and geometric key reference points, realizing the parametric characterization of part features, and combining with spatial relationship rules, a quantifiable and analytical benchmark positioning scheme is constructed. The multi-level knowledge graph architecture is introduced, and the multi-dimensional constraints of part layer, feature layer, point cloud layer and benchmark layer are integrated to construct a machine-analyzable engineering semantic network, laying a foundation for the realization of benchmark intelligent design. This method can significantly improve the scientificity and consistency of benchmark design, reduce the cost of manual trial and error, and provide an adaptive framework for intelligent benchmark design of complex body parts, with high practical application value.
Claims
1. A model-based complex product positioning datum forward design method, the method performs the following operations: characterized in that: The three-dimensional model or two-dimensional model of a complex product is acquired, and the product model is parsed into a part family composed of multiple parts; each part model is parsed to obtain part features, and the part features include a geometric contour, a thickness, and a functional element; the functional element is represented by a spatial point set with coordinates and vector directions, and key reference point sets are extracted from the functional element; for each key reference point, candidate datum point sets are screened according to spatial relationship rules; the candidate datum point sets are evaluated for stability, and if the stability requirements are met, the candidate datum point sets are output as datum point sets, and if the stability requirements are not met, an alarm is sent to update the spatial relationship rules, and the key reference points are extracted again and the candidate datum point sets are screened using the updated spatial relationship rules.
2. The forward design method of claim 1, wherein: The extraction rules of the key reference points include: for a sagittal plane feature, a center point is taken as a key reference point; for a strip-shaped feature, a point with the maximum or minimum value in the X direction, the Y direction or the Z direction of the part coordinate system is taken as a key reference point; for a curved surface feature, a center point and a surface contour boundary point set of the curved surface feature are taken as key reference points; for a hole feature, a center point is taken as a key reference point; and for a line feature, two end points are taken as key reference points.
3. The forward design method according to claim 2, wherein: Curved surfaces with a shortest width greater than 40 mm and an area greater than 0.01 m² are taken as curved surface features.
4. The forward design method of claim 1, wherein: The spatial relationship rules select one or more of a boundary proportion relationship, a distribution distance relationship, a coincidence relationship, and a direction relationship.
5. The method of forward design of claim 4, wherein: The boundary proportional relationship is expressed as: wherein, is the maximum span between reference points in the axial direction; is the total length of the part in the axial direction.
6. The method of forward design of claim 4, wherein: The distribution distance relationship is expressed as: wherein, is a functionally critical reference point; is a fiducial target point.
7. The forward design method of claim 4, wherein: The coincidence relationship is expressed as: the key reference point is the datum point.
8. The method of forward design of claim 4, wherein: The direction relationship is expressed as: any two positioning holes of the same part form a group, the center points of each positioning hole in each group are obtained, a center point connecting line is formed between the two center points, the center point connecting line is projected onto the maximum projection plane to obtain a projection connecting line, and if at least one of the angles between the projection connecting line and the X axis, the Y axis or the Z axis of the world coordinate system is less than 25°, the group of positioning holes is listed as a candidate positioning hole group.
9. The forward design method of claim 1, wherein: Key reference points include functional key reference points and geometric key reference points representing geometric features; the functional key reference points are represented as wherein is a spatial coordinate in the body coordinate system, is a vector direction thereof; the functional key reference points are represented as wherein is a spatial coordinate in the body coordinate system, is a vector direction thereof, is an axial direction; Geometric key reference points are denoted as wherein is a spatial coordinate in a global coordinate system, is its vector direction.
10. The method of forward design of claim 1, wherein: The key reference point set and the datum point set are taken as a datum layer, the part feature point cloud is taken as a point cloud layer, the part feature is taken as a feature layer, and the part model is taken as a part layer; the part layer, the feature layer, the point cloud layer and the datum layer are stored in layers to form a product forward model.
11. The product forward model is represented by a knowledge graph architecture, the part layer is the first layer, the feature layer is the second layer, the point cloud layer is the third layer, and the datum layer is the fourth layer; The part layer of the forward model includes a part family and process connections and process relationships between parts, the feature layer includes part features and a derivation relationship of each part feature from the part, the point cloud layer includes part feature point clouds and a derivation relationship of each point cloud with the part feature, and the datum layer includes a key reference point set and a derivation relationship of each key reference point from the part feature, a derivation relationship of the datum point set and each datum point from the key reference point; the derivation relationship means that the next layer is obtained by calculation from the previous layer.