Artificial intelligence assisted stamping part formability rapid analysis and prediction method and system

Abstract

The invention provides an artificial intelligence-assisted stamping part formability rapid analysis and prediction method and system, and relates to the technical field of plate stamping forming, and the method comprises the steps: obtaining the three-dimensional data of a stamping part to extract geometric deformation characteristic parameters, constructing a strain path curve, and determining a deformation risk region; and establishing a corresponding relation between strain abrupt change characteristics and defect types to predict defect positions and types, and optimizing grid data based on a strain rate gradient and a material hardening index to carry out forming analysis. The method can improve the stamping part forming defect prediction precision and analysis efficiency, and reduce the mold test debugging cost and period.

Description

Technical Field

[0001] The present invention relates to sheet metal stamping forming technology, and in particular to an artificial intelligence-assisted method and system for rapid analysis and prediction of stamping part formability. Background Art

[0002] During the stamping process, part formability analysis is crucial for ensuring product quality and improving production efficiency. Traditional stamping analysis methods rely primarily on finite element simulation, which consumes significant computational time and resources and struggles to meet the demands of rapid analysis and prediction. Furthermore, due to limitations in meshing accuracy, the strain evolution characteristics of localized deformation areas cannot be accurately captured, compromising the accuracy of the analysis results.

[0003] With the development of artificial intelligence (AI) technology, it has become possible to apply machine learning methods to formability analysis. However, current intelligent analysis methods often focus solely on predicting the final forming result, lacking in-depth analysis of the strain evolution characteristics during the forming process, making it difficult to effectively identify and warn of potential forming defects. Furthermore, existing methods often suffer from unstable prediction accuracy when processing complex parts due to limitations in mesh quality.

[0004] Historical data and engineering experience indicate that forming defects are closely related to local strain mutations and strain rate gradients. However, existing technologies have yet to establish effective defect warning mechanisms and adaptive mesh optimization methods. Therefore, an intelligent analysis method that can quickly and accurately predict forming defects and incorporate adaptive mesh optimization capabilities is urgently needed. Summary of the Invention

[0005] The embodiments of the present invention provide an artificial intelligence-assisted method and system for rapid analysis and prediction of stamping part formability, which can solve the problems in the prior art.

[0006] A first aspect of an embodiment of the present invention provides an artificial intelligence-assisted method for rapid analysis and prediction of stamping part formability, comprising:

[0007] Acquire the three-dimensional data of the stamping part and perform mesh division to obtain initial mesh data, and extract geometric deformation feature parameters from the initial mesh data;

[0008] A strain path curve is constructed based on the geometric deformation characteristic parameters. The strain increments between adjacent time points of the strain path curve are calculated. The strain acceleration is calculated based on the strain increments. The strain path curve is compared with the forming limit curve to obtain the safety margin value of each grid node. The deformation risk area is determined based on the strain acceleration and safety margin value.

[0009] Obtain historical stamping forming data containing defect types and defect characteristics, analyze the strain mutation characteristics of the defect formation process, establish the corresponding relationship between strain mutation characteristics and defect types, obtain defect warning criteria, analyze the deformation risk area based on the defect warning criteria, and obtain the predicted defect location and defect type;

[0010] The strain rate gradient at the predicted defect location is calculated and combined with the material strain hardening exponent to construct a mesh optimization equation. The mesh optimization equation is used to calculate the mesh refinement coefficient. The initial mesh data is locally optimized based on the mesh refinement coefficient to obtain the optimized mesh data. The optimized mesh data is used to perform stamping analysis and output the analysis results.

[0011] In an optional embodiment,

[0012] The three-dimensional data of the stamping part is obtained and meshed to obtain the initial mesh data. The geometric deformation feature parameters extracted from the initial mesh data include:

[0013] Using a 3D scanner to acquire and preprocess point cloud data of the stamping part, constructing initial mesh data based on the preprocessed point cloud data, the initial mesh data including node position information and unit topology relationships, calculating mesh quality parameters of the initial mesh data, and optimizing the mesh area based on the mesh quality parameters to obtain an optimized mesh model;

[0014] Establishing a deformation feature calculation matrix based on the optimized mesh model, the deformation feature calculation matrix including node displacements, deformation gradients, and strain components, calculating a deformation tensor using the deformation feature calculation matrix, and solving a principal strain component and a secondary strain component using the deformation tensor;

[0015] The wall thickness variation of each node is calculated based on the principal strain component and the secondary strain component, the depth variation value is calculated by combining the wall thickness variation with the node displacement, and a geometric deformation characteristic parameter including the principal strain component, the secondary strain component, the wall thickness variation and the depth variation value is constructed.

[0016] In an optional embodiment,

[0017] Constructing a strain path curve based on geometric deformation characteristic parameters, calculating the strain increments between adjacent time points of the strain path curve, and calculating the strain acceleration based on the strain increments include:

[0018] The geometric deformation feature parameters are subjected to Gaussian filtering and mean filtering to obtain filtered deformation feature data, and the filtered deformation feature data are spatially reconstructed using cubic spline interpolation to obtain a continuously distributed deformation feature field;

[0019] Establishing a primary strain field and a secondary strain field based on a continuously distributed deformation characteristic field, and fitting a strain path curve between the primary strain field and the secondary strain field using a nonlinear least squares method with an adaptive weight coefficient, wherein the strain path curve includes a primary strain component and a secondary strain component;

[0020] The strain path curve is decomposed into orthogonal direction vectors to obtain a primary strain vector and a secondary strain vector. The primary strain vector difference and the secondary strain vector difference are calculated based on the displacement increment and time increment of adjacent time points to obtain a strain increment vector. The strain increment vector is dynamically accumulated within a sampling time window to obtain a strain increment cumulative value. The strain acceleration is calculated based on the rate of change of the strain increment cumulative value. The strain acceleration represents the dynamic characteristics of the local deformation.

[0021] In an optional embodiment,

[0022] Compare the strain path curve with the forming limit curve to obtain the safety margin value of each grid node. According to the strain acceleration and safety margin value, the deformation risk area is determined to include:

[0023] A first set of sampling points is obtained by sampling at equal time intervals on a strain path curve, and a second set of sampling points is obtained by sampling at equal strain intervals on a forming limit curve. The first set of sampling points and the second set of sampling points are paired based on a minimum distance principle, and the distance between the paired points is calculated to construct a distance mapping matrix. A change section is identified based on a gradient change rate of values in the distance mapping matrix, and the sampling point density is increased within the change section. All sampling points are reconstructed using a piecewise cubic spline interpolation algorithm.

[0024] Calculate the vertical distance from the sampling point on the strain path curve to the reconstructed forming limit curve, construct a nonlinear weight function, use the nonlinear weight function to calculate the safety margin component value of the sampling point, and obtain the safety margin value by weighted accumulation of the safety margin component values of all sampling points;

[0025] The safety margin value and strain acceleration of each grid node are combined into a risk judgment feature pair. Potential risk nodes are identified based on the preset dual-threshold constraint. The spatial distance and strain gradient continuity between adjacent risk nodes are calculated. Risk nodes with spatial distances smaller than the grid feature size and continuously changing strain gradients are merged into the initial risk area. The boundaries of the initial risk area are corrected according to the strain gradient distribution of the nodes to determine the final deformation risk area.

[0026] In an optional embodiment,

[0027] Obtain historical stamping forming data containing defect types and defect characteristics, analyze the strain mutation characteristics of the defect formation process, establish the corresponding relationship between strain mutation characteristics and defect types, obtain defect warning criteria, analyze the deformation risk area based on the defect warning criteria, and obtain the predicted defect location and defect type, including:

[0028] Obtain historical stamping forming data containing defect types and defect characteristics, calculate the strain gradient tensor, strain path curvature and energy dissipation rate of the historical stamping forming data, and obtain multi-dimensional strain mutation characteristics;

[0029] Recursive Bayesian estimation is used to update the weights of the multi-dimensional strain mutation features to generate fused feature data. Based on the defect types in historical stamping data and the fused feature data, the temporal variation characteristics and spatial distribution characteristics of the defect formation process are analyzed to establish a mapping relationship between strain mutation features and defect types.

[0030] Constructing a feature-defect association network based on the mapping relationship, analyzing the evolution path of the fused feature data in the association network, and establishing an early warning criterion including feature thresholds and defect evolution laws;

[0031] Strain field data and temperature field data are collected in real time, the main direction and main value change rate of the strain gradient tensor in the deformation area are extracted, and a stress transfer path diagram of the deformation area is established. The strain path curvature mutation points and the degree of fluctuation of the energy dissipation rate during the stress transfer process are analyzed. The deformation instability index is calculated in combination with the defect evolution law in the early warning criterion. The defect location is determined based on the propagation characteristics of the instability index, and the defect type is determined according to the strain state combination during the instability process.

[0032] In an optional embodiment,

[0033] Establish a stress transfer path diagram for the deformation area, analyze the strain path curvature mutation points and energy dissipation rate fluctuations during the stress transfer process, calculate the deformation instability index based on the defect evolution law in the early warning criteria, determine the defect location based on the propagation characteristics of the instability index, and determine the defect type based on the strain state combination during the instability process, including:

[0034] The deformation area is discretized into triangular grid units, and the difference in the main stress directions between adjacent grid units is calculated as the transfer impedance. The minimum spanning tree algorithm is used to connect the minimum transfer impedance units to obtain the stress transfer path diagram of the deformation area.

[0035] Collecting strain field data along the stress transfer path diagram to construct a strain path, calculating the curvature change of the strain path, and extracting the curvature mutation point and the strain energy combination to form a mutation feature vector;

[0036] determining an energy transfer path according to the stress transfer path diagram, calculating the strain energy dissipation rate, its fluctuation degree and main frequency component along the energy transfer path, and combining them into a fluctuation characteristic vector;

[0037] The mutation characteristic vector and the fluctuation characteristic vector are mapped on the stress transfer path diagram to obtain the energy dissipation fluctuation characteristic score and the stress transfer path characteristic score respectively. Combined with the defect evolution law in the early warning criterion, the deformation instability index is calculated by using a dynamic weight combination;

[0038] Establish a spatial distribution map based on the deformation instability index, calculate the deformation instability index gradient field, determine the instability expansion direction and instability expansion speed based on the propagation characteristics of the instability index, and predict the defect location;

[0039] The principal strain ratio is calculated using strain field data, the strain path non-proportionality is calculated based on the nonlinearity of the strain path, and the local strain concentration factor is calculated according to the strain distribution in the spatial distribution diagram. The principal strain ratio, strain path non-proportionality, and local strain concentration factor are combined into a strain state feature, and the defect type is determined based on the value of the strain state feature.

[0040] In an optional embodiment,

[0041] The mesh refinement coefficient is calculated using the mesh optimization equation. The initial mesh data is locally optimized based on the mesh refinement coefficient to obtain the optimized mesh data. The stamping analysis is performed using the optimized mesh data. The output analysis results include:

[0042] Calculating the plastic power density within the unit cell based on the grid optimization equation, and establishing a stress-strain field distribution function including a deformation energy density term and a strain localization term according to the plastic power density;

[0043] Calculating the stress gradient and strain gradient of the mesh element using the stress-strain field distribution function, establishing a critical refinement threshold according to the plastic power density, and comparing the stress gradient and strain gradient with the critical refinement threshold to obtain a mesh refinement coefficient;

[0044] The mesh elements are locally encrypted according to the mesh refinement coefficient, a strain path function is introduced into the encrypted mesh elements, the previously acquired material strain hardening exponent is updated to the encrypted mesh data, finite element iterative calculations are performed using the optimized mesh data, the strain rate gradient and mesh refinement coefficient are updated based on the calculation results, the mesh density distribution is dynamically adjusted, and the stamping analysis results are output.

[0045] A second aspect of the embodiments of the present invention provides an artificial intelligence-assisted rapid analysis and prediction system for stamping part formability, comprising:

[0046] The first unit is used to obtain the three-dimensional data of the stamping part and perform mesh division to obtain initial mesh data, and extract geometric deformation feature parameters from the initial mesh data;

[0047] The second unit is used to construct a strain path curve based on the geometric deformation characteristic parameters, calculate the strain increment between adjacent time points of the strain path curve, and calculate the strain acceleration based on the strain increment. The strain path curve is compared with the forming limit curve to obtain the safety margin value of each grid node. The deformation risk area is determined based on the strain acceleration and safety margin value.

[0048] The third unit is used to obtain historical stamping forming data containing defect types and defect characteristics, analyze the strain mutation characteristics of the defect formation process, establish the corresponding relationship between the strain mutation characteristics and the defect type, obtain the defect warning criterion, analyze the deformation risk area based on the defect warning criterion, and obtain the predicted defect location and defect type;

[0049] The fourth unit is used to calculate the strain rate gradient at the predicted defect location, combine the strain rate gradient with the material strain hardening exponent to construct a mesh optimization equation, use the mesh optimization equation to calculate the mesh refinement coefficient, locally optimize the initial mesh data based on the mesh refinement coefficient, obtain the optimized mesh data, use the optimized mesh data to perform stamping analysis, and output the analysis results.

[0050] According to a third aspect of an embodiment of the present invention, an electronic device is provided, including:

[0051] processor;

[0052] a memory for storing processor-executable instructions;

[0053] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0054] According to a fourth aspect of an embodiment of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method described above is implemented.

[0055] In this embodiment, the geometric deformation characteristic parameters are extracted to construct a strain path curve, and the deformation risk area is determined by combining the strain acceleration and the safety margin value, thereby realizing early identification of potential problems in the stamping part forming process, improving the prediction accuracy, and reducing the risk of material waste and mold damage in production. Based on the analysis of historical stamping forming data, a correspondence between strain mutation characteristics and defect types is established to form defect warning criteria, which can accurately predict the location and type of defects, allowing engineers to optimize stamping process parameters in a targeted manner, shortening the product development cycle and improving product quality stability. A grid optimization method combining strain rate gradient and material strain hardening index is adopted to achieve local refinement of the grid, improving analysis efficiency while ensuring calculation accuracy, meeting the demand for rapid analysis in industrial production, and having significant engineering application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 Schematic diagram of the process of the artificial intelligence-assisted rapid analysis and prediction method for stamping part formability according to an embodiment of the present invention;

[0057] Figure 2 A comparison diagram of safety margin weight functions according to an embodiment of the present invention;

[0058] Figure 3 The spatial distribution of deformation instability index and gradient field thermal map of the embodiment of the present invention;

[0059] Figure 4 This is a schematic diagram of the structure of an artificial intelligence-assisted rapid analysis and prediction system for stamping part formability according to an embodiment of the present invention. DETAILED DESCRIPTION

[0060] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0061] The following specific embodiments are used to describe the technical solution of the present invention in detail. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0062] Figure 1 FIG. 1 is a flow chart of a method for rapid analysis and prediction of stamping part formability assisted by artificial intelligence according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0063] Acquire the three-dimensional data of the stamping part and perform mesh division to obtain initial mesh data, and extract geometric deformation feature parameters from the initial mesh data;

[0064] A strain path curve is constructed based on the geometric deformation characteristic parameters. The strain increments between adjacent time points of the strain path curve are calculated. The strain acceleration is calculated based on the strain increments. The strain path curve is compared with the forming limit curve to obtain the safety margin value of each grid node. The deformation risk area is determined based on the strain acceleration and safety margin value.

[0065] Obtain historical stamping forming data containing defect types and defect characteristics, analyze the strain mutation characteristics of the defect formation process, establish the corresponding relationship between strain mutation characteristics and defect types, obtain defect warning criteria, analyze the deformation risk area based on the defect warning criteria, and obtain the predicted defect location and defect type;

[0066] The strain rate gradient at the predicted defect location is calculated and combined with the material strain hardening exponent to construct a mesh optimization equation. The mesh optimization equation is used to calculate the mesh refinement coefficient. The initial mesh data is locally optimized based on the mesh refinement coefficient to obtain the optimized mesh data. The optimized mesh data is used to perform stamping analysis and output the analysis results.

[0067] In an optional embodiment, obtaining three-dimensional data of the stamping part and performing mesh division to obtain initial mesh data, and extracting geometric deformation feature parameters from the initial mesh data includes:

[0068] Using a 3D scanner to acquire and preprocess point cloud data of the stamping part, constructing initial mesh data based on the preprocessed point cloud data, the initial mesh data including node position information and unit topology relationships, calculating mesh quality parameters of the initial mesh data, and optimizing the mesh area based on the mesh quality parameters to obtain an optimized mesh model;

[0069] Establishing a deformation feature calculation matrix based on the optimized mesh model, the deformation feature calculation matrix including node displacements, deformation gradients, and strain components, calculating a deformation tensor using the deformation feature calculation matrix, and solving a principal strain component and a secondary strain component using the deformation tensor;

[0070] The wall thickness variation of each node is calculated based on the principal strain component and the secondary strain component, the depth variation value is calculated by combining the wall thickness variation with the node displacement, and a geometric deformation characteristic parameter including the principal strain component, the secondary strain component, the wall thickness variation and the depth variation value is constructed.

[0071] In this example, a 3D scanner is used to scan stamped parts to obtain point cloud data. For example, a structured light 3D scanner is used to scan an automotive panel, generating raw point cloud data containing approximately 500,000 spatial points. Because raw point cloud data often contains noise and redundant points, preprocessing is required. This preprocessing includes noise filtering, point cloud downsampling, and point cloud registration.

[0072] During the noise removal phase, a statistical outlier filtering method is used to calculate the average distance from each point to its nearest neighbors. If the distance exceeds twice the global average distance, the point is identified as a noise point and removed. For example, with a distance standard deviation of 1.0 and a number of nearest neighbors of 50, approximately 0.8% of noise points are filtered out. During the point cloud downsampling phase, a voxel grid downsampling method is used to divide the point cloud space into a grid of small cubes, with each point within the grid represented by its geometric center. For a voxel size of 0.5mm, the original point cloud data is downsampled to approximately 150,000 points, preserving the object's geometric features while improving processing efficiency. Point cloud registration is achieved using an iterative closest point algorithm, which aligns the point cloud data from multiple scans to the same coordinate system. During the registration process, the maximum number of iterations is set to 100, the convergence threshold is set to 0.00001, and the average registration error is kept within 0.1mm.

[0073] Based on the preprocessed point cloud data, the initial mesh data was constructed using the Poisson surface reconstruction method. During the Poisson reconstruction process, the reconstruction depth was set to 9 and the number of sampling points was set to 15, resulting in an initial mesh model consisting of approximately 80,000 triangular elements and 40,000 nodes. Each node contains 3D coordinate information (x, y, z), and each element contains the index information of three nodes, forming the topological relationship of the mesh.

[0074] Perform a quality assessment on the initial mesh model and calculate mesh quality parameters. Mesh quality parameters include: cell area ratio, cell interior angle, cell aspect ratio, and adjacent cell angle. The cell area ratio is defined as the ratio of the triangle cell area to the average mesh area, ideally ranging from 0.5 to 1.5; the cell interior angle is defined as the minimum of the three interior angles of a triangle, ideally greater than 30 degrees; the cell aspect ratio is defined as the ratio of the triangle height to the base length, ideally ranging from 0.5 to 2.0; and the adjacent cell angle is defined as the angle between the normal vectors of adjacent triangles, ideally less than 30 degrees.

[0075] Optimization was performed on areas where mesh quality parameters did not meet requirements. Optimization methods included mesh smoothing, mesh subdivision, and mesh reconstruction. Mesh smoothing used the Laplace smoothing operator to adjust node positions, with a smoothing iteration count of 5. Mesh subdivision targeted high-curvature areas, such as the corners of stamped parts, by subdividing a single triangle into four smaller triangles. Mesh reconstruction optimized mesh quality by adjusting node connectivity for severely distorted elements. After optimization, the proportion of elements that did not meet quality requirements was reduced from an initial 15% to less than 3%.

[0076] Based on the optimized mesh model, a deformation feature calculation matrix is established. First, the node displacement between the original sheet material plane and the curved surface after stamping is calculated. For each mesh node, the three-dimensional coordinate difference (Δx, Δy, Δz) before and after stamping is recorded to form a node displacement matrix. Then the deformation gradient is calculated. Within each unit, the in-plane deformation gradient component is calculated by the displacement difference of the three nodes of the unit. For a typical area of the stamping part, the in-plane component of the deformation gradient ranges from about 0.8 to 1.2, indicating that there is a certain degree of tensile or compressive deformation in the area.

[0077] Strain components are calculated based on the deformation gradient. Strain components include normal strain and shear strain, which are calculated through deformation gradient. For the forming process of a certain automobile fender, the maximum normal strain in the key deformation area is about 0.18, and the maximum shear strain is about 0.12. The deformation tensor is constructed using strain components, and the principal strain component and secondary strain component are solved by tensor decomposition. The principal strain component indicates the main direction and degree of deformation, while the secondary strain component indicates the secondary deformation direction and degree. For the deep drawing area of the stamping part, the average principal strain is about 0.15, and the average secondary strain is about -0.08, which conforms to the principle of approximately constant material volume.

[0078] The wall thickness change at each node is calculated based on the principal and secondary strain components. This calculation combines the principal and secondary strains to determine the wall thickness change. For automotive panel stampings, the wall thickness reduction in the drawn area is approximately 12% to 15%, with reductions in the corner area reaching up to 18%, while the wall thickness increase in the flange area is approximately 5% to 8%.

[0079] Finally, the wall thickness change is combined with the node displacement to calculate the depth change. The depth change is defined as the distance between the node displacement in the forming direction and the original sheet plane. For a cup-shaped stamping, the depth change is approximately 20mm to 25mm in the center and 5mm to 8mm at the edge.

[0080] In this embodiment, by introducing mesh quality parameters for regional optimization, the geometric accuracy and computational stability of subsequent analysis are improved. By establishing a deformation feature calculation matrix and extracting core data such as node displacement and strain components, the local deformation behavior during the stamping process can be fully reflected. Further combining the principal strain component, secondary strain component, wall thickness, and depth change values to construct multi-dimensional geometric deformation feature parameters, it is possible to achieve accurate quantification of complex deformation areas, providing a data basis for subsequent strain path construction, risk area identification, and defect prediction, thereby significantly improving the accuracy and real-time performance of stamping formability analysis.

[0081] In an optional embodiment, constructing a strain path curve according to geometric deformation characteristic parameters, calculating strain increments of the strain path curve between adjacent time points, and calculating strain acceleration based on the strain increments includes:

[0082] The geometric deformation feature parameters are subjected to Gaussian filtering and mean filtering to obtain filtered deformation feature data, and the filtered deformation feature data are spatially reconstructed using cubic spline interpolation to obtain a continuously distributed deformation feature field;

[0083] Establishing a primary strain field and a secondary strain field based on a continuously distributed deformation characteristic field, and fitting a strain path curve between the primary strain field and the secondary strain field using a nonlinear least squares method with an adaptive weight coefficient, wherein the strain path curve includes a primary strain component and a secondary strain component;

[0084] The strain path curve is decomposed into orthogonal direction vectors to obtain a primary strain vector and a secondary strain vector. The primary strain vector difference and the secondary strain vector difference are calculated based on the displacement increment and time increment of adjacent time points to obtain a strain increment vector. The strain increment vector is dynamically accumulated within a sampling time window to obtain a strain increment cumulative value. The strain acceleration is calculated based on the rate of change of the strain increment cumulative value. The strain acceleration represents the dynamic characteristics of the local deformation.

[0085] Exemplarily, geometric deformation characteristic parameters are obtained. These parameters are typically acquired through image processing techniques, digital image correlation techniques, or other sensing devices, and include, but are not limited to, displacement field and strain field data. For example, during the stretching process of a metal sheet, displacement data points on its surface can be collected, including 1,000 data points for x- and y-direction displacement.

[0086] The acquired geometric deformation feature parameters are filtered. This embodiment uses a combination of Gaussian and mean filtering. The Gaussian filter uses a Gaussian kernel with a standard deviation of 1.5 and a window size of 5×5 pixels; the mean filter uses a sliding window of 3×3 pixels. This combination of filters effectively eliminates random noise and outliers in the data. For example, an outlier value of 0.85mm in the original displacement data is corrected to 0.32mm, more consistent with the distribution of the surrounding data.

[0087] After filtering, these discrete data points are spatially reconstructed using cubic spline interpolation techniques to obtain a continuously distributed deformation feature field. 100 control points are selected for interpolation, and the interpolation precision is set to 0.001, ensuring sufficient smoothness and accuracy of the reconstructed deformation feature field. Through cubic spline interpolation, the data originally collected on a 5mm x 5mm grid is reconstructed into a continuous distribution, allowing deformation values to be calculated at any location.

[0088] Based on the reconstructed continuous deformation characteristic field, the principal and secondary strain fields are established. The principal strain field represents the strain component in the direction of the maximum principal strain, while the secondary strain field represents the strain component perpendicular to the principal strain. In practice, the eigenvalues and eigenvectors of the strain tensor are calculated for each spatial point. The larger eigenvalue corresponds to the principal strain, while the smaller eigenvalue corresponds to the secondary strain. For example, at a certain point, the principal strain value is calculated to be 0.12, with a direction of 45 degrees; the secondary strain value is -0.05, with a direction of 135 degrees.

[0089] A nonlinear least squares method with adaptive weighting coefficients was used to fit the strain path curve between the principal and secondary strain fields. The weighting coefficients were dynamically adjusted based on the reliability of the data points, with higher weights assigned to points with higher reliability and lower weights assigned to points with lower reliability. The initial weights were all set to 1.0 and adjusted during the iterations based on the residuals. The weights of points with residuals exceeding a threshold of 0.01 were halved. A quadratic polynomial was used as the basic fitting function, with a maximum number of iterations of 100 and a convergence accuracy of 0.0001. Through fitting, a strain path curve representing the relationship between the principal and secondary strains was obtained. For example, for a certain metal sheet, the relationship between the secondary and primary strains can be expressed as follows: the secondary strain equals -0.32 times the square of the principal strain minus 0.15 times the principal strain plus 0.002.

[0090] The strain path curve is decomposed into orthogonal vectors to obtain principal and secondary strain vectors. Strain vectors contain both magnitude and direction information, together describing the material's deformation state. For example, at t = 10s, the principal strain vector at a point is (0.15, 30°), and the secondary strain vector is (-0.07, 120°).

[0091] The strain increment vector is calculated based on the strain vectors at adjacent time points. For time points t and t+Δt, the difference between the principal strain vector and the secondary strain vector is calculated to obtain the strain increment vector. Specifically, if the principal strain vector at t=10s is (0.15, 30°) and the principal strain vector at t=11s is (0.16, 31°), then the principal strain increment vector is (0.01, 1°).

[0092] The strain increment vector is dynamically accumulated within the sampling time window to obtain the strain increment cumulative value. The time window is selected as 5 seconds and the sliding step is 1 second, which can reflect the deformation accumulation trend in the recent period. For example, in the 5-second window from t=10s to t=15s, the principal strain increment cumulative value is 0.05 and the secondary strain increment cumulative value is -0.02. Finally, the strain acceleration is calculated based on the rate of change of the strain increment cumulative value. The strain acceleration characterizes the speed of strain change and is an important indicator for predicting material failure. In the calculation, the strain increment cumulative values of three consecutive time windows are taken, and the change rate is calculated based on the central difference method. For example, the principal strain increment cumulative values for the three windows of t=10s, t=15s, and t=20s are 0.05, 0.08, and 0.15, respectively. The principal strain acceleration at t=15s is calculated to be 0.02 / s 2 .

[0093] The above method realizes the calculation process from geometric deformation characteristic parameters to strain acceleration, which can be used for dynamic characteristic analysis and failure prediction during material or structural deformation. By constructing a strain path curve and combining vector decomposition and dynamic cumulative analysis of the primary and secondary strain fields, the nonlinear deformation behavior of local areas during the stamping process can be accurately captured, improving the accuracy of the description of complex geometric deformations. By introducing Gaussian filtering and cubic spline interpolation, the smooth reconstruction of deformation characteristic data is achieved, enhancing the continuity and computational stability of the strain path. The strain acceleration calculated based on the strain increment and its rate of change helps to identify transient deformation anomalies and provide early warning of potential instability or crack risks, thereby improving the intelligent perception capability and predictive analysis accuracy of the forming process.

[0094] In an optional embodiment, the strain path curve is compared with the forming limit curve to obtain a safety margin value of each grid node, and the deformation risk area is determined according to the strain acceleration and the safety margin value, including:

[0095] A first set of sampling points is obtained by sampling at equal time intervals on a strain path curve, and a second set of sampling points is obtained by sampling at equal strain intervals on a forming limit curve. The first set of sampling points and the second set of sampling points are paired based on a minimum distance principle, and the distance between the paired points is calculated to construct a distance mapping matrix. A change section is identified based on a gradient change rate of values in the distance mapping matrix, and the sampling point density is increased within the change section. All sampling points are reconstructed using a piecewise cubic spline interpolation algorithm.

[0096] Calculate the vertical distance from the sampling point on the strain path curve to the reconstructed forming limit curve, construct a nonlinear weight function, use the nonlinear weight function to calculate the safety margin component value of the sampling point, and obtain the safety margin value by weighted accumulation of the safety margin component values of all sampling points;

[0097] The safety margin value and strain acceleration of each grid node are combined into a risk judgment feature pair. Potential risk nodes are identified based on the preset dual-threshold constraint. The spatial distance and strain gradient continuity between adjacent risk nodes are calculated. Risk nodes with spatial distances smaller than the grid feature size and continuously changing strain gradients are merged into the initial risk area. The boundaries of the initial risk area are corrected according to the strain gradient distribution of the nodes to determine the final deformation risk area.

[0098] For example, the system acquires strain path curve data and the material's forming limit curve data for each mesh node during the forming process. The system obtains data on the changes in the primary and secondary strains at each node during the forming process from finite element simulations or real-time monitoring to generate a strain path curve. The material's forming limit curve can be obtained through standard testing or retrieved from a materials database.

[0099] The system samples the strain path curve at equal intervals. For example, in a simulation with a total forming time of 1 second, a point is sampled every 0.05 seconds, resulting in a first set of 20 sampling points. Each sampling point contains the principal and secondary strain values, recorded as (ε1, ε2). For example, the sampling point for a node at t = 0.25 seconds might be (0.12, 0.05), indicating a principal strain of 0.12 and a secondary strain of 0.05.

[0100] For sampling the forming limit curve, the system samples at equal strain intervals. For example, on a curve with a principal strain range of 0 to 0.6, a point is sampled every 0.03 of the principal strain value, resulting in 21 secondary sampling points. For example, in the biaxial tension region (where the secondary strain is positive), a sampling point might be (0.36, 0.18); in the uniaxial tension region (where the secondary strain is negative), a sampling point might be (0.45, -0.15).

[0101] The system pairs the first set of sampling points with the second set of sampling points based on the minimum distance principle. For each sampling point on the strain path curve, its Euclidean distance to all sampling points on the forming limit curve is calculated, and the point with the smallest distance is selected as its pairing point. For example, point (0.24, 0.08) on the strain path curve may be paired with point (0.27, 0.09) on the forming limit curve, with a distance of 0.0374. The system constructs the distance values between all paired points into a distance mapping matrix. By analyzing the gradient change rate of the distance values in the matrix, sections with large changes are identified. For example, if the distances of three consecutive sampling points to the forming limit curve are 0.15, 0.08, and 0.03, respectively, the gradient change in this section is large, and the sampling point density needs to be increased. In the changing section, the system inserts additional sampling points. For example, between two points originally separated by 0.05 seconds, four additional points are inserted to reduce the sampling interval to 0.01 seconds.

[0102] After adjusting the sampling point density, the system reconstructs the forming limit curve using a piecewise cubic spline interpolation algorithm to ensure smooth continuity. During reconstruction, appropriate boundary conditions are set for each segment of the curve to ensure continuity of the first- and second-order derivatives of adjacent segments at the connection points. To calculate the safety margin, the system first calculates the vertical distance from each sampling point on the strain path curve to the reconstructed forming limit curve. For example, the vertical distance from point (0.18, 0.06) on the strain path curve to the forming limit curve may be 0.12. The system constructs a nonlinear weighting function so that points closer to the forming limit curve have greater weights. For example, a weighting function W(d) = 1 / (1 + 5d) can be used, where d is the vertical distance. When d = 0.12, the weight is 0.625; when d = 0.05, the weight is 0.8.

[0103] The system uses this nonlinear weighting function to calculate the safety margin component value for each sampling point, which is the product of the vertical distance and the corresponding weight. The safety margin component values of all sampling points are then weighted and accumulated to obtain the safety margin value for that grid node. For example, if the safety margin component values of the sampling points for a node are 0.075, 0.096, 0.048, and so on, the weighted cumulative safety margin value is 0.108.

[0104] At the same time, the strain acceleration, or the rate of change of the strain rate, is calculated for each mesh node. For example, if the principal strain of a node increases from 0.28 to 0.35 at a strain rate of 0.7 between t = 0.4 and t = 0.5 seconds, and from 0.35 to 0.46 at a strain rate of 1.1 between t = 0.5 and t = 0.6 seconds, the strain acceleration of the node at t = 0.5 seconds is 0.4.

[0105] The system combines the safety margin value and strain acceleration of each grid node into a risk determination feature pair. Potential risk nodes are identified based on preset dual-threshold constraints (e.g., safety margin value less than 0.1 and strain acceleration greater than 0.3). For identified risk nodes, the system calculates the spatial distance and strain gradient continuity between them. If the spatial distance between adjacent risk nodes is less than the grid feature size (e.g., 2 mm) and the strain gradient changes continuously (gradient difference less than 15%), these nodes are merged into an initial risk region. For example, if the spatial distance between nodes A (50, 30), B (52, 30), and C (51, 32) is less than 2 mm and the strain gradients are 0.023, 0.025, and 0.026, respectively, with a difference of less than 15%, they can be merged into a single initial risk region. The boundaries of the initial risk region are modified based on the strain gradient distribution of the nodes. At the boundary, if the strain gradient changes suddenly (the rate of change is greater than 30%), the boundary position is adjusted. For example, at the boundary node D (54, 30), the strain gradient changes from 0.025 to 0.016, a rate of change of 36%. This exceeds the 30% threshold, and node D is excluded from the risk zone. In this way, the system determines the final deformation risk zone and provides accurate risk warnings for the stamping process.

[0106] Figure 2 This is a comparison diagram of the safety margin weight function of the embodiment of the present invention, as shown in FIG. Figure 2 As shown in the figure, the comparison of different safety margin weight functions is shown. The solid line in the figure represents the nonlinear weight function of the present technical solution, the dotted line represents the linear weight function used by the traditional Zhenga algorithm, and the dot-dashed line represents the exponential weight function used by the MLFCD algorithm. The nonlinear weight function of the present technical solution has three obvious characteristic ranges: when the vertical distance d<0.3, the function exhibits a high sensitivity characteristic near P1 (0.2, 0.78) with a slope of approximately -1.25; when 0.3≤d<0.6, the function exhibits a medium sensitivity characteristic near P2 (0.4, 0.45) with a slope of approximately -0.78; when d≥0.6, the function exhibits a low sensitivity characteristic near P3 (0.58, 0.10) with a slope of approximately -0.25. This piecewise nonlinear design enables the weight function to exhibit different sensitivities in different risk ranges. Specifically, when the vertical distance is small (close to the forming limit curve), the weight value changes more sensitively, more accurately reflecting the material's forming limit state. When the vertical distance is large (far from the forming limit curve), the weight value changes relatively smoothly, avoiding misjudgments caused by oversensitivity. Compared with the traditional Zhenga algorithm and MLFCD algorithm, the nonlinear weight function of this technical solution exhibits superior recognition characteristics in risk assessment, reducing the missed alarm rate in high-risk areas by 31.7% and the false alarm rate in low-risk areas by 28.9%, while improving overall recognition accuracy by 40.2%.

[0107] The existing technology usually relies on static strain criteria or a single safety margin assessment to identify deformation risks during the stamping process, which makes it difficult to capture dynamic changes in the forming process, especially in areas where the relationship between the strain path and the forming limit is complex and the evolution is drastic. There are problems of risk judgment lag and insufficient regional identification accuracy. This application constructs a distance mapping relationship between the strain path curve and the forming limit curve, introduces gradient change rate analysis, and realizes dynamic sampling encryption of high-change sections, effectively improving the accuracy of curve reconstruction; further combines the nonlinear weight function to perform a differentiated assessment of the safety margin of each sampling point, so that the safety margin value can more truly reflect the deformation margin of each node relative to the instability boundary. At the same time, by introducing strain acceleration into the risk judgment process and constructing a two-variable judgment feature pair, the deficiency of the existing method that it is difficult to identify the rapid deformation trend area with only static indicators is overcome; based on spatial distance and strain gradient continuity, the clustering of potential risk nodes and regional boundary correction are completed, which improves the consistency and accuracy of risk area identification. The above improvements are based on the capture of dynamic evolution characteristics. Without significantly increasing the computational complexity, they achieve accurate early warning and spatial positioning of local forming instability trends, thereby enhancing the timeliness and reliability of risk identification.

[0108] In an optional embodiment, historical stamping forming data including defect types and defect characteristics is obtained, strain mutation characteristics of the defect formation process are analyzed, a correspondence between strain mutation characteristics and defect types is established, and defect warning criteria are obtained. The deformation risk area is analyzed according to the defect warning criteria to obtain predicted defect locations and defect types, including:

[0109] Obtain historical stamping forming data containing defect types and defect characteristics, calculate the strain gradient tensor, strain path curvature and energy dissipation rate of the historical stamping forming data, and obtain multi-dimensional strain mutation characteristics;

[0110] Recursive Bayesian estimation is used to update the weights of the multi-dimensional strain mutation features to generate fused feature data. Based on the defect types in historical stamping data and the fused feature data, the temporal variation characteristics and spatial distribution characteristics of the defect formation process are analyzed to establish a mapping relationship between strain mutation features and defect types.

[0111] Constructing a feature-defect association network based on the mapping relationship, analyzing the evolution path of the fused feature data in the association network, and establishing an early warning criterion including feature thresholds and defect evolution laws;

[0112] Strain field data and temperature field data are collected in real time, the main direction and main value change rate of the strain gradient tensor in the deformation area are extracted, and a stress transfer path diagram of the deformation area is established. The strain path curvature mutation points and the degree of fluctuation of the energy dissipation rate during the stress transfer process are analyzed. The deformation instability index is calculated in combination with the defect evolution law in the early warning criterion. The defect location is determined based on the propagation characteristics of the instability index, and the defect type is determined according to the strain state combination during the instability process.

[0113] For example, historical stamping data containing defect types and characteristics is first acquired. During implementation, 500 sets of stamping test data under different process parameters can be extracted from the database. Each set of data includes strain field data and temperature field data for each stage of the complete forming process, as well as the final workpiece quality rating and defect distribution images. This historical data covers five common defect types: wrinkles, cracks, springback, surface scratches, and material thinning.

[0114] The strain gradient tensor, strain path curvature and energy dissipation rate are calculated for the acquired historical stamping forming data to obtain multi-dimensional strain mutation characteristics. In the specific implementation, the workpiece surface is divided into 10,000 grid cells, and the displacement data of each grid cell under 200 time steps during the forming process are differentially calculated to obtain strain field data. For each grid cell, the strain gradient tensor between adjacent time steps is calculated, including the main direction and amplitude changes; the strain path curvature, that is, the rate of change of the strain direction, is calculated; and the energy dissipation rate, that is, the change in plastic power per unit time, is calculated. For example, during the forming process of a certain area of the workpiece, the main value of the strain gradient tensor changes from 0.05 to 0.25, the strain path curvature increases from 0.02rad / mm to 0.15rad / mm, and the energy dissipation rate increases from 2J / s to 18J / s.

[0115] Recursive Bayesian estimation was used to update the weights of multidimensional strain mutation features to generate fused feature data. Initially, the weights for the strain gradient tensor, strain path curvature, and energy dissipation rate were set to 0.3, 0.3, and 0.4, respectively. These weights were then recursively updated based on the correlation between the defect location and each feature in the historical data. After 50 iterative calculations, the weights converged to 0.45 for the strain gradient tensor, 0.35 for the strain path curvature, and 0.2 for the energy dissipation rate. Using these updated weights, the multidimensional features of each grid cell were weightedly fused to generate a comprehensive feature value.

[0116] Based on the defect types and fusion feature data in historical stamping forming data, the temporal variation characteristics and spatial distribution characteristics of the defect formation process are analyzed, and a mapping relationship between strain mutation characteristics and defect types is established. In specific implementation, for each defect type, the fusion feature data of the defect area and the non-defect area during the forming process are extracted to construct a feature vector. Through comparative analysis, it was identified that wrinkle defects are usually accompanied by an increase of more than 3 times in the strain path curvature in a short period of time (within 15ms); the principal value of the strain gradient tensor in the crack defect area exceeds 0.3 at the end of the forming stage (the last 25% of the forming stage); and the energy dissipation rate fluctuation amplitude in the material thinning area exceeds 5 times that of the surrounding area.

[0117] The mapping relationship is used to construct a feature-defect association network, and the evolution path of the fused feature data in the association network is analyzed to establish an early warning criterion containing feature thresholds and defect evolution laws. During implementation, 200 typical defect samples in historical data are used to construct a five-layer association network, and each layer represents a different forming stage. The network nodes represent the feature state, and the edges represent the state transition probability. Through network analysis, the early warning criteria for each defect type are obtained: when the rate of change of the principal value of the strain gradient tensor in a certain area exceeds 0.02 / s and the duration exceeds 0.5s, it is judged to be a high risk of wrinkles; when the number of mutation points of the strain path curvature is per unit area (100mm 2 ) and the energy dissipation rate fluctuation coefficient is greater than 0.8, it is judged to be at high risk of cracking; when the main direction of the strain gradient tensor changes by more than 30° in adjacent time steps and the energy dissipation rate is lower than 60% of the average value of the surrounding area, it is judged to be at high risk of rebound.

[0118] Real-time strain and temperature field data are collected to extract the principal directions and principal value rates of change of the strain gradient tensor within the deformation region, and a stress transfer path diagram is constructed within the deformation region. Specifically, 12 strain sensors and 8 temperature sensors distributed at key locations in the mold are used with a sampling frequency of 200 Hz to collect data during the forming process in real time. Based on the collected data, the principal directions and principal value rates of change of the strain gradient tensor for each grid cell are calculated, and a stress transfer path diagram is constructed, showing the direction and intensity of stress transfer from high to low stress.

[0119] The strain path curvature mutation points and energy dissipation rate fluctuations during stress transfer are analyzed, and the deformation instability index is calculated based on the defect evolution law in the early warning criterion. During implementation, locations where the strain path curvature changes by more than 150% in a short period of time (50ms) are identified as mutation points. The ratio of the standard deviation to the mean of the energy dissipation rate of each grid cell is calculated as the degree of fluctuation. Based on the early warning criterion, the instability index F is calculated as 0.45 × (strain gradient change rate / threshold) + 0.35 × (number of curvature mutations / threshold) + 0.2 × (energy fluctuation / threshold). When the F value exceeds 1, the area is determined to have a defect risk.

[0120] The defect location is determined based on the propagation characteristics of the instability index, and the defect type is determined based on the combination of strain states during the instability process. During implementation, the spatial distribution and temporal evolution characteristics of areas with an instability index F greater than 1 are analyzed. If the F value diffuses radially and the central F value exceeds 1.5, the center point is located as the starting point of the defect; if the F value is distributed in a band and the bandwidth is less than 5mm, the center line of the band area is located as the defect location. Based on the combined characteristics of the strain state of the defect area, such as the main value of the strain gradient, the curvature peak, and the energy dissipation characteristics, and compared with the early warning criteria, the defect is determined to be a specific type of wrinkle, crack, rebound, surface scratch, or material thinning.

[0121] In this embodiment, by fusion modeling of multi-dimensional strain mutation features in historical stamping forming data, constructing a feature-defect association network and extracting its evolution law, deep mining of defect formation mechanisms and type identification are achieved. By dynamically updating feature weights through recursive Bayesian estimation, the accuracy of expressing the contribution of different features to defect evolution is improved. Real-time collection of strain and temperature field data, combined with stress transfer path and instability index propagation analysis, can achieve early warning, precise positioning and type discrimination of defects. Overall, this method significantly enhances the defect prediction, risk identification and traceability diagnosis capabilities in complex stamping processes, providing support for high-precision quality control and intelligent decision-making.

[0122] In an optional embodiment, a stress transfer path diagram of the deformation region is established, the strain path curvature mutation points and the degree of fluctuation of the energy dissipation rate during the stress transfer process are analyzed, and the deformation instability index is calculated in combination with the defect evolution law in the early warning criterion. The defect location is determined based on the propagation characteristics of the instability index. The defect type is determined based on the combination of strain states during the instability process, including:

[0123] The deformation area is discretized into triangular grid units, and the difference in the main stress directions between adjacent grid units is calculated as the transfer impedance. The minimum spanning tree algorithm is used to connect the minimum transfer impedance units to obtain the stress transfer path diagram of the deformation area.

[0124] Collecting strain field data along the stress transfer path diagram to construct a strain path, calculating the curvature change of the strain path, and extracting the curvature mutation point and the strain energy combination to form a mutation feature vector;

[0125] determining an energy transfer path according to the stress transfer path diagram, calculating the strain energy dissipation rate, its fluctuation degree and main frequency component along the energy transfer path, and combining them into a fluctuation characteristic vector;

[0126] The mutation characteristic vector and the fluctuation characteristic vector are mapped on the stress transfer path diagram to obtain the energy dissipation fluctuation characteristic score and the stress transfer path characteristic score respectively. Combined with the defect evolution law in the early warning criterion, the deformation instability index is calculated by using a dynamic weight combination;

[0127] Establish a spatial distribution map based on the deformation instability index, calculate the deformation instability index gradient field, determine the instability expansion direction and instability expansion speed based on the propagation characteristics of the instability index, and predict the defect location;

[0128] The principal strain ratio is calculated using strain field data, the strain path non-proportionality is calculated based on the nonlinearity of the strain path, and the local strain concentration factor is calculated according to the strain distribution in the spatial distribution diagram. The principal strain ratio, strain path non-proportionality, and local strain concentration factor are combined into a strain state feature, and the defect type is determined based on the value of the strain state feature.

[0129] Exemplarily, for the deformation area to be detected, the full-field strain data of the deformation area is collected by digital image correlation (DIC). In this embodiment, a high-resolution camera is used to photograph the surface of the deformed specimen to obtain grayscale images before and after deformation, and the surface displacement field is calculated by a sub-pixel matching algorithm to obtain the strain distribution information of the deformation area. For example, for a 100mm×100mm metal sheet specimen, a 5-megapixel camera is used to photograph it, and displacement field data with a resolution of 0.05mm / pixel can be obtained. The deformation area is discretized into triangular mesh units. In this embodiment, the Delaunay triangulation algorithm is used to divide the deformation area into triangular units of uniform size, generating a total of about 10,000 triangular mesh units with a unit side length of about 0.5mm. For each triangular unit, its principal stress direction is calculated based on the strain data obtained by DIC.

[0130] The difference in the principal stress directions between adjacent triangular mesh elements is calculated as the transfer impedance. Specifically, for any two adjacent triangular elements A and B, their respective principal stress directions θA and θB are extracted, and the angle difference Δθ = |θA-θB| between the two directions is calculated. The angle difference is normalized to a value between 0 and 1 as the transfer impedance value. For example, when the principal stress directions of the two elements are completely consistent, the transfer impedance is 0; when the principal stress directions of the two elements are perpendicular, the transfer impedance is 1. After calculating the transfer impedance between all adjacent elements, the minimum spanning tree algorithm is used to connect the elements with the minimum transfer impedance. In the implementation process, the Prim algorithm can be used to construct a minimum spanning tree. Starting from a element at the boundary of the deformation area, the adjacent elements with the minimum transfer impedance are gradually added until all elements are connected into an acyclic graph structure, that is, the stress transfer path diagram of the deformation area is obtained.

[0131] Along the path in the stress transfer path diagram, strain field data is collected to construct a strain path. In this embodiment, along each node on the path, the values of the principal strains ε1 and ε2 are recorded to form a strain path {(ε1, 1, ε2, 1), (ε1, 2, ε2, 2), ..., (ε1, n, ε2, n)}. The curvature change of the strain path is then calculated, and the curvature of each point on the path is calculated by the three-point method. When the curvature value exceeds the threshold of 0.5, the point is marked as a curvature mutation point. In practical applications, for aluminum alloy specimens, when loaded to 80% of the yield stress, 5 curvature mutation points are detected in the area around the defect, and the mutation values range from 0.6 to 0.8.

[0132] The energy transfer path is determined based on the stress transfer path diagram, and the unit volume strain energy and its dissipation rate are calculated along the energy transfer path. For each point on the path, the unit volume strain energy U is calculated, and its spatial gradient ΔU / Δx is calculated as the strain energy dissipation rate. The spectral characteristics of the dissipation rate are analyzed using a fast Fourier transform to extract the dominant frequency component and fluctuation amplitude. In this embodiment, the dissipation rate fluctuation amplitude is within ±5% in healthy material areas, while the fluctuation amplitude in defective areas can reach ±30%, with the dominant frequency component concentrated between 0.3Hz and 0.5Hz.

[0133] The mutation characteristic vector and the fluctuation characteristic vector are mapped to the stress transfer path diagram respectively to obtain the energy dissipation fluctuation characteristic score and the stress transfer path characteristic score. In this embodiment, a scoring system of 0 to 100 is adopted, the score of the healthy area is usually less than 20, and the score of the defective area is more than 60. Combined with the defect evolution law of the material, the deformation instability index is calculated by dynamic weight combination. For aluminum alloy materials, the mutation characteristic weight is set to 0.6 and the fluctuation characteristic weight is set to 0.4. The calculated instability index is generally less than 0.2 in the healthy area, and can reach more than 0.7 in the defective area. According to the deformation instability index, a spatial distribution map is established and its gradient field is calculated. The direction with the largest gradient value is identified as the instability expansion direction, and the size of the gradient value corresponds to the instability expansion speed. In practical applications, when there is an area with a gradient value greater than 0.1 / mm in the instability index gradient field, and the area is radially distributed, the center of the area can be determined as the defect position. Finally, the strain field data is used to calculate the principal strain ratio λ = ε1 / ε2, the strain path non-proportionality NPL, and the local strain concentration factor LCF to form the strain state eigenvector (λ, NPL, LCF). The defect type is determined based on the range of this eigenvector's value: when λ > 5 and NPL < 0.2, it is identified as a crack defect; when 1 < λ < 3 and NPL > 0.5, it is identified as a hole defect; and when λ ≈ 1 and LCF > 3, it is identified as an inclusion defect. For example, for a steel plate containing a crack defect, the measured values of λ = 7.2, NPL = 0.15, and LCF = 3.6 successfully identify the crack type defect and accurately locate the defect with an error of less than 1 mm.

[0134] Figure 3 The spatial distribution of deformation instability index and gradient field thermal diagram of the embodiment of the present invention are as follows: Figure 3As shown in the figure, the deformation area is represented by an 8×4 grid, with the horizontal axis ranging from 0-140mm and the vertical axis ranging from 0-60mm. The specific instability index value is clearly displayed in each grid unit, ranging from 0.12 to 0.95, and the color transitions from light blue (low value) to dark blue (high value), forming an intuitive visual gradient. The core unstable area is located at the coordinates (80mm, 40mm), where the instability index reaches the highest value of 0.95. This position is specially marked as the predicted defect location by the black circle. The direction of instability expansion from this point to the surrounding area is marked by three arrows, pointing to the upper left, upper and upper right, indicating that the unstable area has a tendency to expand in multiple directions. The instability index gradient field is steepest in the central area. When spreading from the grid (80mm, 40mm) to the surrounding area, the index value drops rapidly from 0.95 to 0.88, 0.72, etc., forming a clear instability gradient. This spatial distribution map clearly reveals the propagation characteristics of the deformation instability index, providing an intuitive basis for predicting defect locations and determining the direction and speed of instability expansion. Compared with traditional thermodynamic methods, it can more accurately identify potential high-risk areas, especially the identification of high-gradient areas of the instability index, which is of great guiding significance for implementing preventive measures.

[0135] Existing techniques for defect prediction during stamping processes often rely on static thresholds or local strain overruns, making it difficult to effectively identify the dynamic evolution of defect formation. Furthermore, a lack of systematic modeling of stress and energy transfer paths results in inaccurate predictions of defect location and type. This application proposes discretizing the deformation region into a triangular mesh and constructing transfer impedances based on the principal stress direction differences. A minimum spanning tree algorithm is then used to construct a stress transfer path map, dynamically capturing the true stress propagation path. Furthermore, multi-source dynamic feature vectors, such as sudden changes in strain path curvature and energy dissipation fluctuations, are extracted. A dynamic weighted combination approach is then used to construct a deformation instability index, enabling continuous monitoring and early warning of defect evolution trends. Regarding defect identification, this application incorporates principal strain ratios, strain path non-proportionality, and local strain concentration factors as comprehensive strain state features. Compared to traditional methods that rely solely on strain amplitude, this approach significantly improves the discriminability of defect type determination. Based on the coupled stress-strain-energy transfer relationship, this improvement constructs a comprehensive dynamic instability assessment framework, enabling earlier and more accurate identification and location of potential defects in the forming process, significantly enhancing the intelligence and foresight of stamping quality control.

[0136] In an optional embodiment, the mesh refinement coefficient is calculated using the mesh optimization equation, and the initial mesh data is locally optimized according to the mesh refinement coefficient to obtain the optimized mesh data. The stamping forming analysis is performed using the optimized mesh data, and the output analysis results include:

[0137] Calculating the plastic power density within the unit cell based on the grid optimization equation, and establishing a stress-strain field distribution function including a deformation energy density term and a strain localization term according to the plastic power density;

[0138] Calculating the stress gradient and strain gradient of the mesh element using the stress-strain field distribution function, establishing a critical refinement threshold according to the plastic power density, and comparing the stress gradient and strain gradient with the critical refinement threshold to obtain a mesh refinement coefficient;

[0139] The mesh elements are locally encrypted according to the mesh refinement coefficient, a strain path function is introduced into the encrypted mesh elements, the previously acquired material strain hardening exponent is updated to the encrypted mesh data, finite element iterative calculations are performed using the optimized mesh data, the strain rate gradient and mesh refinement coefficient are updated based on the calculation results, the mesh density distribution is dynamically adjusted, and the stamping analysis results are output.

[0140] For example, the part geometry corresponding to the stamped structure is first discretized and converted into the initial mesh model required for structural finite element analysis using two-dimensional or three-dimensional meshing techniques. This initial mesh model contains multiple finite elements, each associated with material properties, boundary conditions, and load paths. The mesh density of the initial mesh is typically set based on the geometric complexity and the forming path. For example, a denser mesh is set in structurally sensitive areas such as part corners, ribs, and orifices, while larger elements are used in flat areas to reduce computational complexity.

[0141] Based on the initial finite element mesh model, a mesh optimization equation is constructed to evaluate the deformation response strength of each mesh unit. The mesh optimization equation is used to comprehensively evaluate the deformation energy density distribution and strain localization trend within the unit, and based on this, the plastic power density index is established. Plastic power density is a parameter that characterizes the plastic energy consumption rate of a material during loading. This parameter can be described by the product relationship between the material hardening behavior and the local strain rate. This index can reflect the degree of plastic activity of the mesh unit under the current loading stage. Areas with higher plastic power density represent more severe deformation and more concentrated risks, so a finer-grained simulation analysis is required.

[0142] In each mesh element, a stress-strain field distribution function is calculated based on the aforementioned plastic power density. This distribution function consists of two main components: a deformation energy density term and a strain localization term. The deformation energy density term describes the energy intensity absorbed per unit volume of material during plastic deformation and is generally directly related to the plastic power density. The strain localization term describes the strain gradient variation trend in a material along a specific direction and can identify potential locations prone to instability, such as local shear bands and necking regions.

[0143] Based on the stress-strain field distribution function, the stress gradient and strain gradient inside each grid unit are extracted. The stress gradient represents the rate of change of stress between different points in the grid, and the strain gradient represents the steepness of the strain distribution in space. High stress or strain gradients usually appear at deformation boundaries, sudden geometry or multi-path stress concentration areas, and are important indicators of potential defects or instability formation. Next, a critical refinement threshold is constructed based on the plastic power density. This threshold serves as the basis for judging whether the grid should be refined or not, and is used to measure whether the current grid unit needs further refinement. Specifically, when the plastic power density of a unit exceeds the critical threshold, or the combination of its stress gradient and strain gradient reaches a specified critical state, it can be determined that the unit has a potential instability trend and requires local refinement.

[0144] In practice, by setting multiple refinement levels corresponding to different critical refinement coefficients, different mesh refinement factors are assigned to regions with different response intensities. The mesh refinement coefficient is used to control the accuracy of cell segmentation; higher values indicate a greater degree of refinement. Based on the refinement coefficient, the initial mesh data is adjusted cell by cell, optimizing the mesh in high-risk areas through local densification, while maintaining the original mesh density or moderately simplifying it in low-risk areas. The mesh's continuity and quality must be maintained during the optimization process, for example by controlling parameters such as cell angle and area ratio to avoid generating overly small or highly distorted cells.

[0145] In this optimized mesh structure, a strain path function is introduced to describe the evolution of the deformation history. The strain path function reflects the changing strain state of a material during different loading stages. It can be obtained by fitting historical strain data and reflects the principal strain direction, strain amplitude, and strain increment characteristics of the material at a specific time point and loading path. The strain path function is mapped to each mesh element within the refined region, allowing these elements to dynamically represent the true deformation path in subsequent analysis.

[0146] The previously acquired material strain hardening exponent is also updated to the newly generated encrypted mesh data. The strain hardening exponent describes a material's ability to harden during plastic deformation. Different materials, or even the same material, exhibit different hardening behaviors at different temperatures and loading rates. Introducing the material strain hardening exponent helps accurately calculate the material's stress response and improves the accuracy of forming analysis.

[0147] After mesh refinement and parameter updates are complete, finite element iterative calculations are performed based on the optimized mesh data. This process uses explicit or implicit finite element methods to dynamically simulate the entire stamping process and output results such as stress fields, strain fields, and thickness distribution in real time. By re-evaluating the strain rate gradient and mesh refinement coefficient at each iteration, the mesh density distribution can be dynamically adjusted. For example, if a trend of rapid strain concentration is detected in the leading edge area of the part during the tenth iteration, the refinement coefficient in this area is increased by one level, and the mesh is further refined to achieve higher simulation accuracy.

[0148] Taking the stamping simulation of automobile door panels as an example, the door panel contour is divided into medium-density meshes in the initial segmentation stage, and the plastic power density threshold is set as the initial judgment standard. During the calculation process, it was found that the strain localization in the corners of the door panels, the roots of the tensile ribs and other locations was obvious, the plastic power density was much higher than the average value, and the stress gradient direction was concentrated. It was decided to increase the refinement coefficient of this area from 1.0 to 2.5, and automatically complete the local encryption. After updating the mesh, the strain hardening exponent variation curve of the dual-phase steel material with temperature is loaded into each mesh unit, and the explicit finite element method is used for iterative solution. In the later iteration, the system continues to dynamically adjust the mesh density based on the strain rate change trend, so that the prediction of the necking position and crack tendency area in the simulation results is more accurate, and highly consistent with the crack position of the test piece.

[0149] Ultimately, the system outputs stamping analysis results, including but not limited to unit stress-strain cloud maps, thickness variation diagrams, and crack risk zone prediction maps. These results can be used to guide actual die compensation, blank holder force adjustment, and process path optimization, significantly improving the yield and stability of actual stamping production.

[0150] In this embodiment, by introducing plastic power density and stress-strain gradient analysis, dynamic optimization of the mesh refinement process is achieved, which can accurately identify areas of severe deformation and locally encrypt the mesh, thereby improving the resolution and accuracy of the finite element analysis. By constructing a strain path function and updating material hardening parameters, the simulation process is made closer to the actual forming behavior, enhancing the physical reality of the model. At the same time, the mesh refinement and iterative adjustment process can adapt to the deformation evolution trend, effectively avoid over-calculation, and improve overall computing efficiency. This solution can significantly improve the stamping simulation's ability to predict instability risks such as cracks and necking, and provide a highly reliable reference for mold design and process optimization.

[0151] Figure 4 FIG. 1 is a schematic diagram of the structure of an artificial intelligence-assisted rapid analysis and prediction system for stamping part formability according to an embodiment of the present invention. Figure 4 As shown, the system includes:

[0152] The first unit is used to obtain the three-dimensional data of the stamping part and perform mesh division to obtain initial mesh data, and extract geometric deformation feature parameters from the initial mesh data;

[0153] The second unit is used to construct a strain path curve based on the geometric deformation characteristic parameters, calculate the strain increment between adjacent time points of the strain path curve, and calculate the strain acceleration based on the strain increment. The strain path curve is compared with the forming limit curve to obtain the safety margin value of each grid node. The deformation risk area is determined based on the strain acceleration and safety margin value.

[0154] The third unit is used to obtain historical stamping forming data containing defect types and defect characteristics, analyze the strain mutation characteristics of the defect formation process, establish the corresponding relationship between the strain mutation characteristics and the defect type, obtain the defect warning criterion, analyze the deformation risk area based on the defect warning criterion, and obtain the predicted defect location and defect type;

[0155] The fourth unit is used to calculate the strain rate gradient at the predicted defect location, combine the strain rate gradient with the material strain hardening exponent to construct a mesh optimization equation, use the mesh optimization equation to calculate the mesh refinement coefficient, locally optimize the initial mesh data based on the mesh refinement coefficient, obtain the optimized mesh data, use the optimized mesh data to perform stamping analysis, and output the analysis results.

[0156] According to a third aspect of an embodiment of the present invention, an electronic device is provided, including:

[0157] processor;

[0158] a memory for storing processor-executable instructions;

[0159] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0160] According to a fourth aspect of an embodiment of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method described above is implemented.

[0161] The present invention may be a method, an apparatus, a system and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for executing various aspects of the present invention.

[0162] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. Artificial intelligence-assisted rapid analysis and prediction method for stamping part formability, characterized by: include: Acquire the three-dimensional data of the stamping part and perform mesh division to obtain initial mesh data, and extract geometric deformation feature parameters from the initial mesh data; A strain path curve is constructed based on the geometric deformation characteristic parameters. The strain increments between adjacent time points of the strain path curve are calculated. The strain acceleration is calculated based on the strain increments. The strain path curve is compared with the forming limit curve to obtain the safety margin value of each grid node. The deformation risk area is determined based on the strain acceleration and safety margin value. Obtain historical stamping forming data containing defect types and defect characteristics, analyze the strain mutation characteristics of the defect formation process, establish the corresponding relationship between strain mutation characteristics and defect types, obtain defect warning criteria, analyze the deformation risk area based on the defect warning criteria, and obtain the predicted defect location and defect type; The strain rate gradient at the predicted defect location is calculated and combined with the material strain hardening exponent to construct a mesh optimization equation. The mesh optimization equation is used to calculate the mesh refinement coefficient. The initial mesh data is locally optimized based on the mesh refinement coefficient to obtain the optimized mesh data. The optimized mesh data is used to perform stamping analysis and output the analysis results.

2. The method according to claim 1, characterized in that The three-dimensional data of the stamping part is obtained and meshed to obtain the initial mesh data. The geometric deformation feature parameters extracted from the initial mesh data include: Using a 3D scanner to acquire and preprocess point cloud data of the stamping part, constructing initial mesh data based on the preprocessed point cloud data, the initial mesh data including node position information and unit topology relationships, calculating mesh quality parameters of the initial mesh data, and optimizing the mesh area based on the mesh quality parameters to obtain an optimized mesh model; Establishing a deformation feature calculation matrix based on the optimized mesh model, the deformation feature calculation matrix including node displacements, deformation gradients, and strain components, calculating a deformation tensor using the deformation feature calculation matrix, and solving a principal strain component and a secondary strain component using the deformation tensor; The wall thickness variation of each node is calculated based on the principal strain component and the secondary strain component, the depth variation value is calculated by combining the wall thickness variation with the node displacement, and a geometric deformation characteristic parameter including the principal strain component, the secondary strain component, the wall thickness variation and the depth variation value is constructed.

3. The method according to claim 1, characterized in that Constructing a strain path curve based on geometric deformation characteristic parameters, calculating the strain increments between adjacent time points of the strain path curve, and calculating the strain acceleration based on the strain increments include: The geometric deformation feature parameters are subjected to Gaussian filtering and mean filtering to obtain filtered deformation feature data, and the filtered deformation feature data are spatially reconstructed using cubic spline interpolation to obtain a continuously distributed deformation feature field; Establishing a primary strain field and a secondary strain field based on a continuously distributed deformation characteristic field, and fitting a strain path curve between the primary strain field and the secondary strain field using a nonlinear least squares method with an adaptive weight coefficient, wherein the strain path curve includes a primary strain component and a secondary strain component; The strain path curve is decomposed into orthogonal direction vectors to obtain a primary strain vector and a secondary strain vector. The primary strain vector difference and the secondary strain vector difference are calculated based on the displacement increment and time increment of adjacent time points to obtain a strain increment vector. The strain increment vector is dynamically accumulated within a sampling time window to obtain a strain increment cumulative value. The strain acceleration is calculated based on the rate of change of the strain increment cumulative value. The strain acceleration represents the dynamic characteristics of the local deformation.

4. The method according to claim 1, wherein Compare the strain path curve with the forming limit curve to obtain the safety margin value of each grid node. According to the strain acceleration and safety margin value, the deformation risk area is determined to include: A first set of sampling points is obtained by sampling at equal time intervals on a strain path curve, and a second set of sampling points is obtained by sampling at equal strain intervals on a forming limit curve. The first set of sampling points and the second set of sampling points are paired based on a minimum distance principle, and the distance between the paired points is calculated to construct a distance mapping matrix. A change section is identified based on a gradient change rate of values in the distance mapping matrix, and the sampling point density is increased within the change section. All sampling points are reconstructed using a piecewise cubic spline interpolation algorithm. Calculate the vertical distance from the sampling point on the strain path curve to the reconstructed forming limit curve, construct a nonlinear weight function, use the nonlinear weight function to calculate the safety margin component value of the sampling point, and obtain the safety margin value by weighted accumulation of the safety margin component values of all sampling points; The safety margin value and strain acceleration of each grid node are combined into a risk judgment feature pair. Potential risk nodes are identified based on the preset dual-threshold constraint. The spatial distance and strain gradient continuity between adjacent risk nodes are calculated. Risk nodes with spatial distances smaller than the grid feature size and continuously changing strain gradients are merged into the initial risk area. The boundaries of the initial risk area are corrected according to the strain gradient distribution of the nodes to determine the final deformation risk area.

5. The method according to claim 1, wherein Obtain historical stamping forming data containing defect types and defect characteristics, analyze the strain mutation characteristics of the defect formation process, establish the corresponding relationship between strain mutation characteristics and defect types, obtain defect warning criteria, analyze the deformation risk area based on the defect warning criteria, and obtain the predicted defect location and defect type, including: Obtain historical stamping forming data containing defect types and defect characteristics, calculate the strain gradient tensor, strain path curvature and energy dissipation rate of the historical stamping forming data, and obtain multi-dimensional strain mutation characteristics; Recursive Bayesian estimation is used to update the weights of the multi-dimensional strain mutation features to generate fused feature data. Based on the defect types in historical stamping data and the fused feature data, the temporal variation characteristics and spatial distribution characteristics of the defect formation process are analyzed to establish a mapping relationship between strain mutation features and defect types. Constructing a feature-defect association network based on the mapping relationship, analyzing the evolution path of the fused feature data in the association network, and establishing an early warning criterion including feature thresholds and defect evolution laws; Strain field data and temperature field data are collected in real time, the main direction and main value change rate of the strain gradient tensor in the deformation area are extracted, and a stress transfer path diagram of the deformation area is established. The strain path curvature mutation points and the degree of fluctuation of the energy dissipation rate during the stress transfer process are analyzed. The deformation instability index is calculated in combination with the defect evolution law in the early warning criterion. The defect location is determined based on the propagation characteristics of the instability index, and the defect type is determined according to the strain state combination during the instability process.

6. The method according to claim 5, characterized in that Establish a stress transfer path diagram for the deformation area, analyze the strain path curvature mutation points and energy dissipation rate fluctuations during the stress transfer process, calculate the deformation instability index based on the defect evolution law in the early warning criteria, determine the defect location based on the propagation characteristics of the instability index, and determine the defect type based on the strain state combination during the instability process, including: The deformation area is discretized into triangular grid units, and the difference in the main stress directions between adjacent grid units is calculated as the transfer impedance. The minimum spanning tree algorithm is used to connect the minimum transfer impedance units to obtain the stress transfer path diagram of the deformation area. Collecting strain field data along the stress transfer path diagram to construct a strain path, calculating the curvature change of the strain path, and extracting the curvature mutation point and the strain energy combination to form a mutation feature vector; determining an energy transfer path according to the stress transfer path diagram, calculating the strain energy dissipation rate, its fluctuation degree and main frequency component along the energy transfer path, and combining them into a fluctuation characteristic vector; The mutation characteristic vector and the fluctuation characteristic vector are mapped on the stress transfer path diagram to obtain the energy dissipation fluctuation characteristic score and the stress transfer path characteristic score respectively. Combined with the defect evolution law in the early warning criterion, the deformation instability index is calculated by using a dynamic weight combination; Establish a spatial distribution map based on the deformation instability index, calculate the deformation instability index gradient field, determine the instability expansion direction and instability expansion speed based on the propagation characteristics of the instability index, and predict the defect location; The principal strain ratio is calculated using strain field data, the strain path non-proportionality is calculated based on the nonlinearity of the strain path, and the local strain concentration factor is calculated according to the strain distribution in the spatial distribution diagram. The principal strain ratio, strain path non-proportionality, and local strain concentration factor are combined into a strain state feature, and the defect type is determined based on the value of the strain state feature.

7. The method according to claim 1, characterized in that The mesh refinement coefficient is calculated using the mesh optimization equation. The initial mesh data is locally optimized based on the mesh refinement coefficient to obtain the optimized mesh data. The stamping analysis is performed using the optimized mesh data. The output analysis results include: Calculating the plastic power density within the unit cell based on the grid optimization equation, and establishing a stress-strain field distribution function including a deformation energy density term and a strain localization term according to the plastic power density; Calculating the stress gradient and strain gradient of the mesh element using the stress-strain field distribution function, establishing a critical refinement threshold according to the plastic power density, and comparing the stress gradient and strain gradient with the critical refinement threshold to obtain a mesh refinement coefficient; The mesh elements are locally encrypted according to the mesh refinement coefficient, a strain path function is introduced into the encrypted mesh elements, the previously acquired material strain hardening exponent is updated to the encrypted mesh data, finite element iterative calculations are performed using the optimized mesh data, the strain rate gradient and mesh refinement coefficient are updated based on the calculation results, the mesh density distribution is dynamically adjusted, and the stamping analysis results are output.

8. An artificial intelligence-assisted rapid analysis and prediction system for stamping part formability, used to implement the method according to any one of claims 1 to 7, characterized in that: include: The first unit is used to obtain the three-dimensional data of the stamping part and perform mesh division to obtain initial mesh data, and extract geometric deformation feature parameters from the initial mesh data; The second unit is used to construct a strain path curve based on the geometric deformation characteristic parameters, calculate the strain increment between adjacent time points of the strain path curve, and calculate the strain acceleration based on the strain increment. The strain path curve is compared with the forming limit curve to obtain the safety margin value of each grid node. The deformation risk area is determined based on the strain acceleration and safety margin value. The third unit is used to obtain historical stamping forming data containing defect types and defect characteristics, analyze the strain mutation characteristics of the defect formation process, establish the corresponding relationship between the strain mutation characteristics and the defect type, obtain the defect warning criterion, analyze the deformation risk area based on the defect warning criterion, and obtain the predicted defect location and defect type; The fourth unit is used to calculate the strain rate gradient at the predicted defect location, combine the strain rate gradient with the material strain hardening exponent to construct a mesh optimization equation, use the mesh optimization equation to calculate the mesh refinement coefficient, locally optimize the initial mesh data based on the mesh refinement coefficient, obtain the optimized mesh data, use the optimized mesh data to perform stamping analysis, and output the analysis results.

9. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images