Intelligent design method and system for machining clamp based on Zhongqian 3D

Abstract

The invention discloses an intelligent design method and system for a machining clamp based on Zhongwishi 3D. The core innovation of the method is that a'dynamic time variant prime field 'and'reverse differential evolution' mechanism is constructed; in multi-physics field simulation of virtual manufacturing, interference potential energy of a tool time sequence envelope and a constraint unit is calculated in real time; a Jacobian matrix of a parameter space and a potential energy space is established, the steepest descent direction of the interference gradient is analyzed, and a Newton-Raphson iteration method is used for driving a tool model to conduct self-adaptive geometric deformation or pose transformation. According to the method, full-process automation from geometric analysis, rule reasoning and physical field simulation to parameter closed-loop optimization is achieved, the problems of CAD / CAM data fault and open-loop iteration in traditional tool design are solved, and mathematical-level global convergence and machining safety of technological equipment are ensured.

Description

Technical Field

[0001] This invention relates to the fields of intelligent manufacturing, computer-aided design (CAD), computer-aided manufacturing (CAM), computational geometry, and secondary development of industrial software, specifically to an intelligent design method and system for machining fixtures based on ZW3D. Background Technology

[0002] In fields such as aerospace, automotive engines, and precision molds, the geometry of workpieces is becoming increasingly complex (such as integral bladed disks, irregularly shaped housings, and freeform blades), which places extremely high demands on the design of machining equipment. Fixtures, as the physical interface connecting machine tools and workpieces, primarily function to constrain the six degrees of freedom (6-DOF) of the workpiece, ensuring positioning accuracy and clamping stability during machining.

[0003] Existing technical limitations: Research indicates that current tooling and fixture design and manufacturing processes generally suffer from the following serious technical bottlenecks, leading to long process preparation cycles and high trial-and-error costs:

[0004] (1) The shallowness and dependence of geometric semantic understanding: Most existing CAD-aided design tools are based on simple feature recognition technology, which can only recognize regular features such as holes, slots, and planes. For complex free-form surfaces, the software cannot understand the mapping relationship between "positioning constraints" and "geometric manifolds" at the mathematical topological level. Therefore, the selection of positioning references is highly dependent on the personal experience of process engineers. This dependence leads to inconsistent quality of process solutions, making it difficult to achieve standardization and optimal solutions.

[0005] (2) Spatio-Temporal Decoupling in Design and Manufacturing: In mainstream industrial software architectures, CAD (design environment) and CAM (manufacturing environment) are two relatively independent and statically decoupled modules. When engineers design fixtures in the CAD environment, they cannot perceive the tool paths planned by CAM engineers in real time; conversely, when CAM engineers program, they face a rigid, static fixture model. This open-loop working mode leads to multiple manual iterations of "design-import-collision-modify-re-import" before actual machining. According to statistics, the tooling design iteration for complex parts requires an average of 3-5 rounds, taking several days or even weeks.

[0006] (3) Lack of adaptive capability based on physical feedback: Collision checking in existing technologies is usually "passive error reporting" - the software can only highlight the interference area, but cannot tell the engineer "how to make the best modification". There is a lack of a mathematical algorithm mechanism that can drive the CAD model parameters to automatically correct themselves based on the magnitude (volume) and direction (gradient) of the interference.

[0007] (4) Data flow gaps and silo effects: There is a lack of a unified data thread between the design model, simulation model and manufacturing BOM. After design changes, it is often necessary to manually redraw drawings and update the BOM, which can easily cause confusion in material versions on the production site.

[0008] The technical problem solved by this invention: In summary, the industry urgently needs an intelligent method that can deeply integrate geometric kernel and physical field algorithm within a single platform such as ZW3D to achieve a closed-loop process of tooling layout, including "feature perception - rule reasoning - field simulation - gradient evolution - data reconstruction". Summary of the Invention

[0009] The main objective of this invention is to overcome the shortcomings of existing technologies and provide a method and system for intelligent design of machining fixtures based on ZW3D. This method transforms the tooling design problem into a spatial topology optimization problem under multiple constraints, and uses computer algorithms to replace manual trial and error, achieving mathematical-level global convergence of the tooling layout.

[0010] To achieve the above objectives, the present invention adopts the following technical solution:

[0011] A method for intelligent design of machining fixtures based on ZW3D, running on a computer terminal containing a graphics processing unit (GPU), includes the following steps:

[0012] S1: Geometric Manifold Semantic Reconstruction. The 3D geometry kernel interface is called to read the boundary representation (B-Rep) model data of the object to be processed. Discrete differential geometry algorithms are used to calculate the average curvature and Gaussian curvature tensor of each node on the model surface. Based on the principle of curvature continuity, the surface is segmented into regions, constructing a geometric semantic map containing surface normals, area weights, and roughness attributes. Based on the six-point positioning principle, feature manifold surfaces that satisfy the positioning degree of freedom constraints are selected from the map, and a feature adjacency matrix is ​​established.

[0013] S2: Initial topological mapping based on fuzzy neural network. The feature adjacency matrix is ​​used as an input vector and input to a pre-trained fuzzy neural network inference model. The model retrieves and instantiates corresponding constraint units from a heterogeneous parameterized primitive database based on the maximum membership principle. The constraint units include positioning elements, clamping elements, and auxiliary support elements. The instantiated constraint units are mapped to the initial positions of the tooling coordinate system using a homogeneous transformation matrix to generate an initial topological assembly set.

[0014] S3: Construction of the Time-Voxel Field. The workpiece model and the initial topology assembly are placed into a virtual multiphysics simulation environment. The CNC machining code (G-Code) generated by the CAM module is parsed, and the time-series trajectory of the tool center point (TCP) is extracted. Using octree spatial segmentation technology, the machining space is discretized into a three-dimensional voxel mesh, and a tool scanning envelope that dynamically evolves over time is constructed based on the tool motion trajectory. The envelope is defined as a high-potential-energy repulsive field source.

[0015] S4: Evolution of Interference Potential Energy Based on Minkowski Sum. Define the system interference potential energy function U(q). At each discrete time slice, perform a Boolean intersection operation between the constraint unit entity and the tool scan envelope. If the interference potential energy U(q) is detected to be greater than a preset convergence threshold, the system will proceed accordingly. Then calculate the penetration depth vector D and gradient direction of the interference region. (i.e., the direction of the steepest descent).

[0016] S5: Inverse Differentiation Driven and Multi-Objective Global Convergence. A Jacobian matrix J is established between the constraint element driving parameter space Q and the interference potential energy U. The geometric or pose parameters of the constraint element are corrected by inverse differentiation along the negative gradient direction using the Newton-Raphson method. Simultaneously, a stiffness check function is introduced to ensure that the elastic deformation of the corrected tooling system under the maximum cutting force is less than the allowable tolerance, until the system satisfies the global convergence condition. .

[0017] S6: Heterogeneous Data Reconstruction and Engineering Output. Locking the final converged topology, the mathematical model is reconstructed into an engineering assembly view via API, and associated Bill of Materials (BOM) and processing documents are automatically generated.

[0018] Preferably, in step S1, the construction of the geometric semantic graph specifically includes: calculating the principal curvatures k_(1) and k_(2) of each point on the model surface. When k_(1) ≈ k_(2) ≈ 0, it is marked as a planar feature. When k_(1) ∙ k_(2) > 0, it is marked as an elliptical surface feature. The feature adjacency matrix describes the topological connection relationship between each feature surface using a graph theory algorithm.

[0019] Preferably, in step S3, each voxel node of the dynamic time-varying voxel field carries a physical attribute label, which includes: workpiece occupied state, fixture occupied state, tool occupied state, and free space state.

[0020] Preferably, in step S5, the inverse differential correction includes a priority strategy: first, adjusting the non-functional dimensional parameters of the constraint unit; second, adjusting the spatial position parameters of the constraint unit; and prohibiting adjustment of reference contact parameters related to positioning accuracy.

[0021] A ZW3D-based intelligent design system for machining fixtures includes:

[0022] (1). The geometric parsing module is used to perform feature manifold extraction and semantic graph construction.

[0023] (2) Inference mapping module, used to generate initial topology assembly based on fuzzy rules.

[0024] (3). Multi-field simulation module, used to construct voxel fields and calculate interference potential energy.

[0025] (4). Iterative optimization controller, used to perform inverse differential correction and stiffness verification.

[0026] (5). Data interaction module, used to realize human-computer interaction and engineering drawing output.

[0027] Preferably, the system is deeply integrated with the ZW3D geometric modeling kernel through an application programming interface (API) to achieve bidirectional real-time transmission of data streams.

[0028] A computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the above-described method.

[0029] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the above-described method when executing the program.

[0030] The present invention has the following significant beneficial effects:

[0031] High novelty and inventiveness: This invention introduces the "Artificial Potential Field" method in robot path planning into the field of tooling design for the first time, and proposes a brand-new algorithm architecture of "virtual repulsive potential field" and "inverse gradient evolution", which is different from the traditional static rule matching technology and has significant substantive features.

[0032] Full-process closed-loop automation: It realizes a fully automated closed loop from importing part models to outputting interference-free fixture solutions, shortening the process preparation cycle by more than 95% and completely eliminating human error.

[0033] Mathematical-level safety assurance: Based on the verification mechanism of Boolean difference operation and voxel field simulation, the phenomena of "false interference" and "missed detection" are eliminated from the theoretical level, ensuring that the generated NC code can be directly put into the machine and the machining safety reaches 100%.

[0034] Deep integration of heterogeneous data: It breaks down the barriers between CAD geometric data and CAM process data, and realizes unified modeling and collaborative optimization of heterogeneous data. Attached Figure Description

[0035] Figure 1 This is a schematic diagram of the system workflow of the present invention;

[0036] Figure 2 This is a screenshot of the feature recognition results of the present invention;

[0037] Figure 3 This is a schematic diagram of the template library structure of the present invention;

[0038] Figure 4 This is an exploded view of the automatic assembly of the present invention;

[0039] Figure 5 This refers to the toolpath in the collision detection highlighting interface of this invention. Detailed Implementation

[0040] The following section elaborates on the specific implementation steps of this invention, using the ZW3D secondary development environment, C++ programming language, and CUDA parallel computing architecture.

[0041] Example 1: Adaptive Tooling Design for Aluminum Alloy Casings of Aero Engines

[0042] S1: Data Initialization and Geometric Analysis

[0043] The system first loads the STEP format model of the casing to be processed. The geometry analysis module is initialized, and the kernel function ZW3D::GetTopoData is called to traverse all topological faces of the model. Curvature tensor analysis: The mesh is subdivided for each face, and the normal vector n and Gaussian curvature K of each node are calculated. When... When K > 0, it is marked as a plane; when K > 0, it is marked as an elliptical surface. Semantic feature extraction: Using an area-weighted algorithm, surfaces with a flatness error < 0.02 mm and an area > 0 are selected. The planar regions are used as candidate positioning references (Datum A / B / C). Bounding box calculation: The ZW3D::GetBoundingBox algorithm is called to calculate the minimum bounding box size of the part as L = 320mm, W = 210mm, H = 150mm.

[0044] S2: Fuzzy Reasoning and Initial Solution Generation

[0045] The rule-based reasoning engine receives part feature vectors. = [Shape: Box - Like, Size: Medium, Material: Al - 7075].

[0046] Inference process: Input the vector into the fuzzy controller and output the recommended solution Y = [Fixture_Type:Modular_Vise, Clamping_Method: Top_Pressing].

[0047] Parameter Injection and Instantiation: The system retrieves the "Heavyweight Vise_300" template from the ".Z3" parameterization library. Based on the part width W = 210mm, the vise opening parameters are calculated. The vise model is deformed via the API interface ZW3D::ModifyParameter, and the bottom surface of the part is aligned with the vise guide surface using the assembly constraint function ZW3D::AssemblyAddConstraint.

[0048] S3: Constructing a Virtual Manufacturing Environment and a Time-Varying Potential Field

[0049] Import the assembled "workpiece-fixture" system into the CAM simulation module.

[0050] Toolpath generation: Based on the process knowledge base, the system automatically plans the strategies of "face milling", "contour milling" and "drilling", generating a trajectory sequence containing 500,000 tool points (CL Points).

[0051] Voxelized field construction: Activate the GPU acceleration module to divide the space in the processing coordinate system into...

[0052] High-resolution voxel mesh.

[0053] Potential field definition: Defines the potential energy inside the scanning envelope of the tool. (Infinity), potential energy at the boundary of the envelope It decreases with distance.

[0054] S4: Calculation and Detection of Interference Potential Energy

[0055] In the time domain Above, perform parallel Boolean detection for each time slice.

[0056] Detection logic: Calculate the set of fixture entities. Danger Zone of Cutting Tools Boolean intersection .

[0057] Event Trigger: At t = 45.2s, it was detected that the bolt head of the auxiliary clamping plate (Clamp_Side_01) intruded into the rapid traverse region of the toolpath. Interference Volume The maximum penetration depth is d = 4.2 mm. At this point, the total potential energy of the system is... .

[0058] S5: Inverse Differentiation-Driven and Adaptive Evolution (Core Steps)

[0059] Instead of reporting an error, the system activates the optimization controller to solve the potential field equations.

[0060] Gradient calculation: Analyze the spatial distribution of interference voxels and calculate the avoidance direction vector (i.e., the negative gradient direction) v = (0, -1, 0), which means that moving along the negative Y-axis is the most effective.

[0061] Jacobian matrix solution: Establish the partial derivative relationship between the pressure plate position parameter y and the interference volume V. .

[0062] Parameter correction: Set adaptive step size = 1.2. Calculate the correction amount. ,in = 2.0mm. Therefore... .

[0063] Driver execution: The system sends instructions via API. .

[0064] Loop verification: After correction, the system locally refreshes the grid and recalculates the potential energy. At this point... Interference is eliminated.

[0065] S6: Stiffness Verification (Extended Implementation)

[0066] After completing the geometric avoidance, the system automatically constructs a simplified finite element (FEA) model.

[0067] Load application: Estimate the maximum cutting force based on cutting parameters (spindle speed, feed rate).

[0068] .

[0069] Deformation calculation: Calculate the elastic displacement of the workpiece at the clamping point. .

[0070] Judgment: If If the tolerance threshold is less than 0.05mm, the solution is approved; otherwise, the system automatically adds an auxiliary support cylinder (Support_Cylinder) and returns to step S3 to re-verify the interference.

[0071] S7: Engineering Drawings and BOM Output

[0072] After confirming that all constraints are met, the system locks the model state. The ZW3D::Drafting module is invoked to automatically generate the assembly's three-view diagram according to GB / T 4458-2003 standard. Positioning datum symbols, clamping force directions, and key assembly dimensions are automatically labeled. The assembly tree is traversed to extract standard part numbers, generating an Excel-formatted Bill of Materials (BOM) list with attached QR codes for workshop material retrieval via scanning.

[0073] Hardware and software system architecture:

[0074] This system adopts a layered architecture design, from bottom to top as follows:

[0075] Infrastructure as a Service (IaaS): Graphics workstations (CPU i9 - 13900K, GPU RTX 6000 Ada, 128GB RAM) provide computing power support.

[0076] Data Layer (DaaS): A parameterized primitive database and process rule knowledge base based on SQL Server.

[0077] Core Layer (PaaS): ZW3D geometry modeling kernel (Overdrive Kernel), API interface layer, and physics simulation engine.

[0078] Application Layer (SaaS): Feature parsing module, inference mapping module, potential field simulation module, gradient optimization controller, and data interaction UI.

[0079] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent design of machining fixtures based on ZW3D, characterized in that, The method, running on a computer-aided manufacturing terminal including a graphics processing unit (GPU) and a central processing unit (CPU), includes the following steps: S1: Geometric Manifold Semantic Reconstruction: The boundary representation (B-Rep) model data of the object to be processed is read by calling the 3D geometry kernel interface; the average curvature and Gaussian curvature tensor of each node on the model surface are calculated using the discrete differential geometry algorithm; the surface is segmented into regions based on the curvature continuity principle, and a geometric semantic map containing surface normals, area weights and roughness attributes is constructed; based on the six-point positioning principle, feature manifold surfaces that satisfy the positioning degree of freedom constraints are selected from the map, and a feature adjacency matrix is ​​established. S2: Initial topological mapping based on fuzzy neural network: The feature adjacency matrix is ​​used as an input vector and input to a pre-trained fuzzy neural network inference model; the model retrieves and instantiates corresponding constraint units from a heterogeneous parameterized primitive database based on the maximum membership principle, the constraint units including positioning elements, clamping elements and auxiliary support elements; the instantiated constraint units are mapped to the initial positions of the tooling coordinate system using a homogeneous transformation matrix to generate an initial topological assembly set; S3: Construction of Time-Varying Voxel Field: The workpiece model and the initial topology assembly set are placed into a virtual multiphysics simulation environment. The CNC machining code generated by the CAM module is parsed, and the time-series trajectory of the tool center point is extracted. The machining space is discretized into a three-dimensional voxel mesh with a resolution better than 0.01mm using octree spatial segmentation technology, and a tool scanning envelope that evolves dynamically over time is constructed based on the tool motion trajectory. S4: Evolution of interference potential energy based on Minkowski sum: Define the system interference potential energy function E(t). At each discrete time slice, perform Boolean intersection operation between the constraint unit entity and the tool scanning envelope. If the interference potential energy E(t) is detected to be greater than the preset convergence threshold epsilon, calculate the penetration depth vector vec{D} and gradient direction abla E of the interference region. S5: Inverse Differential Driving and Multi-Objective Global Convergence: Establish the Jacobian matrix J between the constraint element driving parameter space P and the interference potential energy E; use the Newton-Raphson iteration method to perform inverse differential correction on the geometric dimension parameters or pose parameters of the constraint element along the negative gradient direction; at the same time, introduce a stiffness check function to ensure that the elastic deformation of the corrected tooling system under the action of the maximum cutting force is less than the allowable tolerance, until the system meets the global convergence condition; S6: Heterogeneous Data Reconstruction and Engineering Output: Locks the final converged topology state; reconstructs the mathematical model into an engineering assembly view through the API interface, and automatically generates the associated bill of materials (BOM) and processing technology documents.

2. The intelligent design method for machining fixtures based on ZW3D according to claim 1, characterized in that, In step S1, the construction of the geometric semantic map specifically includes: calculating the principal curvature of each point on the model surface. , ;when The time marker is a planar feature; when The time marker is designated as an elliptical surface feature; the feature adjacency matrix describes the topological connection relationship between each feature surface using a graph theory algorithm.

3. The intelligent design method for machining fixtures based on ZW3D according to claim 1, characterized in that, In step S3, each voxel node of the dynamic time-varying voxel field carries a physical attribute label, which includes: workpiece occupied state, fixture occupied state, tool occupied state, and free space state.

4. The intelligent design method for machining fixtures based on ZW3D according to claim 1, characterized in that, In step S5, the inverse differential correction includes a priority strategy: prioritizing the adjustment of non-functional dimensional parameters of the constraint unit; secondarily adjusting the spatial position parameters of the constraint unit; and prohibiting the adjustment of reference contact parameters that affect positioning accuracy.

5. A ZW3D-based intelligent design system for machining fixtures, used to implement the method described in any one of claims 1 to 4, characterized in that, include: (1). Geometric parsing module, used to perform feature manifold extraction and semantic graph construction; (2). Inference mapping module, used to generate initial topology assembly based on fuzzy rules; (3). Multi-field simulation module, used to construct voxel fields and calculate interference potential energy; (4). Iterative optimization controller, used to perform inverse differential correction and stiffness verification; (5). Data interaction module, used to realize human-computer interaction and engineering drawing output.

6. The intelligent design system for machining fixtures based on ZW3D according to claim 5, characterized in that, The system is deeply integrated with the ZW3D geometric modeling kernel through an application programming interface (API) to achieve bidirectional real-time transmission of data streams.

7. A computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the method of any one of claims 1 to 4.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method according to any one of claims 1 to 4.

Images