Method for automatically generating center plane of cooling water channel of hot forming insert based on three-dimensional modeling

Abstract

The present application relates to the field of computer aided design and manufacturing, and discloses a hot forming insert cooling water channel center surface automatic generation method based on three-dimensional modeling, which comprises the following steps: identifying a mold working surface and constructing a water channel constraint parameter set based on a thermal load density parameter; converting the constraint parameter into a topological connection weight, and generating an initial center trajectory line set in a continuous symbol distance field; after generating an initial grid model through cross-section circle skinning, adopting a Bezier surface sheet and performing iterative fitting based on a gradient descent mechanism, including normal and deformation penalty, and outputting a smooth center surface model; and finally performing wall thickness interference detection and triggering local trajectory line reconstruction. The present application realizes the self-adaptation of cooling water channel layout to thermal load, avoids geometric interference while guaranteeing cooling efficiency, and outputs mold water channel geometric data which can be directly manufactured.

Description

Technical Field

[0001] This application belongs to the interdisciplinary field of computer-aided design and manufacturing, specifically involving a method for automatically generating the center surface of cooling water channels for thermoformed inserts based on 3D modeling. Background Technology

[0002] High-strength steel sheet hot forming is a core manufacturing method for overcoming the limitations of lightweighting and passive safety in automotive bodies. High-strength boron alloy steel sheets within the austenitizing temperature range are transferred to a stamping die for closed forming. The die insert, as the contact component directly bearing mechanical and thermal loads, must undergo rapid quenching and cooling of the hot-formed part within the holding pressure cycle. The topology, spatial layout, and solid wall thickness distribution of the cooling water channels within the insert directly determine the heat transfer efficiency of the fluid medium and the overall thermal resistance network of the die, thus influencing the uniformity of the martensitic phase transformation within the formed workpiece and the tensile strength of the final structural component.

[0003] Extracting and defining the three-dimensional center surface of the cooling water channel is the initial step in realizing the geometric design of the insert. The extracted center surface is used as the geometric reference track for constructing the fluid channel entity. Current mold design methods often exhibit a high degree of dependence on the subjective experience of engineers. Two-dimensional manual offset based on section sketches is difficult to handle the boundaries of irregularly shaped thermoformed inserts with large chamfers and hyperbolic shapes. When the center trajectory surface generated by conventional parametric translation encounters a highly curved working surface, it is prone to normal vector twisting, local self-intersection of the trajectory surface, and abrupt changes in cusps.

[0004] Existing technologies utilize Boolean operations to directly generate conformal water channel design spaces from the inlet and outlet selection surfaces of the mold model. There are also related studies on generating offset curved surfaces based on surfaces and projecting spiral curves to construct scanning channel entities. These methods establish a technical framework for generating water channel trajectories based on surface geometric equidistant mapping of conventional injection mold cores. However, thermoforming inserts face highly uneven and unsteady heat flux impacts, and simple geometric equidistant offsets ignore the differences in local surface heat load density. If cooling water channels are planned with conventional constant spacing in deep cavity corner areas where high heat loads converge, local heat transfer bottlenecks are easily induced. Surface generation mechanisms relying on direct offsetting of discrete meshes lack geometric continuity constraints; the constructed center surface often fails to generate subsequent CNC machining toolpaths due to mesh quality degradation, and may even result in water channel hole wall penetration interference defects. Summary of the Invention

[0005] The purpose of this invention is to provide an automatic generation method for the center surface of cooling water channels in thermoformed inserts based on three-dimensional modeling. This method solves the technical problems of existing technologies, which are prone to trajectory self-intersection and interference when generating the center surface of cooling water channels on thermoformed inserts with variable curvature due to constant geometric equidistant offset, and are unable to adapt to the spatial jump characteristics of local thermal load.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0007] A method for automatically generating the center surface of cooling water channels for thermoformed inserts based on 3D modeling includes the following steps:

[0008] Step 1: Establish a three-dimensional geometric model of the thermoformed insert and identify the working surface area and the cooling cavity wall area;

[0009] Step II: Discretize the working surface area into a grid, extract the surface feature parameters of each grid node in the working surface area, and obtain the heat load density parameters of each grid node according to the preset heat load calculation rules.

[0010] Step III: Based on the surface feature parameters and the heat load density parameters, construct a set of waterway design constraint parameters. The set of waterway design constraint parameters includes at least the local waterway spacing parameters and the local target wall thickness parameters corresponding to each of the grid nodes.

[0011] Step IV: Establish a continuous symbolic distance field in the three-dimensional space where the working surface area is located. Based on the local target wall thickness parameter, convert the local waterway spacing parameter into topological connection weights in the symbolic distance field, and generate an initial trajectory line set of the waterway centerline based on the topological connection weights.

[0012] Step V: Establish a cross-sectional circle perpendicular to the tangent direction of the trajectory line along each trajectory line in the initial trajectory line set, obtain the topological matching relationship between adjacent cross-sectional circles, and perform surface skinning on all cross-sectional circles according to the topological matching relationship to generate the initial mesh model of the center surface;

[0013] Step VI: Geometrically fit the initial mesh model of the center surface using a Bézier surface patch with a preset polynomial order, construct an iterative objective function including a normal deviation penalty term and a mesh deformation penalty term, and adjust the control vertex coordinates of the Bézier surface patch using a gradient descent mechanism until the iterative objective function meets the convergence condition, and output a smooth center surface model.

[0014] Step VII: Perform wall thickness interference detection based on spatial voxels on the smoothed center surface model, identify the wall thickness warning area where the distance of the spatial entity is less than the local target wall thickness parameter, trigger local trajectory line reconstruction for the wall thickness warning area, and output the cooling water channel center surface data.

[0015] As a preferred embodiment of the present invention, in step II, the specific implementation process of discretizing the working surface region into a mesh is as follows:

[0016] The Delaunay triangulation algorithm is used to adaptively mesh the working surface region. When the rate of curvature change exceeds the set extreme value, the mesh edge length is halved and the mesh is refined.

[0017] The specific process of extracting surface feature parameters is as follows: using local surface quadratic polynomial fitting to obtain the first basic form and the constant coefficients of the second basic form of the surface, and then calculating the Gaussian curvature and average curvature of each grid node.

[0018] By combining thermodynamic boundary condition parameters, the steady-state heat conduction differential equation of the working surface region within a set quenching cycle is solved using the finite difference method, and the scalar heat flux absorbed by each grid node is calculated as the heat load density parameter.

[0019] As a preferred embodiment of the present invention, in step III, the mathematical relationship for constructing the local waterway spacing parameters is as follows:

[0020]

[0021] in, To be allocated to the first Local waterway spacing parameters for each grid node. Based on the baseline value of the basic spacing, For the first Heat load density parameters of each grid node This is the lower bound of the set of minimum values ​​for the heat load density parameters of all grid nodes. This is the upper bound of the set of maximum values ​​for the heat load density parameters of all grid nodes. This is the heat load sensitivity adjustment coefficient. For spatial curvature compensation weights, For the first The average curvature of each grid node It is the absolute value symbol;

[0022] By calculating the global arithmetic mean heat load density, and combining the geometric baseline value of the foundation wall thickness with the wall thickness safety redundancy factor, the load is mapped and generated to the [missing information]. The local target wall thickness parameters of each grid node.

[0023] Furthermore, in step IV, the process of establishing a continuous symbolic distance field includes:

[0024] Calculate the scalar symbolic distance value of any spatial coordinate point in three-dimensional free space relative to the continuous point set model of the working surface region;

[0025] A connected undirected graph is constructed within the symbolic distance field, with the spatial voxel center as the undirected graph node, and the topological connection weight dominates the edge weight of the undirected graph; the local waterway spacing parameter controls the step size of the path search or the neighborhood connection range to achieve the transformation;

[0026] The fast travel algorithm is used to solve the path cost minimization integral function, which includes wall thickness deviation penalty, bending energy convergence penalty and torsional energy penalty. The wall thickness deviation penalty term is calculated based on the continuous local target wall thickness field obtained by interpolation from the discrete local target wall thickness parameters. The continuous spatial discrete polyline in the cost convergence state is output to form the initial trajectory line set.

[0027] In step V, when obtaining the topological matching relationship between adjacent cross-sectional circles, a feature point alignment control process is executed:

[0028] A local orthogonal coordinate system is generated by discretely sampling nodes along the trajectory line using the Flyner frame to draw the cross-sectional circle, the diameter of which is fixed as the nominal diameter of the cooling water system;

[0029] A homogeneous rotation matrix is ​​constructed with the tangential vector of the current trajectory line as the rotation axis. The optimal rotation alignment adjustment angle is calculated to minimize the sum of the squared distances of the three-dimensional coordinates of the discrete feature points on the two cross-sectional circles. After correcting the phase using the optimal rotation alignment adjustment angle, quadrilateral patches are generated by non-uniform rational B-spline interpolation, and then spliced ​​to generate the initial mesh model of the center surface.

[0030] In step VI, the iterative objective function, which includes a normal deviation penalty term and a mesh deformation penalty term, is as follows:

[0031]

[0032] in, Quantize the numerical values ​​for the iterative objective function. The scalar represents the total number of surface sampling points of the initial mesh model on the central face. The fitted Bézier surface is in the first... The unit normal vector at each sampling point This refers to the unit normal vector extracted at the corresponding point in the working surface area. This is an absolute value conversion operation. For the dot product operation of three-dimensional vectors in space, The normal consistency penalty coefficient is a constant. To control the constant of the mesh compliance penalty coefficient, Let be the square of the Euclidean norm of the vector. To control the 3D coordinate matrix of the vertex; for boundary control vertices, the coordinates of neighboring vertices are supplemented by a boundary symmetric extension method; the 3D coordinates of each control vertex are updated along the negative direction of the geometric gradient matrix vector until they are within the minimum tolerance range.

[0033] As a preferred embodiment of the present invention, in step VII, the implementation logic for performing wall thickness interferometry detection based on spatial voxels is as follows:

[0034] Discrete voxelization is performed on the internal solid space of the thermoformed insert three-dimensional geometric entity to construct a three-dimensional spatial octree index structure;

[0035] Calculate the geometric distance from the center plane spatial detection point to the working surface area, and the geometric distance from the center plane spatial detection point to the cooling cavity wall area; extract the smaller of the two geometric distance parameters as the minimum physical wall thickness value of the local spatial entity;

[0036] The minimum physical wall thickness value of the local spatial entity is compared with the wall thickness parameter of the local target that is associated with the same coordinate space grid node through a spatial mapping relationship. When a violation minimum value exists, a three-dimensional spatial connected domain marking expansion operation is initiated for the set of discrete danger points in the warning area stored in the temporary warning register set to define the wall thickness warning area.

[0037] As a preferred embodiment of the present invention, in step VII, when triggering the reconstruction of the local trajectory line for the wall thickness warning area:

[0038] A globally continuous three-dimensional thermal field function is established within the solid volume of the working surface region. The radial basis function three-dimensional spatial interpolation model is called to perform spatial voxel-level interpolation on the scattered heat load density parameters on the surface, generating an extended heat load density scalar value.

[0039] In the process of solving the large linear equation system for calculating the radial basis weight coefficients, a Tikhonov regularization stabilization module based on singular value decomposition is embedded to process the ill-conditioned condition number and obtain the smoothly decaying extended heat load density scalar value; three-dimensional spatial analytical differentiation is performed on the continuous extended heat load density scalar value to extract the extended heat load gradient field vector.

[0040] As a preferred embodiment of the present invention, when performing the local trajectory reconstruction for the wall thickness warning area, a three-dimensional composite space correction vector is constructed:

[0041]

[0042] in, For a three-dimensional composite space correction vector, This is a three-dimensional spatial repulsive force vector generated based on the wall thickness interferometry detection and pointing away from the normal to the interference boundary. The traction stiffness coefficient is the coefficient for thermal load gradient. Let be the instantaneous vector value of the extended heat load gradient field vector at the node position of the trajectory line. This is the normalization constant for the heat load gradient. Let be the Euclidean norm of the vector. It is a natural exponential function;

[0043] The trajectory line nodes are guided by the three-dimensional composite space correction vector to perform spatial twisting and orbit change until the minimum physical wall thickness of the local spatial entity exceeds the safe thickness distance threshold baseline.

[0044] This invention also discloses an automatic generation system for the center surface of cooling water channels of thermoformed inserts based on three-dimensional modeling, used to execute the automatic generation method for the center surface of cooling water channels of thermoformed inserts based on three-dimensional modeling as described above, including:

[0045] The 3D model recognition module is used to create a 3D geometric model of the thermoformed insert and identify the working surface area and the cooling cavity wall area.

[0046] The grid field feature extraction module is used to discretize the working surface area into a grid, extract the surface feature parameters of each grid node in the working surface area, and obtain the heat load density parameters of each grid node according to the preset heat load calculation rules.

[0047] The constraint parameter matrix construction module is used to construct a set of waterway design constraint parameters based on the surface feature parameters and the heat load density parameters. The set of waterway design constraint parameters includes at least the local waterway spacing parameters and the local target wall thickness parameters corresponding to each of the grid nodes.

[0048] The topology trajectory optimization module is used to establish a continuous symbolic distance field in the three-dimensional space where the working surface area is located, and based on the local target wall thickness parameter, convert the local waterway spacing parameter into topology connection weights in the symbolic distance field, and generate an initial trajectory line set of the waterway centerline based on the topology connection weights.

[0049] The surface skinning rendering module is used to establish cross-sectional circles perpendicular to the tangent direction of each trajectory line in the initial trajectory line set, obtain the topological matching relationship between adjacent cross-sectional circles, perform surface skinning processing on all cross-sectional circles according to the topological matching relationship, and generate the initial mesh model of the center surface.

[0050] The high-order geometric fitting iteration module is used to geometrically fit the initial mesh model of the center surface using a Bézier surface patch with a preset polynomial order, construct an iterative objective function including a normal deviation penalty term and a mesh deformation penalty term, adjust the control vertex coordinates of the Bézier surface patch using a gradient descent mechanism until the iterative objective function meets the convergence condition, and output a smooth center surface model.

[0051] The anti-interference composite reconstruction module is used to perform wall thickness interference detection based on spatial voxels on the smooth center surface model, identify the wall thickness warning area where the distance of the spatial entity is less than the local target wall thickness parameter, trigger the reconstruction of the local trajectory line for the wall thickness warning area, and output the cooling water channel center surface data.

[0052] As a preferred technical solution of the present invention, the system's internal registry contains a set of microscopic thermal diffusivity constants and a database of variable-temperature thermal conductivity corresponding to special superhard alloy billets; the dynamic thermophysical properties of the material are used to perform secondary calibration and scaling on the base spacing reference value variable and the base wall thickness geometric reference value variable, and the spacing compensation correction factor is adjusted by combining the dimensionless Nusselt number variable control equation constructed based on fluid dynamics parameters, thereby coordinating the control of the local water channel spacing parameters.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] This invention breaks away from the constant geometric equidistant offset rule in existing technologies by constructing a dynamic mapping mechanism between heat load density parameters and geometric design parameters. This allows the spacing and wall thickness of the cooling water channels to adapt to differences in heat load distribution on the workpiece surface. In areas with high heat flux convergence, the method drives the local water channel spacing parameters and local target wall thickness parameters to adaptively decrease, increasing the cooling contact area and reducing heat transfer resistance, thereby eliminating the cooling blind zone defects in conventional designs.

[0055] At the trajectory planning level, a continuous signed distance field in three-dimensional space is used to replace the traditional parametric domain equidistant offset, avoiding trajectory self-intersection and curve folding phenomena at strong curvature and acute angle boundaries, thus ensuring the geometric validity and topological correctness of the generated path. By constructing an iterative objective function that integrates normal deviation penalty and mesh deformation penalty, and using a gradient descent mechanism to perform high-order geometric fitting on the Bézier surface patch, the output smooth center surface model is strictly parallel to the working surface and has parametric smoothing characteristics, avoiding subsequent manufacturing failures caused by mesh quality degradation.

[0056] Furthermore, the composite potential field dynamic reconstruction mechanism introduced in this invention, upon detecting wall thickness interference, does not perform a single geometric avoidance but instead fuses the extended thermal load gradient field vector to synthesize a composite correction vector. This pushes the trajectory line nodes away from the safe range of the wall thickness warning zone, while simultaneously guiding the nodes to undergo spatial torsion and trajectory change along the thermal gradient direction, ensuring that the cooling water path closely follows the outer boundary of the high heat flux density zone. This collaborative control logic simultaneously guarantees the structural strength and ultimate cooling efficiency of the mold, effectively suppressing the peak temperature and thermal stress concentration factor of the mold working surface while maintaining a safe wall thickness margin, significantly improving the uniformity of the martensitic phase transformation of the formed part, and achieving a balanced optimization of mechanical safety and thermodynamic cooling efficiency. Attached Figure Description

[0057] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained from these drawings without creative effort.

[0058] Figure 1 This is an overall flowchart of the method described in this invention.

[0059] Figure 2 This is a flowchart of the wall thickness interferometry detection and local trajectory reconstruction of the present invention.

[0060] Figure 3 This is a flowchart of the composite potential field dynamics reconstruction sub-process of the present invention. Detailed Implementation

[0061] In the following description, only certain exemplary embodiments are briefly described. As those skilled in the art will recognize, the described embodiments can be modified in various ways without departing from the spirit or scope of the embodiments of the invention. Therefore, the drawings and description are considered to be exemplary in nature and not restrictive.

[0062] The following is in conjunction with the appendix Figures 1-3 The embodiments of the present invention will be described in detail below.

[0063] Example 1: This example discloses an automatic generation system for the center surface of cooling water channels of thermoformed inserts based on 3D modeling, including:

[0064] The 3D model recognition module is used to create a 3D geometric model of the thermoformed insert and identify the working surface area and the cooling cavity wall area.

[0065] The grid field feature extraction module is used to discretize the working surface area into a grid, extract the surface feature parameters of each grid node in the working surface area, and obtain the heat load density parameters of each grid node according to the preset heat load calculation rules.

[0066] The constraint parameter matrix construction module is used to construct a set of waterway design constraint parameters based on the surface feature parameters and the heat load density parameters. The set of waterway design constraint parameters includes at least the local waterway spacing parameters and the local target wall thickness parameters corresponding to each of the grid nodes.

[0067] The topology trajectory optimization module is used to establish a continuous symbolic distance field in the three-dimensional space where the working surface area is located, and based on the local target wall thickness parameter, convert the local waterway spacing parameter into topology connection weights in the symbolic distance field, and generate an initial trajectory line set of the waterway centerline based on the topology connection weights.

[0068] The surface skinning rendering module is used to establish cross-sectional circles perpendicular to the tangent direction of each trajectory line in the initial trajectory line set, obtain the topological matching relationship between adjacent cross-sectional circles, perform surface skinning processing on all cross-sectional circles according to the topological matching relationship, and generate the initial mesh model of the center surface.

[0069] The high-order geometric fitting iteration module is used to geometrically fit the initial mesh model of the center surface using a Bézier surface patch with a preset polynomial order, construct an iterative objective function including a normal deviation penalty term and a mesh deformation penalty term, adjust the control vertex coordinates of the Bézier surface patch using a gradient descent mechanism until the iterative objective function meets the convergence condition, and output a smooth center surface model.

[0070] The anti-interference composite reconstruction module is used to perform wall thickness interference detection based on spatial voxels on the smooth center surface model, identify the wall thickness warning area where the distance of the spatial entity is less than the local target wall thickness parameter, trigger the reconstruction of the local trajectory line for the wall thickness warning area, and output the cooling water channel center surface data.

[0071] For ease of implementation, Table 1 below shows the commonly used preferred ranges for the key adjustment coefficients in this application. Those skilled in the art can make fine adjustments within this range according to the specific insert material and quenching conditions to obtain stable and effective optimization results.

[0072] Table 1 provides a reference for the commonly used preferred ranges of key adjustment coefficients in industry.

[0073] Heat load sensitivity adjustment coefficient Spatial curvature compensation weight Wall thickness safety redundancy factor Normal consistency penalty coefficient constant Control mesh compliance penalty coefficient constant Repulsive stiffness coefficient constant Thermal load gradient traction stiffness coefficient

[0074] The automatic generation method for the center surface of the cooling water channel of the thermoformed insert based on 3D modeling in this embodiment includes the following steps:

[0075] Step I: The 3D model recognition module establishes a 3D geometric model of the thermoformed insert, identifying the working surface area and the cooling cavity wall area. Acquiring the 3D topological data of the thermoformed insert's physical entity relies on multi-view structured light scanning data fusion technology. The 3D model recognition module uses a non-contact optical scanning device to acquire the initial point cloud; the scanning accuracy of the non-contact optical scanning device is set to no less than 0.01mm (i.e., the maximum measurement error is less than 0.01mm).

[0076] The thermoformed insert is fixed on a 3D rotating stage, and data acquisition is performed from multiple angles to generate local point cloud datasets with multiple independent viewpoints. The 3D model recognition module inputs the local point cloud datasets into the registration and fusion unit, which applies an iterative nearest-point algorithm to eliminate the cumulative rigid body transformation error during the multi-view conversion process.

[0077] To quantify the alignment accuracy of the 3D model recognition module, a spatial error objective functional between the target point cloud and the source point cloud is constructed:

[0078] in, This is the cumulative mean square error scalar during the point cloud registration process. This represents the total number of feature point pairs in the overlapping region that participate in the error calculation. Let be the orthogonal rotation matrix for transforming the source point cloud coordinate system to the target point cloud coordinate system. The first in the source point cloud set Three-dimensional coordinate vectors of feature points Let be the translation vector that transforms the source point cloud coordinate system to the target point cloud coordinate system. For the target point cloud set and Matching the first Three-dimensional coordinate vectors of feature points The square of the Euclidean norm of the vector is given. The registration and fusion unit continuously calculates the optimal orthogonal rotation matrix and translation vector until the cumulative mean square error scalar between the local point cloud datasets of adjacent views converges to below 0.015 mm, outputting a complete 3D topological point cloud.

[0079] The 3D model recognition module applies a moving least squares algorithm to smooth and denoise the 3D topological point cloud and reconstruct the surface, generating a 3D geometric model containing parametric surface information. Within the 3D geometric model, the module extracts a set of faces whose geometric normal vectors point to the external free space, marking this set as the working surface region. It also extracts a set of faces whose geometric normal vectors point to the internal sealed cavity and possess fluid inflow / outflow boundary conditions, marking this set as the cooling cavity wall region. The module then extracts the geometric boundary coordinate systems of the working surface region and the cooling cavity wall region, transforming them to a global reference system to generate 3D spatial boundary frame data. This 3D spatial boundary frame data serves as the absolute geometric reference for subsequent spatial optimization and interferometry detection. After generating the 3D spatial boundary frame data, the module transmits it to the mesh field feature extraction module for feature calculation.

[0080] Step II: The mesh field feature extraction module receives the 3D spatial boundary frame data, discretizes the working surface area into a mesh, extracts the surface feature parameters of each mesh node within the working surface area, and obtains the heat load density parameters of each mesh node according to the preset heat load calculation rules. The mesh field feature extraction module uses the Delaunay triangulation algorithm to adaptively mesh the parameter domain of the working surface area. During the adaptive meshing process, the mesh edge length is dynamically adjusted according to the curvature change rate of the working surface area. For areas where the curvature change rate exceeds the set curvature extreme value constraint, a mesh edge length halving and densification operation is performed. The set curvature extreme value constraint is preset according to the geometric complexity of the inserts, typically taken as 0.1 mm⁻¹. The mesh field feature extraction module outputs the discretized mesh model after meshing. For any mesh node within the discretized mesh model, the mesh field feature extraction module uses local surface quadratic polynomial fitting to obtain the first and second fundamental forms of the surface with constant coefficients. Based on the first and second fundamental forms of the surface with constant coefficients, the specific calculation formula for extracting the surface feature parameters of each mesh node within the working surface area is as follows:

[0081] ;

[0082] in, For the first Gaussian curvature of each grid node For the first The average curvature of each grid node For the first Parameters of the first fundamental form of the surface at each mesh node The squared term of the directional partial derivative, For the first Parameters of the first fundamental form of the surface at each mesh node With parameters The product of directional partial derivatives, For the first Parameters of the first fundamental form of the surface at each mesh node The squared term of the directional partial derivative, For the first Parameters of the second fundamental form of the surface at each grid node Projection of the second partial derivative of the direction, For the first The mixed partial derivative projection terms of the second fundamental form of the surface at each grid node For the first Parameters of the second fundamental form of the surface at each grid node The projection term of the second partial derivative of the direction. Gaussian curvature and mean curvature together constitute the surface characteristic parameters, used to characterize the local convexity and concavity properties and the degree of spatial curvature of the working surface region.

[0083] The mesh field feature extraction module reads the thermodynamic boundary condition parameters input from the stamping process database. The thermal load mainly originates from heat conduction in contact with the high-temperature sheet metal. Combining the thermodynamic boundary condition parameters, the mesh field feature extraction module solves the steady-state heat conduction differential equation of the working surface region within a set quenching cycle using the finite difference method, calculating the scalar heat flux absorbed by each mesh node. The expansion of the steady-state heat conduction differential equation under the Laplace operator is as follows:

[0084]

[0085] in, For divergence operators, Let be the thermal conductivity constant of the thermoformed insert solid material. Three-dimensional spatial coordinates The temperature gradient vector at that point, Three-dimensional spatial coordinates The temperature value at the specified location is used. The surface heat flux boundary value obtained by solving the differential equation is defined as the scalar heat flux, and this scalar heat flux is defined as the heat load density parameter. The surface characteristic parameters and heat load density parameters of the mesh nodes are pressed into the node attribute matrix in parallel. The node attribute matrix serves as the basic input structure for multiphysics data mapping and is transmitted to the constraint parameter matrix construction module.

[0086] Step III: The constraint parameter matrix construction module receives the node attribute matrix. Based on the surface feature parameters and heat load density parameters contained within the node attribute matrix, it constructs a set of water channel design constraint parameters. This set includes at least the local water channel spacing parameters and local target wall thickness parameters corresponding to each grid node. A dynamic mapping mechanism for the local water channel spacing parameters and local target wall thickness parameters is established, enabling the cooling water channel layout to adapt to differences in heat load density distribution. The mathematical derivation of the mapping relationship is as follows:

[0087] ;

[0088] ;

[0089] in, To be allocated to the first Local waterway spacing parameters for each grid node. Based on the baseline value of the basic spacing, For the first Heat load density parameters of each grid node This is the lower bound of the set of minimum values ​​for the heat load density parameters of all grid nodes. This is the upper bound of the set of maximum values ​​for the heat load density parameters of all grid nodes. This is the heat load sensitivity adjustment coefficient. For spatial curvature compensation weights, For the first Extraction of average curvature parameter values ​​for each grid node. It is the absolute value symbol. To be allocated to the first Local target wall thickness parameters for each grid node. Based on the geometric reference value of the basic wall thickness The global arithmetic mean heat load density calculated for the working surface area. The wall thickness safety redundancy factor is... This is a logarithmic function with the natural constant as its base. After obtaining the local waterway spacing parameters and local target wall thickness parameters of all grid nodes, the constraint parameter matrix construction module encapsulates the local waterway spacing parameters and local target wall thickness parameters into the waterway design constraint parameter set. The waterway design constraint parameter set is then submitted to the topology trajectory optimization module to perform path optimization.

[0090] Step IV: The topology trajectory optimization module establishes a continuous symbolic distance field in the three-dimensional space of the working surface area, transforming the local waterway spacing parameters into topological connection weights within the symbolic distance field. Based on these topological connection weights, it generates an initial set of trajectory lines for the waterway centerlines. Establishing a continuous symbolic distance field aims to avoid irreversible self-intersecting singularities generated by equidistant offsets in the parameter domain at sharp corners of complex surfaces. The formula for calculating the continuous symbolic distance field is:

[0091] in, For any spatial coordinate point in three-dimensional free space scalar sign distance value, This is a continuous point set model for the working surface region. These serve as reference points for traversing the continuous point set model within the working surface region. Let be the Euclidean norm of the vector. For algebraic symbol extraction functions, For reference point The outward normal unit vector pointing outwards from the working surface. This is the vector dot product operator. When the spatial coordinates point... When located in free space outside the working surface region, the scalar sign distance value is positive; when When located within a solid region inside a thermoformed insert, the scalar sign distance value is negative, and path search is performed only within the insert region where the sign is negative.

[0092] After calculating the global symbolic distance field, the topology trajectory optimization module constructs a connected undirected graph within the symbolic distance field. The nodes of the undirected graph are spatial voxel centers, and the edge weights are dominated by topological connection weights. The topological connection weights are inversely proportional to the local waterway spacing parameter; that is, the smaller the local waterway spacing parameter, the lower the edge weight of the undirected graph, and the more preferentially the connection edges of that region are selected during path search. Simultaneously, the local waterway spacing parameter directly limits the neighborhood connection range of the path search, only searching for adjacent voxel nodes whose distance from the current voxel node does not exceed the corresponding local waterway spacing parameter, thus completing the conversion from parameter to weight. The process of obtaining the initial trajectory line set includes a path cost optimization integration step, and the path cost minimization integral function is:

[0093] in, The cumulative total cost function value of the trajectory line to be calculated. It is the set of integral paths for a continuous trajectory curve. The differential arc length parameter of the trajectory curve. Let be the spatial three-dimensional coordinates of the point corresponding to the differential arc length. To map the discrete local target wall thickness parameters to points using trilinear interpolation. The continuous local target wall thickness parameters at the location. The local spatial curvature calculated from the trajectory curve. The local spatial torsion calculated from the trajectory curve. To control the wall thickness deviation error, To limit the convergence weight of bending energy in cases of sharp curve bends, To avoid torsional energy penalty weights for abnormal spatial distortion, the topology trajectory optimization module applies a fast traversal algorithm to solve the integral function that minimizes path cost, outputting multiple continuous spatial discrete polylines under cost convergence. The topology trajectory optimization module smooths these continuous spatial discrete polylines and writes them into the initial trajectory line set. The initial trajectory line set contains transition segments connecting equidistant lines at different levels; the lower limit of the curvature radius of these transition segments is strictly constrained by the maximum principal curvature parameter of the working surface region. The initial trajectory line set is then output to the surface skinning rendering module.

[0094] Step V: The surface skinning rendering module establishes cross-sectional circles perpendicular to the tangent direction of each trajectory line within the initial trajectory line set, obtains the topological matching relationship between adjacent cross-sectional circles, and performs surface skinning processing on all cross-sectional circles according to the topological matching relationship to generate the initial mesh model of the center surface. The surface skinning rendering module reads the initial trajectory line set and uses the Flyner frame to discretize sampling nodes along the trajectory lines to generate a series of local orthogonal coordinate systems. Cross-sectional circles are drawn within the normal section of the local orthogonal coordinate system.

[0095] The diameter of the cross-sectional circles is fixed to the nominal diameter of the cooling water system, which is preset according to the cooling flow requirements, typically 8-12mm. After generating the sequence of cross-sectional circles, the surface skinning rendering module performs topology matching calculations. When obtaining the topology matching relationship between adjacent cross-sectional circles, to avoid spatial distortion of the skinning mesh, a feature point alignment formula is applied:

[0096]

[0097] in, The optimal rotation alignment adjustment angle calculated between two adjacent front and rear cross-section circles. This represents the total number of discrete feature points along the circumference of a single cross-section. For the first The first sampling node on the cross-sectional circle Three-dimensional coordinates of discrete feature points For the first The first sampling node on the cross-sectional circle Three-dimensional coordinates of discrete feature points The rotation angle is the rotation axis centered on the tangential vector of the current trajectory line. The corresponding homogeneous rotation matrix.

[0098] The surface skinning rendering module applies optimal rotation alignment adjustment angles to correct the initial phase offset of the circular cross-section. Following the topological connection order of feature points after phase correction, non-uniform rational B-spline interpolation is used to generate quadrilateral patches. All quadrilateral patches are stitched together, outputting the initial mesh model of the center face without overall smoothing. The initial mesh model data of the center face is then transferred to the high-order geometry fitting iteration module.

[0099] Step VI: The high-order geometric fitting iteration module receives the initial mesh model of the center surface, uses a Bézier surface patch with a preset polynomial order to perform geometric fitting on the initial mesh model of the center surface, constructs an iterative objective function that includes a normal deviation penalty term and a mesh deformation penalty term, and uses the gradient descent mechanism to adjust the coordinates of the control vertices of the Bézier surface patch until the iterative objective function meets the convergence condition, and outputs a smooth center surface model.

[0100] The high-order geometric fitting iteration module extracts the local parametric coordinate domain of the initial mesh model of the center plane. A control vertex array is set within this local parametric coordinate domain. The initial coordinates of the control vertex array are obtained by least-squares spatial approximation of the surface polygonal mesh vertices of the initial mesh model of the center plane. A mathematical expression for the Bézier surface patch with a preset polynomial order is established:

[0101]

[0102] in, For parameter coordinates The spatial three-dimensional coordinates output by mapping points on the Bezier surface. For Bézier surfaces in parameters The preset order limit value for direction setting. For Bézier surfaces in parameters The preset order limit value for direction setting. This is the three-dimensional coordinate matrix established in three-dimensional space for the control vertex network of the Bezier surface patch. For calculation parameters The Bernstein basis functions that influence the weights of the direction nodes. For calculation parameters The Bernstein basis function that influences the weights of the direction nodes, with parameters... and The range of values ​​is .

[0103] To overcome the normal vector distortion and surface wrinkling caused by the initial mesh model of the center face, an iterative objective function is constructed, including a normal deviation penalty term and a mesh deformation penalty term. Solving the iterative objective function aims to find the optimal control vertex network coordinate arrangement that minimizes the overall surface physical energy state. The iterative objective function including the normal deviation penalty term and the mesh deformation penalty term is as follows:

[0104] in, The iterative objective function calculated to control the current state of the vertex array is quantized. The scalar represents the total number of sampling points used to perform discretization sampling on the surface of the initial mesh model at the center plane. The fitted Bézier surface is in the first... The unit normal vector obtained at each sampling point This is the unit normal vector extracted at the corresponding point on the working surface area. For scalar absolute value conversion operations, For the dot product operation of three-dimensional vectors in space, The constant is the normal consistency penalty coefficient to limit the deviation of the included angle between the normals in space. To suppress control grid compliance penalty coefficient constants for controlling vertex distance oscillations, the summation term... This represents the summation over all control vertices of a Bézier surface patch. , or , The boundary control vertices are calculated using a boundary-symmetric extension method. or The equivalent value. The objective function maintains the path from the high-temperature heat source to the cooling fluid in the shortest orthogonal physical state by forcing the surface normal to be parallel to the outer working surface normal. The Laplace smoothing term prevents spatial jumps between adjacent control vertices. The higher-order geometric fitting iterative module uses the gradient descent mechanism to adjust the update rule of the control vertex coordinates of the Bézier surface patch as follows:

[0105]

[0106] in, For the experience of the first The latest 3D coordinates of the control vertex after sub-gradient iteration calculation For being in the first The three-dimensional coordinates of the current controlling vertex in the next iteration state. The learning rate parameter is used to control the iteration step size. The geometric gradient matrix vector is formed by the partial derivatives of the iterative objective function with respect to the corresponding control vertex.

[0107] The high-order geometric fitting iterative module monitors the descent gradient slope of the quantized values ​​of the iterative objective function. When the difference between the quantized values ​​of two adjacent gradient iterations falls within a set minimum tolerance range, the iterative objective function is deemed to have met the global convergence condition. This minimum tolerance range is preset according to the processing accuracy requirements, typically 1e-6 mm. The parameter update loop of the gradient descent mechanism is then stopped, the topological position framework of the control vertex network in three-dimensional space is locked, and a smoothed center plane model is generated. The smoothed center plane model data command stream is transmitted to the anti-interference composite reconstruction module.

[0108] Step VII: The anti-interference composite reconstruction module receives the smoothed center surface model, performs wall thickness interference detection based on spatial voxels on the smoothed center surface model, identifies the wall thickness warning area where the distance of the spatial entity is less than the local target wall thickness parameter, triggers the reconstruction of the local trajectory line for the wall thickness warning area, and outputs the cooling water channel center surface data.

[0109] The anti-interference composite reconstruction module performs discrete voxelization partitioning on the internal solid space of the 3D geometric solid, constructing a 3D spatial octree index structure. The formula for calculating the local spatial solid wall thickness is:

[0110]

[0111] in, Let's take any space probe point selected on the central plane. The minimum physical wall thickness of the local spatial entity is calculated by ray-based measurements taken from the origin as the starting reference point. For high-density discrete points on the surface covered by the point set of the working surface region, These are the discrete boundary points of the back-side inner cavity covered by the point set of the cooling cavity wall region. This is a continuous point set model of the cooling cavity wall region. The calculation formula compares the geometric distance from a point on the central surface to the high-temperature working surface outside the insert with the geometric distance extended to the internal cooling cavity, performing a minimum value interception. The anti-interference composite reconstruction module compares the difference between the minimum physical wall thickness value of the local spatial entity and the local target wall thickness parameter mapped to the same coordinate space grid node through the nearest point projection.

[0112] When the minimum physical wall thickness of a local spatial entity is less than the local target wall thickness parameter, the monitoring unit extracts the three-dimensional spatial points into a temporary warning register set. For the set of discrete danger points in the warning area stored in the temporary warning register set, the anti-interference composite reconstruction module initiates a three-dimensional spatial connected domain marking expansion operation, aggregating and stitching isolated danger points into a warning area bounding box, thus identifying the wall thickness warning area in the three-dimensional model.

[0113] After the anti-interference composite reconstruction module locks the wall thickness warning zone, it forcibly triggers a local trajectory reconstruction command for that zone. An artificial potential field repulsion vector is used to perform a displacement mechanical correction calculation on the trajectory mesh center that deviates from the safe thickness track.

[0114]

[0115] in, This refers to the three-dimensional spatial repulsive resultant force vector applied to the misaligned deformation nodes of the waterway trajectory line. To adjust the repulsive force stiffness coefficient constant of the repulsive force field pushing away the strength level, To identify the sequence of all extracted danger points in the wall thickness warning zone, The three-dimensional coordinates of discrete danger points in the warning zone, located at the boundary of the wall thickness warning zone. This provides the three-dimensional coordinates of the core force-bearing center node, which urgently needs to undergo spatial anti-interference displacement push-away correction. The three-dimensional spatial repulsive force vector drives the core force-bearing center node to translate deeper into the solid. After confirming that the corrected smooth center surface model passes the anti-interference closed-loop verification, the anti-interference composite reconstruction module converts the topology array into a standard three-dimensional model transmission file format, such as STEP or IGES, and outputs the cooling water channel center surface data.

[0116] Comparative Example 1: Comparative Example 1 is established for reference. Comparative Example 1 replicates the conventional process logic in the industry that relies on constant spacing and a single equidistant translation rule for conformal flow channel arrangement.

[0117] An initial three-dimensional geometric topology data matrix for the thermoformed insert, identical to that in Example 1, was established. The model was also divided into an external high-temperature working surface region and an internal low-temperature cooling cavity wall region. The mesh node feature capture mechanism for the heat load density parameter in Example 1 was removed.

[0118] Bypassing all physical heat transfer analysis procedures, the basic parametric continuous surface data of the working area is directly captured. A fixed global water channel spacing constant value, independent of spatial morphology, is forcibly bound to this basic parametric continuous surface data. This water channel spacing constant value is set to 25 mm in the control panel. Simultaneously, a physical wall thickness constant depth limit data is permanently written into the global registry, with the wall thickness constant depth limit data fixed at 10 mm.

[0119] The underlying 2D planar projection isopleth calculation component is invoked. This component directly instructs the outer boundary contour of the working surface region to perform geometric topological offset calculations with a fixed step size of 25 mm, generating a 2D parametric topological contour network composed of a series of mechanical parallel structures. The reverse mapping array tool is then used to project and snap the 2D parametric topological contour network back onto the 3D undulating physical shell of the working surface.

[0120] By capturing the absolute normal in the opposite direction of each point on the physical shell of the 3D undulating working surface, all the adsorbed contour mesh points are uniformly pressed vertically downwards into the depth of the mold by a constant physical distance of 10 mm, generating a series of geometric center tracks suspended and stagnant inside the alloy steel entity. Around these geometric center tracks generated by existing technology, a solid sweep additive modeling process with a fixed pipe diameter of 10 mm is initiated, and a 3D Boolean subtractive modeling operation is performed on the water channel cavity entity.

[0121] When the geometric center track line passes through a localized area of ​​raised bends on the outer surface of the insert where the radius of curvature is reduced to less than 12 mm, the constant depth downward pressure logic directly triggers the interpenetration of adjacent flow channel segments within a narrow local volume.

[0122] The three-dimensional Boolean subtractive process of the water channel cavity solid leaves a large number of intersecting dead holes that are impossible to manufacture. After abandoning the dynamic density adjustment elastic mechanism that tracks the local heat load density parameter, the hot-stamped sheet inside the insert impacts the sharp high-temperature convergence point area where heat absorption is most intense, and cannot get enough refrigerant flow to cover and flush it.

[0123] Heat congestion occurred in the high-temperature convergence area, and the local temperature measurement nodes of the insert showed high-temperature distortion exceeding the safety warning threshold by more than 75°C. The abandonment of the post-processed multi-dimensional collision wall thickness back-off verification logic module resulted in the generated water channel cavity being extremely close to the countersunk hole structure on the back side, causing the weakest point of the physical isolation layer between the two geometric features to drop to 0.94 mm. The minimum safe wall thickness threshold for special superhard alloys under 7.5 MPa pressure is 0.8 mm. Although this value is slightly higher than the threshold, the conventional engineering design safety margin recommends ≥0.15 mm. This solution has a safety margin of only 0.14 mm, posing a significant risk of failure due to processing deviations.

[0124] Example 2: Based on the continuous logic established in Example 1, Example 2 changes the core variable dependency relationship between the local water channel spacing parameter and the local target wall thickness parameter in the underlying mathematical mapping model, and extends and integrates the dynamic thermophysical property constant response module of the mold material into the system-level node attribute matrix.

[0125] The core steel component of the hot forming die is forged from a special superhard alloy doped with molybdenum and vanadium. Different batches of the special superhard alloy forging blanks exhibit uneven thermal diffusivity with minute oscillating peaks. The set of constant values ​​of the microscopic thermal diffusivity parameter of the special superhard alloy and the floating data pool of the dynamic thermal conductivity parameter are merged into the pre-calculation sequence pool of the physics field and imported into the control core. The floating data pool of the dynamic thermal conductivity parameter records the measured thermal conductivity values ​​of different batches of special superhard alloys in the temperature range of 20°C-600°C.

[0126] For the baseline values ​​of foundation spacing and foundation wall thickness, a secondary calibration scaling based on the actual material thermal conductivity is performed. The calibration formula is as follows:

[0127] ,

[0128] in, and These are the initial values ​​of the foundation spacing and foundation wall thickness geometric datum, respectively, based on standard reference materials. The measured thermal conductivity is for the current batch of special superhard alloys. This is the thermal conductivity of the standard reference alloy. Also, the repulsive force stiffness constant. Adjustments are made synchronously based on the actual thermal conductivity of the material. When it increases, The wall thickness is increased synchronously to provide a stronger pushing force to ensure that the high thermal conductivity does not fail due to excessive wall thinning.

[0129] In extrapolating and calculating the convective heat transfer efficiency of the fluid, the model introduces a dimensionless Nusselt number variable governing equation based on fluid dynamic parameters. This equation is used to determine the convective heat transfer index within the rough pipe wall. A network of spacing compensation correction factors, dependent on the evolution of fluid dynamic viscous force parameters, is also established. The Nusselt number calculation formula is as follows: ,in The convective heat transfer coefficient of the coolant. The nominal inner diameter of the waterway. Let be the thermal conductivity of the cooling medium. Define the spacing compensation coefficient. , and The relationship is: when Increasing the spacing indicates improved convective heat transfer efficiency, and the water channel spacing can be appropriately widened. This increases accordingly. The final corrected baseline spacing value is... This allows for coordinated control of the local waterway spacing parameters.

[0130] The local water channel spacing parameters assigned to the target discretized mesh nodes are not only controlled by the geometric attenuation pressure caused by the working surface heat load density parameters, but also by the positive proportional amplification effect of the global average thermal conductivity performance level of the special ultra-hard alloy billet, which offsets the feedback influence. High-density materials with high deep-layer heat transfer speeds can tolerate the system appropriately widening and sparsening the local flow channel mesh, thereby reducing the processing and production time.

[0131] When performing the integral term calculation of the path cost minimization integral function, to address the sparse mesh defect at the far end caused by excessively fast material transfer speed, the system forcibly increases the contraction constraint tension, which includes wall thickness deviation weights, to prevent excessively wide flow channels from causing the near-end wall thickness to significantly exceed the baseline cooling and wetting critical depth limit. When relying on artificial potential fields for risk avoidance, the baseline constant-order repulsive force stiffness coefficient used to disperse displaced dangerous nodes is increased. A thrust at the level of gravity acceleration ensures that the complex local network interference reconstruction loop ends within a shorter step size iteration, enhancing the algorithm's robust adaptation and fault tolerance boundary to abrupt changes in microscopic material properties.

[0132] Example 3: Based on the continuous curved surface mesh and symbolic distance field topology logic established in Example 1, Example 3 deeply reconstructs the local trajectory reconstruction mechanism for the wall thickness warning zone. When dealing with extremely narrow regions where the wall thickness of local spatial entities is limited, the artificial potential field dynamics theory relies solely on repulsive force vectors to forcibly push the trajectory lines deep into the metal matrix of the thermoformed insert. This unidirectional geometric avoidance action strips away the core physical function of the cooling water system, inducing an elongation of the heat transfer distance between the local cooling medium and the heat source. This causes the surface of the wall thickness warning zone, subjected to high thermal shock, to rapidly lose heat exchange efficiency, resulting in local heat accumulation and annealing softening defects in the mold material.

[0133] Example 3 incorporates a thermodynamic flow-guiding algorithm kernel based on a three-dimensional spatial interpolation model of radial basis functions into the background system architecture, upgrading the unidirectional geometric repulsion to a composite potential field dynamic reconstruction network that integrates thermal load gradient traction vectors.

[0134] Specifically, the discretized mesh model is read, and the heat load density parameters bound to each mesh node within the working surface area are extracted. These heat load density parameters are essentially attached to the two-dimensional manifold surface of the working surface area and cannot directly generate vectorized directional guidance computational power for the cooling water channel center surface hidden in the three-dimensional space inside the thermoformed insert. The background system triggers a multi-dimensional space expansion instruction module, forcibly establishing a globally continuous three-dimensional thermal field function within the solid volume of the thermoformed insert.

[0135] To reduce computational burden, the construction of the three-dimensional thermal field function does not employ the computationally redundant three-dimensional finite element method for solving steady-state heat conduction. Instead, it directly calls the radial basis function three-dimensional spatial interpolation model to perform spatial voxel-level interpolation derivation of the scattered heat load density parameters on the surface. The derivation formula is compiled into low-level executable code by the core tensor processor.

[0136]

[0137] in, For any coordinate point in three-dimensional space The extended heat load density scalar value calculated from the above. This represents the total number of discrete heat source points extracted by the system within the point set of the working surface region using a spatial downsampling algorithm. Assigning the first... through the inverse solution of the matrix equation system Discrete points of heat source Radial basis weighting coefficients, For the first The three-dimensional coordinates of discrete points of the heat source To control the smoothing kernel width parameter of the Gaussian radial basis function in three-dimensional space to decay the gradient, it is recommended to take a value of 1.2 to 2 times the average side length of the working surface mesh.

[0138] A distance matrix is ​​constructed using the coordinates of discrete points of the heat source. The heat load density parameters inherent to these discrete points are extracted to form the target column vector. The distance matrix is ​​multiplied by an unknown weight vector containing radial basis weight coefficients, and the product is equated to the target column vector, thus constructing a large-scale linear equation system with massive dimensions. Solving this large-scale linear equation system encounters the extremely dense distribution of the point cloud, causing the distance matrix to exhibit ill-conditioned condition number properties. To address this, the system forcibly embeds a Tikhonov regularization stabilization module based on singular value decomposition into the matrix inversion pipeline, reconstructing the original objective functional into a ridge regression expression that minimizes the sum of the square norm of the prediction error and the square norm of the weight vector.

[0139] By introducing a regularized damping factor, truncated masking suppression is applied to the small singular values ​​of the distance matrix. The regularized damping factor is preset based on the condition number of the distance matrix, and its value is typically [value missing]. The constants within the range ensure that the calculated radial basis weight coefficient vector has smoothness and physical fidelity, ensuring that the extended heat load density scalar value exhibits a smooth decay pattern that conforms to Fourier's law of heat conduction in the region transitioning from the surface to the interior of the solid.

[0140] After completing the scalar field mapping across the entire space, the vector guiding index that determines the fluid's directional tendency is extracted. Three-dimensional analytical differentiation is performed on the continuous extended heat load density scalar values ​​to obtain the extended heat load gradient field vector. The partial differential expression for the extended heat load gradient field vector is:

[0141]

[0142] in, To extend the scalar value of heat load density at point The three-dimensional orthogonal direction first-order partial derivative vector array is extracted from the mapping. The extended heat load gradient field vector points to the absolute direction of the most intense increase in local heat accumulation at every microscopic voxel point within the three-dimensional space.

[0143] For the voxel-based wall thickness interferometry detection process, the anti-interference composite reconstruction module also identifies the wall thickness warning zone where the distance to the spatial entity is less than the local target wall thickness parameter, and intercepts the node coordinates to be corrected on the trajectory line that triggers the interferometry alarm. Freeze the existing single repulsion correction logic and start the composite space correction vector synthesis unit. The computation unit retrieves the generated repulsion vector. repulsive vector The heat load gradient traction vector, with the extended heat load gradient field vector as its core, is injected into the middle. The complete mathematical formula for the three-dimensional composite space correction vector is established as follows:

[0144]

[0145] in, A three-dimensional composite space correction vector (hereinafter referred to as composite correction vector) is used to simultaneously take into account interference prevention and heat transfer efficiency. This refers to the three-dimensional repulsive force vector generated by the repulsive field mentioned above. The thermal load gradient traction stiffness coefficient is used to adjust the traction mechanical level strength under thermal load gradient. This represents the instantaneous vector value at the current node position of the extended heat load gradient field vector constructed within three-dimensional space for the heat load density parameters of the working surface region. This is the heat load gradient normalization constant, predefined based on the maximum modulus of the heat load density parameter in the working area, used to eliminate the absolute magnitude difference of the gradient.

[0146] Repulsive vector Its function is to push the waterway centerline trajectory node in the initial trajectory line set into the safe interior of the mold along the reverse path of the normal of the working surface area, thereby eliminating the risk of breakage.

[0147] If the working surface area corresponding to the current wall thickness warning zone happens to be the core quenching zone where extremely high heat flux converges, forcibly pushing it will cause the cooling water channel to move away from the core quenching zone. At this time, the magnitude of the extended heat load gradient field vector calculated at the node increases sharply. According to the composite correction vector formula, the system imparts a huge heat load gradient traction force. The heat load gradient traction vector pulls the node to slide laterally along the tangential contour of the mold surface. While avoiding local minimum wall thickness obstacles, the trajectory line node is forced to travel along the three-dimensional spatial contour ridge with the highest temperature gradient, producing a spatial twisting and trajectory-changing effect, crossing the interference danger zone and precisely wrapping around the nearby safe secondary high-temperature zone with high heat conduction value.

[0148] After replacing the repulsive vector with a composite correction vector, the master node of the anti-interference composite reconstruction module dispatches node coordinate displacement integral update tasks to the distributed parallel computing cluster, continuously monitoring the changes in the local entity wall thickness monitoring pointer. When the minimum physical wall thickness of the local spatial entity first exceeds the safe thickness distance threshold baseline, the matrix operator of the anti-interference composite reconstruction module applies a reverse damping coefficient to attenuate and truncate the magnitude of the composite correction vector, preventing the trajectory line nodes from overshooting the limit. The reconstructed discrete polyline segments are then fed into a high-frequency tubular feature point matching and alignment process, mobilizing the Bezier surface patch iterative network to perform local fine-tuning and smoothing, generating the final cooling water channel center surface data carrying the composite potential field optimization gene.

[0149] A three-dimensional transient thermodynamic fluid-structure interaction verification platform was constructed to perform digital twin operation condition verification on the output cooling water channel center surface data entity. A special super-hard alloy forged steel property package and a turbulent coolant medium model with a pressure level of 20 MPa were loaded, and a closed-loop quenching and pressure holding cycle test with a duration of 12 s was performed on boron steel sheet material in austenitic state at 930 °C with a hot-formed insert in contact.

[0150] The ultimate performance of the center plane geometry generated by the repulsion of a single artificial potential field (Scheme 1) and the center plane geometry generated by the dynamic reconstruction of a composite potential field (Scheme 2) are compared. The highest peak temperature node data of the working surface region of the mold, the maximum thermal stress concentration factor data of the entire mold, the average volume fraction of martensitic phase transformation of the entire formed part, and the minimum safe wall thickness margin data are extracted and a verification comparison data table is output.

[0151] Table 2. Comparison data of dynamic reconstruction verification of composite potential field;

[0152] Peak temperature in the working area 185.4 °C 142.7 °C Decrease by 42.7 °C Maximum thermal stress coefficient across the entire domain 4.82 MPa / °C 3.15 MPa / °C Reduced by 34.6% Average volume fraction of martensitic phase transformation 92.1% 98.6% Increased by 6.5% Minimum safety wall thickness margin for solids 0.85 mm 0.88 mm Maintaining security boundaries

[0153] Note: The maximum thermal stress coefficient refers to the maximum thermal stress value generated inside the mold under a unit temperature gradient change. The larger the coefficient, the higher the risk of thermal fatigue cracking of the mold.

[0154] The analysis of the physical indicators in the fluid-structure interaction verification comparison data table disclosed in Table 2 reveals several key aspects. Scheme 1, by simply retreating deeper into the extremely narrow, acute-angle interference zone, resulted in a 30 mm long cooling blind zone in the local high-heat-flux area. The peak temperature in the working surface region surged above 180 °C, and the extreme temperature gradient induced lattice distortion within the mold. Scheme 2, on the other hand, utilizes a spatial lateral sliding avoidance tactic based on the thermal load gradient traction vector. While maintaining a minimum safe wall thickness margin of over 0.85 mm, it guides the water channels to tightly embrace the outer boundary of the high-heat-flux-density region. The peak temperature in the working surface region is suppressed within the 150 °C cold cycle safety threshold, and the significantly reduced thermal stress concentration factor eliminates the physical fatigue hazard of localized cracking in the insert. The increase in the average volume fraction of martensitic phase transformation across the entire formed part verifies the engineering application value of the composite potential field reconstruction logic in overcoming the fundamental physical limitations of geometric avoidance and heat transfer dead zones. The output cooling water channel center surface data has the processing maturity to be directly applied to high-end CNC machine tool production lines for physical manufacturing.

[0155] Example 4: Based on the composite potential field dynamics reconstruction logic and local trajectory line reconstruction network established in Example 3, this example deeply explores the physical conduction potential of extreme heat flux in confined space, and introduces a variable cross-section morphology mapping mechanism based on heat transfer anisotropy and a fluid-structure interaction micro-turbulence reconstruction network.

[0156] Conformal cooling water channels are often constrained by the constant-diameter drill bits or ball end mills used in machining, resulting in a fixed cylindrical cross-section. When the water channel trajectory is pushed into a narrow safety zone outside the high heat flux density area under the guidance of a three-dimensional composite space correction vector, maintaining a constant circular cross-section leads to a geometrically limited heat exchange surface area between the cooling medium and the sidewall of the thermoformed insert. To overcome this geometric limitation, the anti-interference composite reconstruction module incorporates a cross-section topology deformation engine during the local trajectory reconstruction stage.

[0157] The anti-interference composite reconstruction module locks the trajectory segments undergoing local trajectory reconstruction within the wall thickness warning zone. For the discrete core force center nodes on the trajectory segments, the cross-section topology deformation engine retrieves the three-dimensional composite space correction vector synthesized from the repulsive force vector and the thermal load gradient traction vector. It extracts the local trajectory line tangential vector at the current force center node position. The cross-section topology deformation engine performs a spatial vector cross product operation on the three-dimensional composite space correction vector and the local trajectory line tangential vector to generate the cross-section major axis reference vector. This cross-section major axis reference vector is physically perpendicular to the direction of the most intense heat flow, ensuring that subsequent cross-section deformation unfolds along the isothermal surface and blocking physical expansion towards the extremely thin solid wall thickness.

[0158] Establish a continuous cross-sectional deformation scale mapping function. The cross-sectional topology deformation engine reads the extended thermal load gradient field vector at the node spatial coordinates. Extract the modulus of the extended thermal load gradient field vector at the corresponding node position. Input the extracted modulus into the variable cross-sectional major and minor axis scaling function. The underlying logic of the variable cross-sectional major and minor axis scaling function is set as follows:

[0159]

[0160] in, For the node The output of the variable cross-section major and minor axis scaling function at the location is a quantized scalar. This is the morphological expansion sensitivity constant, a dimensionless parameter ranging from 0.85 to 1.5. To extend the magnitude scalar of the thermal load gradient field vector, This represents the upper bound of the set of maximum values ​​for the heat load density parameters of all grid nodes. The quantized scalar output of the variable cross-section major and minor axis scaling function directly reflects the thermal resistance breaking demand caused by local heat accumulation.

[0161] The cross-section topology deformation engine establishes the principle of baseline area conservation to ensure that the fluid does not experience drastic macroscopic pressure drop changes in the variable cross-section section. The standard circular cross-sectional area calculated from the nominal diameter of the cooling water system is used as the baseline area constraint. By simultaneously establishing the baseline area constraint and the proportional function of the major and minor axes of the variable cross-section, a quantized scalar is output, and a nonlinear equation system is established to solve for the physical length parameters of the major and minor axes of the variable cross-section ellipse. The calculated physical length parameter of the major axis is greater than the nominal diameter of the cooling water system, while the physical length parameter of the minor axis is less than the nominal diameter of the cooling water system.

[0162] The cross-section topology deformation engine uses the calculated major and minor axis physical length parameters to draw a variable cross-section elliptical profile in a locally orthogonal coordinate system. The major axis direction of the variable cross-section ellipse is forcibly aligned and parallel to the cross-section major axis reference vector. At this point, the minor axis direction of the variable cross-section ellipse automatically points towards the direction of the extended thermal load gradient field vector. The standard cross-section circles at corresponding nodes are entirely replaced with the variable cross-section ellipse. Outside the safety boundary of the wall thickness warning zone, the flow channel evolves into a flat and wide elliptical ribbon-like channel. The shorter minor axis avoids the minimum physical wall thickness of the local spatial entity, while the longer major axis significantly increases the fluid wetting area parallel to the high-temperature working surface region.

[0163] A topological transition zone is formed in the area where the cross-sectional circle and the variable cross-section ellipse meet at the beginning and end along the trajectory line. The cross-sectional topological deformation engine extracts the feature point sequences of the cross-sectional circle and the variable cross-section ellipse. A cubic Hermitian spline curve is used to construct an interpolation matrix, and the derivative constraints of the control points along the tangent of the trajectory line are calculated. A variable cross-section mesh transition patch spanning the topological transition zone is generated. The variable cross-section mesh transition patch maintains the first level of the surface. Continuity is maintained to prevent cavitation and stripping of the fluid boundary layer. The variable cross-section mesh transition patch is spliced ​​with the remaining normal pipe segment skin to generate an updated initial mesh model for the center surface.

[0164] The introduction of the variable cross-section ellipse widens the flow cross-section, resulting in a slight decrease in local flow velocity under the control of the fluid dynamics continuity equation. This velocity decrease weakens the convective heat transfer coefficient at the liquid-solid interface. The anti-interference composite reconstruction module embeds a fluid-structure interaction micro-turbulence reconstruction network within the variable cross-section ellipse. This network generates turbulence microstructures pointing towards the interior of the cooling channel along the corresponding side of the cavity wall of the variable cross-section ellipse.

[0165] The turbulence reconstruction network calculates the shortest spatial projection distance from discrete feature points on the edge of the variable cross-section ellipse to the point set model of the working surface region. The shortest spatial projection distance is compared with the local target wall thickness parameter mapped to the same location. The semi-elliptical arc segment on one side of the edge where the shortest spatial projection distance is less than the local target wall thickness parameter is extracted and marked as the microstructure growth baseline. The microstructure growth baseline is precisely aligned and locked onto the windward side of the thermal shock that urgently needs enhanced heat transfer.

[0166] Activate the computational fluid dynamics background evaluation component to solve the Navier-Stokes equations and obtain the background flow field and Reynolds number distribution within the variable cross-section of the cooling channel. Based on the Reynolds number distribution, construct a formula for adjusting the microstructure height.

[0167]

[0168] in, To be allocated to the microstructure growth baseline The radial growth height of the perturbation microstructure at each discrete point To prevent the critical height extreme constant of flow resistance, it is set to 0.15 times the physical length parameter of the minor axis of the variable cross-section ellipse. The dimensionless parameter for the local fluid Reynolds number corresponding to the 3D mesh node. The velocity coupling attenuation coefficient controls the saturation response of height as velocity increases. For the first variable cross section ellipse The spatial geometric angle cosine variable between the local surface curvature normal vector at a discrete point and the reference vector of the major axis of the cross section. As the discrete point approaches the pole of the ellipse, the spatial geometric angle decreases, and the height of the microstructure gradually reaches its peak.

[0169] Obtain the radial growth height of all perturbation microstructures assigned to the microstructure growth baseline nodes. The perturbation reconstruction network reverses the normal direction and performs coordinate space offset operations on the mesh vertices inward into the solid space. The offset trajectory exhibits a streamlined ridge protrusion shape, constructing an asymmetric perturbation rib solid structure mesh arranged in an interlaced array. The longitudinal spacing of the interlaced array is set to 2.5 times the maximum radial growth height of the perturbation microstructure. The asymmetric perturbation rib solid structure mesh triggers localized turbulent vortices, stripping away the thermally impeded fluid boundary layer and forcibly entraining the low-temperature fluid in the deep core region of the refrigerant to the heated wall surface of the mold. The perturbation reconstruction network completely merges the asymmetric perturbation rib solid structure mesh into the cooling water channel center plane data stream, outputting a complete microscopic thermal control topology file for additive manufacturing.

[0170] A control and verification group was set up. The control and verification group extracted the center surface model of the circular tube shape (circular tube structure) from Example 3 without cross-sectional deformation. The experimental group extracted the composite center surface model (variable cross-section turbulence structure) generated in Example 4, which included a mesh of a variable cross-section ellipse and asymmetric turbulence ribs. Both models were thermoformed into solid parts using a selective laser melting metal 3D printer with identical parameters. An infrared thermal imaging monitoring array and an inlet / outlet high-frequency pressure sensor array were deployed, and a standard high-strength steel plate hot stamping test cycle was performed. Core performance comparison data were extracted.

[0171] Table 3. Comparison of microscopic topological evolution performance of fluid-solid interfaces;

[0172] Local heat transfer coefficient of the core heat load area 14500 W / (m²·K) 21850 W / (m²·K) Increase by 7350 W / (m²·K) Wall thickness warning zone transient peak mode temperature 142.7 °C 118.4 °C Decrease by 24.3 °C Overall thermal cycling time of the insert 8.5 s 6.2 s Shortened by 2.3 s Single-cycle internal cooling hydraulic pressure surge 0.42 MPa 0.58 MPa Increase by 0.16 MPa (within safe limits)

[0173] Table 3 details the thermodynamic changes brought about by the cross-sectional evolution and microstructure compensation based on heat transfer anisotropy. Within the core heat load zone, the local heat transfer coefficient increases dramatically due to the stretching of the wetting width by the variable cross-section ellipse and the strong horseshoe-shaped turbulent vortex shedding induced by the asymmetric turbulence ribs. The transient peak modulus temperature in the wall thickness warning zone is further suppressed to an ultra-high safety line of 118.4 °C. The overall thermal cycle recovery time of the insert is significantly shortened, meaning that the production cycle limit of the automated stamping production line is physically lifted. Even at the cost of a slight loss of hydrodynamic friction resistance and a 0.16 MPa increase in the cooling pressure drop per cycle, this back pressure load is still far below the design threshold of industrial-grade high-pressure cooling pumps.

[0174] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0175] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. It should be noted that any modifications, equivalent substitutions, and improvements 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 automatically generating the center surface of cooling water channels for thermoformed inserts based on 3D modeling, characterized in that, Includes the following steps: Step 1: Establish a three-dimensional geometric model of the thermoformed insert and identify the working surface area and the cooling cavity wall area; Step II: Discretize the working surface area into a grid, extract the surface feature parameters of each grid node in the working surface area, and obtain the heat load density parameters of each grid node according to the preset heat load calculation rules. Step III: Based on the surface feature parameters and the heat load density parameters, construct a set of waterway design constraint parameters. The set of waterway design constraint parameters includes at least the local waterway spacing parameters and the local target wall thickness parameters corresponding to each of the grid nodes. Step IV: Establish a continuous symbolic distance field in the three-dimensional space where the working surface area is located. Based on the local target wall thickness parameter, convert the local waterway spacing parameter into topological connection weights in the symbolic distance field, and generate an initial trajectory line set of the waterway centerline based on the topological connection weights. Step V: Establish a cross-sectional circle perpendicular to the tangent direction of the trajectory line along each trajectory line in the initial trajectory line set, obtain the topological matching relationship between adjacent cross-sectional circles, and perform surface skinning on all cross-sectional circles according to the topological matching relationship to generate the initial mesh model of the center surface; Step VI: Geometrically fit the initial mesh model of the center surface using a Bézier surface patch with a preset polynomial order, construct an iterative objective function including a normal deviation penalty term and a mesh deformation penalty term, and adjust the control vertex coordinates of the Bézier surface patch using a gradient descent mechanism until the iterative objective function meets the convergence condition, and output a smooth center surface model. Step VII: Perform wall thickness interference detection based on spatial voxels on the smoothed center surface model, identify the wall thickness warning area where the distance of the spatial entity is less than the local target wall thickness parameter, trigger local trajectory line reconstruction for the wall thickness warning area, and output the cooling water channel center surface data.

2. The method for automatically generating the center surface of the cooling water channel of a thermoformed insert based on three-dimensional modeling according to claim 1, characterized in that, In step II, the specific implementation process of discretizing the working surface region into a mesh is as follows: The Delaunay triangulation algorithm is used to adaptively mesh the working surface region. When the rate of curvature change exceeds the set extreme value, the mesh edge length is halved and the mesh is refined. The specific process of extracting surface feature parameters is as follows: using local surface quadratic polynomial fitting to obtain the first basic form and the constant coefficients of the second basic form of the surface, and then calculating the Gaussian curvature and average curvature of each grid node. By combining thermodynamic boundary condition parameters, the steady-state heat conduction differential equation of the working surface region within a set quenching cycle is solved using the finite difference method, and the scalar heat flux absorbed by each grid node is calculated as the heat load density parameter.

3. The method for automatically generating the center surface of the cooling water channel of a thermoformed insert based on three-dimensional modeling according to claim 2, characterized in that, In step III, the mathematical relationship for constructing the local waterway spacing parameters is as follows: in, To be allocated to the first Local waterway spacing parameters for each grid node. Based on the baseline value of the basic spacing, For the first Heat load density parameters of each grid node This is the lower bound of the set of minimum values ​​for the heat load density parameters of all grid nodes. This is the upper bound of the set of maximum values ​​for the heat load density parameters of all grid nodes. This is the heat load sensitivity adjustment coefficient. For spatial curvature compensation weights, For the first The average curvature of each grid node It is the absolute value symbol; By calculating the global arithmetic mean heat load density, and combining the geometric baseline value of the foundation wall thickness with the wall thickness safety redundancy factor, the load is mapped and generated to the [missing information]. The local target wall thickness parameters of each grid node.

4. The method for automatically generating the center surface of the cooling water channel of a thermoformed insert based on three-dimensional modeling according to claim 3, characterized in that, Step IV, the process of establishing a continuous symbolic distance field includes: Calculate the scalar symbolic distance value of any spatial coordinate point in three-dimensional free space relative to the continuous point set model of the working surface region; A connected undirected graph is constructed within the symbolic distance field, with the spatial voxel center as the undirected graph node, and the topological connection weight dominates the edge weight of the undirected graph; the local waterway spacing parameter controls the step size of the path search or the neighborhood connection range to achieve the transformation; The fast travel algorithm is used to solve the path cost minimization integral function, which includes wall thickness deviation penalty, bending energy convergence penalty and torsional energy penalty. The wall thickness deviation penalty term is calculated based on the continuous local target wall thickness field obtained by interpolation from discrete local target wall thickness parameters. The continuous spatial discrete polyline in the cost convergence state is output to form the initial trajectory line set.

5. The method for automatically generating the center surface of the cooling water channel of a thermoformed insert based on three-dimensional modeling according to claim 4, characterized in that, In step V, when obtaining the topological matching relationship between adjacent cross-sectional circles, the feature point alignment control process is executed: A local orthogonal coordinate system is generated by discretely sampling nodes along the trajectory line using the Flyner frame to draw the cross-sectional circle, the diameter of which is fixed as the nominal diameter of the cooling water system; A homogeneous rotation matrix is ​​constructed with the tangential vector of the current trajectory line as the rotation axis. The optimal rotation alignment adjustment angle is calculated to minimize the sum of the squared distances of the three-dimensional coordinates of the discrete feature points on the two cross-sectional circles. After correcting the phase using the optimal rotation alignment adjustment angle, quadrilateral patches are generated by non-uniform rational B-spline interpolation, and then spliced ​​to generate the initial mesh model of the center surface.

6. The method for automatically generating the center surface of the cooling water channel of a thermoformed insert based on three-dimensional modeling according to claim 5, characterized in that, In step VI, the iterative objective function, which includes a normal deviation penalty term and a mesh deformation penalty term, is as follows: in, Quantize the numerical values ​​for the iterative objective function. The scalar represents the total number of surface sampling points of the initial mesh model on the central face. The fitted Bézier surface is in the first... The unit normal vector at each sampling point This refers to the unit normal vector extracted at the corresponding point in the working surface area. This is an absolute value conversion operation. For the dot product operation of three-dimensional vectors in space, The normal consistency penalty coefficient is a constant. To control the constant of the mesh compliance penalty coefficient, Let be the square of the Euclidean norm of the vector. To control the 3D coordinate matrix of the vertex; for boundary control vertices, the coordinates of neighboring vertices are supplemented by a boundary symmetric extension method; the 3D coordinates of each control vertex are updated along the negative direction of the geometric gradient matrix vector until they are within the minimum tolerance range.

7. The method for automatically generating the center surface of the cooling water channel of a thermoformed insert based on three-dimensional modeling according to claim 6, characterized in that, In step VII, the implementation logic for performing wall thickness interferometry detection based on spatial voxels is as follows: Discrete voxelization is performed on the internal solid space of the thermoformed insert three-dimensional geometric entity to construct a three-dimensional spatial octree index structure; Calculate the geometric distance from the center plane spatial detection point to the working surface area, and the geometric distance from the center plane spatial detection point to the cooling cavity wall area; extract the smaller of the two geometric distance parameters as the minimum physical wall thickness value of the local spatial entity; The minimum physical wall thickness value of the local spatial entity is compared with the wall thickness parameter of the local target that is associated with the same coordinate space grid node through a spatial mapping relationship. When a violation minimum value exists, a three-dimensional spatial connected domain marking expansion operation is initiated for the set of discrete danger points in the warning area stored in the temporary warning register set to define the wall thickness warning area.

8. The method for automatically generating the center surface of the cooling water channel of a thermoformed insert based on three-dimensional modeling according to claim 7, characterized in that, In step VII, when the local trajectory line reconstruction for the wall thickness warning area is triggered: A globally continuous three-dimensional thermal field function is established within the solid volume of the working surface region. The radial basis function three-dimensional spatial interpolation model is called to perform spatial voxel-level interpolation on the scattered heat load density parameters on the surface, generating an extended heat load density scalar value. In the process of solving the large linear equation system for calculating the radial basis weight coefficients, a Tikhonov regularization stabilization module based on singular value decomposition is embedded to process the ill-conditioned condition number and obtain the smoothly decaying extended heat load density scalar value; three-dimensional spatial analytical differentiation is performed on the continuous extended heat load density scalar value to extract the extended heat load gradient field vector.

9. The method for automatically generating the center surface of the cooling water channel of a thermoformed insert based on three-dimensional modeling as described in claim 8, characterized in that, When reconstructing the local trajectory line for the wall thickness warning area, a three-dimensional composite space correction vector is constructed: in, For a three-dimensional composite space correction vector, This is a three-dimensional spatial repulsive force vector generated based on the wall thickness interferometry detection and pointing away from the normal to the interference boundary. The traction stiffness coefficient is the coefficient for thermal load gradient. Let be the instantaneous vector value of the extended heat load gradient field vector at the node position of the trajectory line. This is the normalization constant for the heat load gradient. Let be the Euclidean norm of the vector. It is a natural exponential function; The trajectory line nodes are guided by the three-dimensional composite space correction vector to perform spatial twisting and orbit change until the minimum physical wall thickness of the local spatial entity exceeds the safe thickness distance threshold baseline.

10. The method for automatically generating the center surface of the cooling water channel of a thermoformed insert based on three-dimensional modeling according to claim 9, characterized in that, For the trajectory segments within the wall thickness warning zone that have undergone the local trajectory reconstruction, perform a variable cross-section shape mapping operation based on heat transfer anisotropy: Obtain the three-dimensional composite space correction vector and the local trajectory line tangential vector of the corresponding node on the trajectory line segment reconstructed by the local trajectory line; calculate the vector cross product of the three-dimensional composite space correction vector and the local trajectory line tangential vector; and generate the cross section major axis reference vector. Extract the magnitude of the extended heat load gradient field vector at the corresponding node position, and construct a variable cross-section major and minor axis scaling function; Using the nominal diameter of the cooling water system as the reference area constraint, and combining the output value of the variable cross-section major and minor axis ratio function, the physical length parameters of the major axis and minor axis of the variable cross-section ellipse are calculated. The cross-sectional circle at the corresponding node is replaced with the variable cross-sectional ellipse, the major axis of the variable cross-sectional ellipse being parallel to the reference vector of the major axis of the cross-section; cubic Hermit spline curves are used to generate a variable cross-sectional mesh transition patch that spans the topological transition zone between the cross-sectional circle and the variable cross-sectional ellipse, and the patches are spliced ​​together to generate an updated initial mesh model of the center surface.

Images