A smart die casting manufacturability analysis method and apparatus

Abstract

This invention provides an intelligent method and apparatus for die casting manufacturability analysis. The method converts the casting model into a triangular mesh and calculates its structural features. A candidate set of draft directions is obtained through the triangular mesh. For each draft direction, the openable region, under-openable region, and non-openable region are determined. Based on the under-openable and non-openable regions, candidate regions for each slider to be placed and its core-pulling direction are determined. Based on the above, various manufacturability indices for adding sliders are constructed, and the results of multiple manufacturability indices for core pulling of each slider are analyzed using a scoring model, generating and outputting an analysis report. This method comprehensively considers the material properties, geometric features, draft scheme, and specific die casting process conditions of the casting, with draft scheme design as the core, to achieve the evaluation and optimization of draft schemes. This significantly improves the automation level of the die casting product draft scheme design process, significantly reduces the design time, and effectively reduces operational difficulty.

Description

Technical Field

[0001] This invention relates to the field of casting engineering technology, and in particular to an intelligent method and apparatus for analyzing the manufacturability of die casting. Background Technology

[0002] With the development of modern manufacturing, software tools such as computer-aided design and computer-aided manufacturing play an important role in casting processes. Casting is widely used in the manufacture of metal and plastic products, but it faces challenges in improving production efficiency and product quality.

[0003] The casting process involves pouring liquid material into a mold, allowing it to solidify, and then pulling out the finished product. Drafting is a critical step that directly affects product quality and mold life. Currently, drafting feasibility assessments primarily rely on analyzing the angle between the casting surface normal and the draft direction, but existing technologies have some limitations.

[0004] First, the selection of the draft direction relies on the engineer's experience. Existing tools offer limited choices and lack an intelligent evaluation mechanism based on geometric features and process constraints, which may lead to the omission of the optimal draft direction, thereby increasing manufacturing risks. Second, existing tools can only rely on manual identification of basic undercut areas when identifying the core-pulling area of ​​the slider, reducing design efficiency. Finally, draft analysis mainly focuses on geometric features and fails to fully consider material properties and process rules, making it difficult to conduct a reasonable draft feasibility score, thus affecting structural optimization.

[0005] In conclusion, it is imperative to develop an integrated and intelligent casting manufacturability analysis tool to improve the practical application effect of casting processes. Summary of the Invention

[0006] In view of this, embodiments of the present invention provide an intelligent die casting manufacturability analysis method and apparatus to solve the problems of difficult and incomplete casting manufacturability analysis.

[0007] To achieve the above objectives, the embodiments of the present invention provide the following technical solutions:

[0008] The first aspect of this invention discloses an intelligent die-casting manufacturability analysis method, the method comprising:

[0009] The parametric geometric surfaces of the casting model are converted into corresponding uniform triangular meshes, and the structural features of each parametric geometric surface of the casting model are calculated based on the uniform triangular meshes corresponding to each parametric geometric surface.

[0010] Based on the uniform triangular mesh corresponding to each parametric geometric surface, the principal axis of inertia of the casting model and the principal axis of the bounding box of the casting model are calculated, and a rotational transformation is performed based on the principal axis of inertia and the principal axis of the bounding box to construct a candidate set of draft directions.

[0011] For any draft direction in the candidate set, the draft direction is defined as the moving mold opening direction, and the opposite direction of the draft direction is defined as the stationary mold opening direction. Based on the moving mold opening direction and the stationary mold opening direction, ray detection is performed to obtain a set of moldable facets.

[0012] Based on any one of the openable moldable surfaces in the set of openable moldable surfaces and the mold opening direction, the openable moldable area and the under-openable moldable area are determined; the mold opening direction is either the moving mold opening direction or the stationary mold opening direction.

[0013] The candidate areas and candidate directions of the slider to be arranged are determined based on the under-opened mold area and the unopenable mold area; the unopenable mold area is the area that cannot be reached by both the moving mold opening direction and the stationary mold opening direction.

[0014] Select a target slider candidate direction from the set of slider candidate directions, and use the target slider candidate direction as the core-pulling direction of the slider corresponding to the candidate area of ​​the slider to be arranged; the target slider candidate direction is the slider candidate direction that minimizes the total area of ​​the under-opened area or the unopenable area and maximizes the minimum draft angle value.

[0015] Based on the material properties of the casting model, the structural features of each parametric geometric surface, the moldable area, the under-molding area, the mold-unopenable area, the candidate area of ​​the slider to be arranged, and the core-pulling direction of the slider corresponding to each candidate area of ​​the slider to be arranged, a feature vector of the draft direction is constructed.

[0016] The feature vectors of each draft direction are input into the scoring model to obtain multiple draft manufacturability index results for each draft direction. An analysis report is generated and output based on the multiple draft manufacturability index results.

[0017] Preferably, the step of calculating the structural features of each parametric geometric surface of the casting model based on the uniform triangular mesh corresponding to each parametric geometric surface includes:

[0018] Obtain point cloud data and topological connectivity information from the uniform triangular mesh corresponding to each parametric geometric surface;

[0019] All point cloud data and topology connection information are input into the recognition model, and the structural features of each parameterized geometric surface are output.

[0020] Preferably, the step of performing a rotational transformation based on the principal axis of inertia and the principal axis of the bounding box to construct a candidate set of draft directions includes:

[0021] Around the inertial main axis and the main axis of the enclosure box, a uniform rotation transformation is performed according to a preset angle to obtain each axis.

[0022] Based on the principal axis of inertia and the principal axis of the bounding box, as well as the various axes, a candidate set of draft directions is constructed.

[0023] Preferably, the step of performing ray testing based on the opening direction of the moving mold and the opening direction of the stationary mold to obtain a set of openable moldable facets includes:

[0024] Parallel rays are emitted into the interior of the casting model along the opening direction of the moving mold and the opening direction of the stationary mold.

[0025] Calculate the intersection points between each parallel ray and the triangular mesh of each parameterized geometric surface, and mark the visible side of the triangular facet in the triangular mesh where each intersection point is located as the moldable facet.

[0026] The set of openable moldable surfaces corresponding to the draft direction is determined based on all openable moldable surfaces.

[0027] Preferably, determining the moldable area and the under-molded area based on any moldable surface in the set of moldable surfaces and the mold opening direction includes:

[0028] For any moldable surface in the set of moldable surfaces, calculate the draft angle based on the normal at the three vertices of the moldable surface and the mold opening direction;

[0029] When the draft angle is greater than a preset threshold, the moldable surface at the intersection point is marked as a moldable area;

[0030] When the draft angle is not greater than a preset threshold, the moldable surface at the intersection point is marked as an under-molded area.

[0031] Preferably, determining the candidate region and candidate direction set of the slider to be arranged based on the under-molding region and the un-molding region includes:

[0032] Based on the adjacency relationship between the under-molded area and the un-molded area on the triangular mesh, the connected regions of the under-molded area and the un-molded area are extracted;

[0033] Each of the connected regions is marked as a candidate region for the slider to be placed;

[0034] For any candidate region of the slider to be placed, the normal directions of all triangular facets contained in the candidate region of the slider to be placed are used as the candidate directions of the slider, and a set of candidate directions of the slider is obtained.

[0035] Preferably, the training process of the scoring model includes:

[0036] Obtain historical casting sample data, which includes sample feature vectors;

[0037] Initialize the deep learning model to obtain the initial scoring model;

[0038] The sample feature vector is input into the initial scoring model to obtain the manufacturability score, draft angle, and risk level;

[0039] Calculate the loss function between the manufacturability score, the draft angle, and the risk level, and the actual manufacturability score, actual draft angle, and actual risk level corresponding to the sample feature vector;

[0040] If the loss function converges, the initial scoring model is marked as a scoring model;

[0041] If the loss function does not converge, the parameters of the initial scoring model are modified, and the process returns to the step of inputting the sample feature vector into the initial scoring model to obtain the manufacturability score, draft angle, and risk level.

[0042] A second aspect of this invention discloses an intelligent die-casting manufacturability analysis device, the device comprising:

[0043] The transformation model representation unit is used to convert the parametric geometric surfaces of the casting model into corresponding triangular meshes, and to calculate the structural features of each parametric geometric surface of the casting model based on the uniform triangular meshes corresponding to each parametric geometric surface.

[0044] The first construction unit is used to calculate the principal axis of inertia of the casting model and the principal axis of the bounding box of the casting model based on the uniform triangular mesh corresponding to each parameterized geometric surface, and to perform rotational transformation based on the principal axis of inertia and the principal axis of the bounding box to construct a candidate set of draft directions.

[0045] The definition unit defines any draft direction in the candidate set as the moving mold opening direction and the opposite direction of the draft direction as the stationary mold opening direction. Based on the moving mold opening direction and the stationary mold opening direction, ray detection is performed to obtain a set of moldable facets.

[0046] The first determining unit is used to determine the moldable area and the under-molding area based on any moldable surface in the set of moldable surfaces and the mold opening direction; the mold opening direction is the moving mold opening direction or the stationary mold opening direction.

[0047] The second determining unit is used to determine the candidate area and candidate direction set of the slider to be arranged based on the under-opened mold area and the unopenable mold area; the unopenable mold area is the area that cannot be reached by both the moving mold opening direction and the stationary mold opening direction.

[0048] The selection unit is used to select a target slider candidate direction from the set of slider candidate directions, and use the target slider candidate direction as the core-pulling direction of the slider corresponding to the candidate area of ​​the slider to be arranged; the target slider candidate direction is the slider candidate direction that minimizes the total area of ​​the under-opened area or the unopenable area and maximizes the minimum draft angle value.

[0049] The second construction unit is used to construct the feature vector of the draft direction based on the material properties of the casting model, the structural features of each parameterized geometric surface, the moldable area, the under-molding area, the mold-unopenable area, the candidate area of ​​the slider to be arranged, and the core-pulling direction of the slider corresponding to each candidate area of ​​the slider to be arranged.

[0050] The scoring unit is used to input the feature vector of each draft direction into the scoring model to obtain multiple draft manufacturability index results for each draft direction, and generate and output an analysis report based on the multiple draft manufacturability index results.

[0051] Preferably, the conversion model representation unit includes:

[0052] The acquisition module is used to acquire point cloud data and topological connection information in the uniform triangular mesh corresponding to each parameterized geometric surface;

[0053] The recognition module is used to input all point cloud data and topological connection information into the recognition model and output the structural features of each parameterized geometric surface.

[0054] Preferably, the first building unit includes:

[0055] The deflection module is used to uniformly rotate around the inertial spindle and the spindle of the enclosure box at a preset angle to obtain various axes;

[0056] A construction module is used to construct a candidate set of draft directions based on the principal axis of inertia, the principal axis of the bounding box, and various axial directions.

[0057] Based on the above embodiments of the present invention, an intelligent die-casting manufacturability analysis method and apparatus are provided. This method converts the casting model into a triangular mesh and calculates structural features. A candidate set of draft directions is obtained through the triangular mesh. For each draft direction, the openable region, under-openable region, and non-openable region are determined. Based on the under-openable and non-openable regions, candidate regions for each slider to be placed and its core-pulling direction are determined. Based on the above, various manufacturability indices for adding sliders are constructed, and the results of multiple manufacturability indices for core pulling of each slider are analyzed using a scoring model, generating and outputting an analysis report. This method comprehensively considers the material properties, geometric features, draft scheme, and specific die-casting process conditions of the casting, with draft scheme design as the core, to achieve the evaluation and optimization of draft schemes. This significantly improves the automation level of the die-casting product draft scheme design process, significantly reduces the design time, and effectively reduces the learning curve for technical personnel. Attached Figure Description

[0058] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0059] Figure 1 A flowchart illustrating an intelligent die-casting manufacturability analysis method provided in this embodiment of the invention;

[0060] Figure 2 This is a schematic diagram of the draft area division provided in an embodiment of the present invention;

[0061] Figure 3 A schematic diagram showing the position of the rounded corner region on a three-dimensional model according to an embodiment of the present invention;

[0062] Figure 4 This is a schematic diagram showing the location of the reinforced area on a three-dimensional model according to an embodiment of the present invention.

[0063] Figure 5 This is a schematic diagram of thickness analysis on a three-dimensional model provided in an embodiment of the present invention;

[0064] Figure 6 This is a structural block diagram of an intelligent die-casting manufacturability analysis device provided in an embodiment of the present invention. Detailed Implementation

[0065] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0066] In this application, the terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0067] As can be seen from the background technology, the current draft feasibility assessment in the casting process mainly relies on the experience of engineers, lacking intelligent analysis and methods that comprehensively consider material properties and process rules, resulting in low design efficiency and increased manufacturing risks.

[0068] Therefore, this invention provides an intelligent die-casting manufacturability analysis method and apparatus. The method converts the casting model into a triangular mesh and calculates structural features. A candidate set of draft directions is obtained through the triangular mesh. For each draft direction, openable areas, under-openable areas, and non-openable areas are determined. Based on the under-openable and non-openable areas, candidate areas for each slider to be placed and its core-pulling direction are determined. Based on the above, various manufacturability indices for adding sliders are constructed, and the results of multiple manufacturability indices for core pulling of each slider are analyzed using a scoring model, generating and outputting an analysis report. The method comprehensively considers the material properties, geometric features, draft scheme, and specific die-casting process conditions of the casting, with draft scheme design as the core, to achieve the evaluation and optimization of draft schemes. This significantly improves the automation level of the die-casting product draft scheme design process, significantly reduces the design time, and effectively reduces the learning curve for technical personnel.

[0069] See Figure 1 The diagram illustrates a flowchart of an intelligent die-casting manufacturability analysis method provided by an embodiment of the present invention. The method includes:

[0070] Step S101: Convert the parametric geometric surfaces of the casting model into corresponding uniform triangular meshes, and calculate the structural features of each parametric geometric surface of the casting model based on the uniform triangular meshes corresponding to each parametric geometric surface.

[0071] It should be noted that the Boundary Representation (BRep) model is a solid composed of continuous parametric curves and surfaces.

[0072] In the specific implementation step S101, the casting model input by the user is received, the parameterized geometric surfaces of the casting model are converted into corresponding uniform discrete triangular meshes, and a two-way mapping relationship between each parameterized geometric surface and the uniform triangular mesh is established; the structural features of each parameterized geometric surface of the casting model are calculated based on the three-level mesh of each parameterized geometric surface.

[0073] Specifically, the process of calculating structural features is as follows (processes A1 to A2):

[0074] Process A1: Obtain point cloud data and topology connectivity information from the uniform triangular mesh corresponding to each parametric geometric surface.

[0075] It is understandable that point cloud data refers to the three-dimensional coordinates of vertices in a triangular mesh; topological link information refers to the connection relationships between vertices, edges, and faces in a triangular mesh.

[0076] Process A2: Input all point cloud data and topology connection information into the recognition model and output the structural features of each parameterized geometric surface.

[0077] In implementing process A2, all point cloud data and topological connection information are input into the recognition model as features. The recognition model identifies the structural features of each parameterized geometric surface, including the small hole area, rounded corner area, sharp edge area, thin wall area and stiffened area in the casting model.

[0078] In some embodiments, by combining the mapping relationship between geometric surfaces and triangular meshes, a corresponding point set can be obtained for each CAD geometric surface, thereby reducing the complexity of point cloud segmentation and clustering. These point sets are then fitted with geometric features such as small holes, rounded corners, sharp edges, thin walls, and stiffeners, and cross-validated using thickness attributes to improve the accuracy and stability of feature recognition results.

[0079] In practical applications, the dimensions, volume, surface area, and other geometric properties of the bounding box of the casting model are calculated based on the triangular mesh of each geometric face. Furthermore, the local thickness, area, volume, and other properties of the casting model are calculated based on the material properties (such as density and Young's modulus) input by the user.

[0080] Understandably, the material properties of the casting model, the structural features of its parametric geometric surfaces, and the geometric properties of its bounding box, such as dimensions, volume, and surface area, play a crucial role in the subsequent evaluation of the mold-making scheme. For example, when the Young's modulus of the mold material is high, its elastic deformation capacity is weak. This makes it difficult to alleviate geometric mismatch and contact interference through deformation during the drafting process, which can easily lead to localized contact stress concentration and an increase in peak draft force. This can result in problems such as surface damage, structural failure, or draft failure, thus affecting the selection of the draft angle and ultimately the evaluation of the mold-making scheme.

[0081] Step S102: Calculate the principal axis of inertia of the casting model and the principal axis of the bounding box of the casting model based on the uniform triangular mesh corresponding to each parametric geometric surface, and perform rotational transformation based on the principal axis of inertia and the principal axis of the bounding box to construct a candidate set of draft directions.

[0082] It should be noted that the principal axes of inertia are determined based on the centroid of the casting model. This can be achieved by calculating the centroid coordinates of all vertices in a uniform triangular mesh (i.e., summing and averaging them) to obtain the centroid coordinates. Next, the overall inertia matrix is ​​calculated based on the volume mesh, and eigenvalue decomposition is performed on the inertia matrix to obtain three orthogonal eigenvectors. These eigenvectors are the directions of the principal axes of inertia. The principal axes of the bounding box, on the other hand, are calculated using principal component analysis or other strategies to represent the three principal axes of the overall shape of the model. The specific calculation methods for the principal axes of inertia of the casting model and the principal axes of the bounding box are not limited here; the method depends on the actual application.

[0083] It is understandable that a uniform rotation transformation is performed around the principal axis of inertia and the principal axis of the bounding box at a preset angle to obtain various axes; then, based on the principal axis of inertia, the principal axis of the bounding box, and each axis, a candidate set of draft directions is constructed.

[0084] Among them, the draft direction is evenly distributed and representative on the unit spherical surface.

[0085] Step S103: For any draft direction in the candidate set, define the draft direction as the moving mold opening direction, define the opposite direction of the draft direction as the static mold opening direction, and perform ray detection based on the moving mold opening direction and the static mold opening direction to obtain a set of moldable facets.

[0086] See the embodiments of the present invention. Figure 2 The red area represents the draft area on the moving mold side, the green area represents the draft area on the stationary mold side, and the white area represents the draft area on the slider side, which will be discussed later. This is the candidate area for placing the slider.

[0087] It should be noted that the specific implementation process of obtaining the set of openable mold facets based on ray testing according to the opening direction of the moving mold and the opening direction of the stationary mold is as follows (processes B1 to B3):

[0088] Process B1: Along the opening direction of the moving mold and the opening direction of the stationary mold, emit parallel rays into the interior of the casting model.

[0089] Procedure B2: Calculate the intersection points between each parallel ray and the triangular mesh of each parameterized geometric surface, and mark the visible side of the triangular facet in the triangular mesh at each intersection point as the moldable facet.

[0090] It should be noted that the intersection of each parallel ray with the triangular mesh of each parametric geometric surface is to solve the intersection point of each parallel ray with the outer surface of the casting model when it is emitted into the interior of the casting model.

[0091] Process B3: Determine the set of openable mold surfaces corresponding to the draft direction based on all openable mold surfaces.

[0092] In practical applications, for processes B1 to B3 mentioned above, the ray intersection process can be accelerated by using an AABB tree based on a triangular mesh.

[0093] Step S104: Determine the moldable area and the under-molded area based on any moldable surface in the set of moldable surfaces and the mold opening direction.

[0094] It should be noted that the mold opening direction can be either the moving mold opening direction or the stationary mold opening direction. That is, if the moldable surface is located in the draft area of ​​the moving mold, the mold opening direction is the moving mold opening direction; if the moldable surface is located in the draft area of ​​the stationary mold, the mold opening direction is the stationary mold opening direction. If it is at the interface between the moving and stationary molds, either the moving mold opening direction or the stationary mold opening direction is acceptable.

[0095] In the specific implementation step S104, for any openable moldable surface in the set of openable moldable surfaces, the draft angle is calculated based on the normals at the three vertices of the openable moldable surface and the mold opening direction. Then, it is determined whether the draft angle is greater than a preset threshold. When the draft angle is greater than the preset threshold, the openable moldable surface at the intersection point is marked as an openable moldable region; when the draft angle is not greater than the preset threshold, the openable moldable surface at the intersection point is marked as an under-opened moldable region.

[0096] For example: Calculate the angle θ1 between the normal and the parallel ray at the three vertices of the moldable facet. The draft angle θ2 of the triangular facet is θ1 - 90 (PI / 2). When θ2 is less than the preset minimum draft angle, the moldable facet at the intersection is marked as an under-molded area.

[0097] Step S105: Determine the candidate area and candidate direction set of the slider to be arranged based on the under-opened mold area and the mold-unopenable mold area.

[0098] It should be noted that the unopenable mold area is the area that cannot be reached in either the moving mold opening direction or the stationary mold opening direction.

[0099] In the specific implementation step S105, based on the adjacency relationship between the under-molded area and the un-molded area on the triangular mesh, the connected regions of the under-molded area and the un-molded area are extracted. Then, each connected region is marked as a candidate region for the slider to be placed; for any candidate region for the slider to be placed, the normal directions of all triangular faces contained in the candidate region are used as the candidate directions for the slider, thus obtaining a set of candidate directions for the slider.

[0100] Understandably, the purpose of adding sliders is because certain facets cannot be handled in either the moving mold opening direction or the stationary mold opening direction. Therefore, it is necessary to add separate candidate areas for the sliders to be placed and slider opening directions to ensure that the facet can be successfully molded. In this case, this area is neither the moving mold side draft area nor the stationary mold side draft area, but rather... Figure 2 The white area shown is the draft area on the slider side, which is the candidate area where the slider is to be placed.

[0101] Step S106: Select the target slider candidate direction from the slider candidate direction set, and use the target slider candidate direction as the core-pulling direction of the slider corresponding to the candidate area of ​​the slider to be arranged.

[0102] It should be noted that the target slider candidate direction is the slider candidate direction that minimizes the total area of ​​the under-opened or unopened mold area and maximizes the minimum draft angle value.

[0103] In the specific implementation step S106, for each candidate slider direction in the candidate slider direction set, the draft angle distribution of each triangular facet within the connected region under the candidate slider direction is calculated. This allows for the statistical analysis of the total area of ​​under-opened or unopened faces within the candidate slider direction, as well as the minimum draft angle value within the connected region. Finally, the candidate direction that minimizes the total area of ​​under-opened or unopened faces and maximizes the minimum draft angle value is selected as the core-pulling direction of the corresponding slider within the connected region.

[0104] Understandably, there may be multiple candidate areas for placing sliders within a mold, and each area that cannot be molded requires the addition of a corresponding slider. Figure 2 In the example, the protruding parts (white areas) that cannot be opened on both the moving mold side and the stationary mold side will be blocked, so a slider needs to be added to achieve mold opening.

[0105] Step S107: Based on the material properties of the casting model, the structural features of each parametric geometric surface, the moldable area, the under-molding area, the mold-unopenable area, the candidate area of ​​the slider to be arranged, and the core-pulling direction of the slider corresponding to each candidate area of ​​the slider to be arranged, construct the feature vector of the draft direction.

[0106] In the specific implementation of step S107, for each draft direction, based on the above material properties, the structural features of each geometric surface, the moldable area, the under-molded area, the non-molded area, the candidate area of ​​the slider to be arranged, and the core-pulling direction of the slider corresponding to the candidate area of ​​each slider to be arranged, a feature vector corresponding to that draft direction is constructed.

[0107] It should be noted that the feature vector includes at least the area ratio of the moldable area, the under-molded area, and the non-molded area, the number and total area of ​​the candidate areas for the slider to be arranged, and the area statistics of the thin-walled area, the small hole area, the sharp edge area, the rounded corner area, and the ribbed area under different draft states.

[0108] Step S108: Input the feature vector of each draft direction into the scoring model to obtain the results of multiple draft manufacturability indices for each draft direction. Generate and output an analysis report based on the results of multiple draft manufacturability indices.

[0109] It should be noted that the results of multiple draft manufacturability indicators include at least:

[0110] 1. Comprehensive manufacturability score and draft risk level for this draft direction.

[0111] 2. The predicted area ratios of the moldable area, the under-molded area, and the non-moldable area, as well as the predicted number and area of ​​candidate areas for the sliders to be placed.

[0112] 3. Among structural features such as regions, small hole regions, sharp edge regions, rounded corner regions, and reinforced regions, the proportion of areas in a state of under-molding or inability to be demolded, and the corresponding risk coefficients.

[0113] 4. Based on different structural feature types, recommend corresponding minimum draft angle parameters to guide the design of local structure glue addition, glue reduction, and draft angle optimization.

[0114] Understandably, during the inference phase of the scoring model, for each draft direction in the candidate set of draft directions, the deep learning scoring module is invoked to calculate the draft manufacturability score and various evaluation index scores corresponding to the draft direction. Subsequently, the candidate draft directions are sorted according to these scores, and several draft directions with higher scores are selected as target draft directions.

[0115] Next, the target draft direction and its corresponding evaluation index scores are mapped onto the three-dimensional model of the casting. The corresponding CAD geometric surfaces and their triangular mesh areas are parametrically marked and displayed in color, thereby realizing the automatic recommendation and visual verification of the casting draft scheme.

[0116] In practical applications, based on the above geometric preprocessing results, structural feature recognition results, draft analysis results, and multiple draft manufacturability index results output by the scoring model, a draft manufacturability analysis report for castings is automatically generated and output.

[0117] It should be noted that the analysis report should include at least the following basic information: the bounding box dimensions, volume, surface area, mass, and material properties of the casting model. Additionally, the analysis report should provide the target draft direction and manufacturability scores for each draft direction, the area ratio of the undercut and undercut regions, and the number and spatial distribution of candidate regions for the sliders to be placed.

[0118] In addition, the analysis report should include a schematic diagram of the location of structural features such as orifice areas, rounded corner areas, thin-walled areas, and stiffened areas on the 3D model, as well as their relationship with the draft area. Simultaneously, a list of high-risk areas based on the local thickness field and draft angle distribution should be provided, including the corresponding CAD surface number, spatial location, risk type, and risk level. The schematic diagram of the location of rounded corner areas on the 3D model is shown below. Figure 3 As shown, the red, green, and blue areas represent the model's rounded corner recognition effect. A schematic diagram illustrating the location of the stiffened areas on the 3D model is shown below. Figure 4 As shown. And, a schematic diagram of the thickness analysis on the 3D model is shown below. Figure 5 As shown.

[0119] Finally, the analysis report should provide local structural optimization suggestions recommended by the scoring model, including suggestions on adding or reducing adhesive to thin-walled areas, undercut areas, and stress concentration areas, adjusting local draft angles, and modifying parameters such as the position of the slider core-pulling area.

[0120] In some specific embodiments, the scoring model is pre-trained based on a deep learning model and historical casting sample data. The specific training process is as follows (processes C1 to C6):

[0121] Process C1: Obtain historical casting sample data.

[0122] It should be noted that the historical casting sample data includes sample feature vectors, which are constructed from the results of draft area features, thickness attributes, material attributes, and structural features.

[0123] Understandably, historical casting sample data also includes, but is not limited to, casting models, casting material properties, actual draft directions and their draft angle configurations for multiple automatic draft directions, draft area division results, structural feature annotations (including small hole areas, rounded corner areas, sharp edge areas, thin-walled areas, and stiffened areas), slider arrangement area division results, as well as draft manufacturability scores, risk levels, and evaluation scores for each evaluation indicator given by process or mold designers.

[0124] Process C2: Initialize the deep learning model to obtain the initial scoring model.

[0125] It should be noted that deep learning models can be multi-layer feedforward neural networks.

[0126] It is understood that the scoring model is equivalent to the relevant technologies and working mechanisms of existing reward models. In this embodiment of the invention, geometric features, process features, and material features are uniformly represented using high-dimensional semantics such as vectors. The scoring conversion layer then converts these features into specific score (reward) values. In terms of workflow, the scoring model constructs a mapping relationship between case features and human evaluations in the input dataset, captures semantic associations, analyzes these existing feature data through a multilayer perceptron, and evaluates the scoring results of the answers in the manufacturability analysis indicators.

[0127] Process C3: Input the sample feature vector into the initial scoring model to obtain the manufacturability score, draft angle, and risk level.

[0128] Process C4: Calculate the loss function between the manufacturability score, draft angle, and risk level and the actual manufacturability score, actual draft angle, and actual risk level corresponding to the sample feature vector.

[0129] Process C5: If the loss function converges, then mark the initial scoring model as the scoring model.

[0130] Procedure C6: If the loss function does not converge, modify the parameters of the initial scoring model and return to procedure C3.

[0131] It should be further explained that during training, a feature vector composed of draft area features, thickness attributes, material properties, and structural features is used as the model input. Simultaneously, the comprehensive draft direction score corresponding to this feature vector, the recommended minimum draft angle parameter for different structural types, and the draft risk level are used as supervision labels. In this way, the initial scoring model is jointly trained across multiple tasks, enabling it to simultaneously approximate both the human scoring function and the empirical rules of draft angles, and to learn the correlations between various evaluation indicators.

[0132] In this embodiment of the invention, the respective advantages of the BRep model and the triangular mesh representation model are fully utilized, comprehensively considering the material properties, geometric features, draft scheme, and specific manufacturing process conditions of the casting. Combined with a deep learning model, the draft direction and draft angle are intelligently evaluated, and a detailed manufacturability analysis report is generated, thereby improving the accuracy and automation of casting draft scheme design. This effectively optimizes the draft scheme, improves production efficiency, and reduces potential manufacturing risks.

[0133] Corresponding to the intelligent die-casting manufacturability analysis method provided in the above embodiments of the present invention, see also... Figure 6 The diagram shows a structural block diagram of an intelligent die-casting manufacturability analysis device provided in an embodiment of the present invention.

[0134] The device includes: a conversion model representation unit 601, a first construction unit 602, a definition unit 603, a first determination unit 604, a second determination unit 605, a selection unit 606, a second construction unit 607, and a scoring unit 608.

[0135] The transformation model representation unit 601 is used to convert the parametric geometric surfaces of the casting model into corresponding triangular meshes, and to calculate the structural features of each parametric geometric surface of the casting model based on the uniform triangular mesh corresponding to each parametric geometric surface.

[0136] The first building unit 602 is used to calculate the principal axis of inertia of the casting model and the principal axis of the bounding box of the casting model based on the uniform triangular mesh corresponding to each parameterized geometric surface, and to perform rotational transformation based on the principal axis of inertia and the principal axis of the bounding box to construct a candidate set of draft directions.

[0137] The definition unit 603 is used to define any draft direction in the candidate set as the moving mold opening direction, define the opposite direction of the draft direction as the stationary mold opening direction, and perform ray detection based on the moving mold opening direction and the stationary mold opening direction to obtain a set of moldable facets.

[0138] The first determining unit 604 is used to determine the moldable area and the under-molded area based on any moldable surface in the set of moldable surfaces and the mold opening direction; the mold opening direction is either the moving mold opening direction or the stationary mold opening direction.

[0139] The second determining unit 605 is used to determine the candidate area and candidate direction set of the slider to be arranged based on the under-opened mold area and the unopenable mold area; the unopenable mold area is the area that cannot be reached by either the moving mold opening direction or the stationary mold opening direction.

[0140] The selection unit 606 is used to select a target slider candidate direction from the slider candidate direction set, and use the target slider candidate direction as the core-pulling direction of the slider corresponding to the candidate area of ​​the slider to be arranged; the target slider candidate direction is the slider candidate direction that minimizes the total area of ​​the under-opened area or the area that cannot be opened, and maximizes the minimum draft angle value.

[0141] The second building unit 607 is used to construct a feature vector of the draft direction based on the material properties of the casting model, the structural features of each parameterized geometric surface, the moldable area, the under-molded area, the moldable area, the candidate area of ​​the slider to be arranged, and the core-pulling direction of the slider corresponding to the candidate area of ​​each slider to be arranged.

[0142] The scoring unit 608 is used to input the feature vector of each automatic draft direction into the scoring model to obtain the results of multiple draft manufacturability indices for each automatic draft direction, and generate and output an analysis report based on the results of multiple draft manufacturability indices.

[0143] In this embodiment of the invention, the respective advantages of the BRep model and the triangular mesh representation model are fully utilized, comprehensively considering the material properties, geometric features, draft scheme, and specific manufacturing process conditions of the casting. Combined with a deep learning model, the draft direction and draft angle are intelligently evaluated, and a detailed manufacturability analysis report is generated, thereby improving the accuracy and automation of casting draft scheme design. This effectively optimizes the draft scheme, improves production efficiency, and reduces potential manufacturing risks.

[0144] Combination Figure 6 The content shown is represented by the conversion model representation unit 601, which includes an acquisition module and a recognition module.

[0145] The acquisition module is used to acquire point cloud data and topological connection information in the uniform triangular mesh corresponding to each parametric geometric surface.

[0146] The recognition module is used to input all point cloud data and topological connection information into the recognition model and output the structural features of each parameterized geometric surface.

[0147] Combination Figure 6 The content shown, the first building unit 602, includes: a deflection module and a building module.

[0148] The deflection module is used to uniformly rotate the spindle of the inertia spindle and the spindle of the bounding box around the spindle at a preset angle to obtain each axis.

[0149] The building module is used to construct a candidate set of draft directions based on the principal axis of inertia and the principal axis of the bounding box, as well as the various axes.

[0150] Combination Figure 6The content shown defines unit 603, which includes: a transmission module, a calculation module, and a determination module.

[0151] The emitting module is used to emit parallel rays into the interior of the casting model along the opening directions of the moving mold and the stationary mold.

[0152] The calculation module is used to calculate the intersection points between each parallel ray and the triangular mesh of each parameterized geometric surface, and to mark the visible side of the triangular facet in the triangular mesh at each intersection point as a moldable facet.

[0153] The determination module is used to determine the set of openable mold surfaces corresponding to the draft direction based on all openable mold surfaces.

[0154] Combination Figure 6 The first determining unit 604, as shown, is specifically used for: calculating the draft angle for any openable moldable surface in the set of openable moldable surfaces based on the normal and mold opening direction at the three vertices of the openable moldable surface; when the draft angle is greater than a preset threshold, marking the openable moldable surface at the intersection point as an openable moldable region; when the draft angle is not greater than the preset threshold, marking the openable moldable surface at the intersection point as an under-molded region.

[0155] Combination Figure 6 As shown, the second determining unit 605 is specifically used for: extracting the connected regions of the under-molded region and the un-molded region based on the adjacency relationship of the under-molded region and the un-molded region on the triangular mesh; marking each connected region as a candidate region for the slider to be arranged; and for any candidate region for the slider to be arranged, using the normal directions of all triangular faces contained in the candidate region as the slider candidate directions to obtain a set of slider candidate directions.

[0156] Combination Figure 6 The device, as shown, also includes: an initialization unit, an input unit, a calculation unit, a marking unit, and a modification unit.

[0157] The initialization unit is used to initialize the deep learning model and obtain the initial scoring model.

[0158] The input unit is used to input the sample feature vector into the initial scoring model to obtain the manufacturability score, draft angle, and risk level.

[0159] The calculation unit is used to calculate the loss function between the manufacturability score, draft angle, and risk level and the actual manufacturability score, actual draft angle, and actual risk level corresponding to the sample feature vector.

[0160] The labeling unit is used to label the initial scoring model as the scoring model if the loss function converges.

[0161] The modification unit is used to modify the parameters of the initial scoring model if the loss function fails to converge, and then return to the execution input unit.

[0162] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, for system or system embodiments, since they are basically similar to method embodiments, the description is relatively simple, and relevant parts can be referred to the descriptions in the method embodiments. The systems and system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0163] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0164] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method of intelligent die cast manufacturability analysis, characterized by, The method includes: The parametric geometric surfaces of the casting model are converted into corresponding uniform triangular meshes, and the structural features of each parametric geometric surface of the casting model are calculated based on the uniform triangular meshes corresponding to each parametric geometric surface. Based on the uniform triangular mesh corresponding to each parametric geometric surface, the principal axis of inertia of the casting model and the principal axis of the bounding box of the casting model are calculated, and a rotational transformation is performed based on the principal axis of inertia and the principal axis of the bounding box to construct a candidate set of draft directions. For any draft direction in the candidate set, the draft direction is defined as the moving mold opening direction, and the opposite direction of the draft direction is defined as the stationary mold opening direction. Based on the moving mold opening direction and the stationary mold opening direction, ray detection is performed to obtain a set of moldable facets. Based on any one of the openable moldable surfaces in the set of openable moldable surfaces and the mold opening direction, the openable moldable area and the under-openable moldable area are determined; the mold opening direction is either the moving mold opening direction or the stationary mold opening direction. The candidate areas and candidate directions of the slider to be arranged are determined based on the under-opened mold area and the unopenable mold area; the unopenable mold area is the area that cannot be reached by both the moving mold opening direction and the stationary mold opening direction. Select a target slider candidate direction from the set of slider candidate directions, and use the target slider candidate direction as the core-pulling direction of the slider corresponding to the candidate area of ​​the slider to be arranged; the target slider candidate direction is the slider candidate direction that minimizes the total area of ​​the under-opened area or the unopenable area and maximizes the minimum draft angle value. Based on the material properties of the casting model, the structural features of each parametric geometric surface, the moldable area, the under-molding area, the mold-unopenable area, the candidate area of ​​the slider to be arranged, and the core-pulling direction of the slider corresponding to each candidate area of ​​the slider to be arranged, a feature vector of the draft direction is constructed. The feature vectors of each draft direction are input into the scoring model to obtain multiple draft manufacturability index results for each draft direction. An analysis report is generated and output based on the multiple draft manufacturability index results.

2. The method of claim 1, wherein, The calculation of the structural features of each parametric geometric surface of the casting model based on the uniform triangular mesh corresponding to each parametric geometric surface includes: Obtain point cloud data and topological connectivity information from the uniform triangular mesh corresponding to each parametric geometric surface; All point cloud data and topology connection information are input into the recognition model, and the structural features of each parameterized geometric surface are output.

3. The method according to claim 1, characterized in that, Rotational transformations are performed based on the principal axis of inertia and the principal axis of the bounding box to construct a candidate set of draft directions, including: Around the inertial main axis and the main axis of the enclosure box, a uniform rotation transformation is performed according to a preset angle to obtain each axis. Based on the principal axis of inertia and the principal axis of the bounding box, as well as the various axes, a candidate set of draft directions is constructed.

4. The method according to claim 1, characterized in that, The ray-firing inspection based on the opening direction of the moving mold and the opening direction of the stationary mold yields a set of openable moldable facets, including: Parallel rays are emitted into the interior of the casting model along the opening direction of the moving mold and the opening direction of the stationary mold. Calculate the intersection points between each parallel ray and the triangular mesh of each parameterized geometric surface, and mark the visible side of the triangular facet in the triangular mesh where each intersection point is located as the moldable facet. The set of openable moldable surfaces corresponding to the draft direction is determined based on all openable moldable surfaces.

5. The method according to claim 4, characterized in that, The step of determining the moldable area and the under-molding area based on any moldable surface in the set of moldable surfaces and the mold opening direction includes: For any openable moldable surface in the set of openable moldable surfaces, calculate the draft angle based on the normal at the three vertices of the openable moldable surface and the mold opening direction; When the draft angle is greater than a preset threshold, the moldable surface at the intersection point is marked as a moldable area; When the draft angle is not greater than a preset threshold, the moldable surface at the intersection point is marked as an under-molded area.

6. The method according to claim 1, characterized in that, The step of determining the candidate region and candidate direction set of the slider to be arranged based on the under-molding region and the un-molding region includes: Based on the adjacency relationship between the under-molded area and the un-molded area on the triangular mesh, the connected regions of the under-molded area and the un-molded area are extracted; Each of the connected regions is marked as a candidate region for the slider to be placed; For any candidate region of the slider to be placed, the normal directions of all triangular facets contained in the candidate region of the slider to be placed are used as the candidate directions of the slider, and a set of candidate directions of the slider is obtained.

7. The method according to claim 1, characterized in that, The training process of the scoring model includes: Obtain historical casting sample data, which includes sample feature vectors; Initialize the deep learning model to obtain the initial scoring model; The sample feature vector is input into the initial scoring model to obtain the manufacturability score, draft angle, and risk level; Calculate the loss function between the manufacturability score, the draft angle, and the risk level, and the actual manufacturability score, actual draft angle, and actual risk level corresponding to the sample feature vector; If the loss function converges, the initial scoring model is marked as a scoring model; If the loss function does not converge, the parameters of the initial scoring model are modified, and the process returns to the step of inputting the sample feature vector into the initial scoring model to obtain the manufacturability score, draft angle, and risk level.

8. An intelligent die-casting manufacturability analysis device, characterized in that, The device includes: The transformation model representation unit is used to convert the parametric geometric surfaces of the casting model into corresponding uniform triangular meshes, and to calculate the structural features of each parametric geometric surface of the casting model based on the uniform triangular meshes corresponding to each parametric geometric surface. The first construction unit is used to calculate the principal axis of inertia of the casting model and the principal axis of the bounding box of the casting model based on the uniform triangular mesh corresponding to each parameterized geometric surface, and to perform rotational transformation based on the principal axis of inertia and the principal axis of the bounding box to construct a candidate set of draft directions. A definition unit is used to define any draft direction in the candidate set as the moving mold opening direction, define the opposite direction of the draft direction as the stationary mold opening direction, and perform ray detection based on the moving mold opening direction and the stationary mold opening direction to obtain a set of moldable facets. The first determining unit is used to determine the moldable area and the under-molding area based on any moldable surface in the set of moldable surfaces and the mold opening direction; the mold opening direction is the moving mold opening direction or the stationary mold opening direction. The second determining unit is used to determine the candidate area and candidate direction set of the slider to be arranged based on the under-opened mold area and the unopenable mold area; the unopenable mold area is the area that cannot be reached by both the moving mold opening direction and the stationary mold opening direction. The selection unit is used to select a target slider candidate direction from the set of slider candidate directions, and use the target slider candidate direction as the core-pulling direction of the slider corresponding to the candidate area of ​​the slider to be arranged; the target slider candidate direction is the slider candidate direction that minimizes the total area of ​​the under-opened area or the unopenable area and maximizes the minimum draft angle value. The second construction unit is used to construct the feature vector of the draft direction based on the material properties of the casting model, the structural features of each parameterized geometric surface, the moldable area, the under-molding area, the mold-unopenable area, the candidate area of ​​the slider to be arranged, and the core-pulling direction of the slider corresponding to each candidate area of ​​the slider to be arranged. The scoring unit is used to input the feature vector of each draft direction into the scoring model to obtain multiple draft manufacturability index results for each draft direction, and generate and output an analysis report based on the multiple draft manufacturability index results.

9. The apparatus according to claim 8, characterized in that, The transformation model representation unit includes: The acquisition module is used to acquire point cloud data and topological connectivity information in the uniform triangular mesh corresponding to each parameterized geometric surface; The recognition module is used to input all point cloud data and topological connection information into the recognition model and output the structural features of each parameterized geometric surface.

10. The apparatus according to claim 8, characterized in that, The first building unit includes: The deflection module is used to uniformly rotate around the inertial spindle and the spindle of the enclosure box according to a preset angle to obtain various axes; A construction module is used to construct a candidate set of draft directions based on the inertial axis, the bounding box axis, and various axes.

Images