Contour accuracy compensation method and system for precision hot-press forming mold

Abstract

The present invention relates to a contour accuracy compensation method and system for a precision hot-press forming mold. The method comprises: delineating grids on a three-dimensional model of a hot-press forming mold to obtain mold surface grid data; performing curvature analysis to obtain a high-curvature region and a low-curvature region; performing high-density sampling on the high-curvature region, and performing low-density sampling on the low-curvature region, so as to obtain a sampling point set; using a coordinate measuring machine to perform three-dimensional coordinate measurement on the sampling point set to obtain surface contour data; performing comparative analysis on the surface contour data and a theoretical model to obtain a contour error distribution; on the basis of the contour error distribution, using an inverse distance weighted interpolation algorithm to perform interpolation operation on an entire mold surface to obtain a mold surface compensation amount distribution; and on the basis of the mold surface compensation amount distribution, correcting the three-dimensional model of the hot-press forming mold to generate a compensated mold processing model. The implementation of the present invention realizes intelligent and automated contour accuracy compensation.

Description

Contour accuracy compensation method and system for precision hot press forming die TECHNICAL FIELD

[0001] The present application relates to the technical field of mold contour compensation, particularly relates to a contour accuracy compensation method and system for precision hot press forming die. BACKGROUND

[0002] The forming accuracy of precision hot press forming technology directly affects the performance and reliability of products. However, in actual production, due to the influence of material properties, process parameters and die deformation and other factors, the actual contour of the hot press forming part often deviates from the theoretical design, affecting product quality.

[0003] Traditional mold design and manufacturing methods are difficult to effectively predict and compensate for these deviations, resulting in insufficient product accuracy, the need for multiple trial molding and mold repair, and increased production costs and cycle time. However, the existing contour accuracy compensation method still has some problems, such as unreasonable sampling point distribution, inaccurate curvature analysis, and low efficiency of interpolation algorithm, which further leads to low accuracy and efficiency of the existing technology. SUMMARY

[0004] The main purpose of the present application is to provide a contour accuracy compensation method and system for precision hot press forming die to realize the intelligentization and automation of contour accuracy compensation.

[0005] To achieve the above purpose, the present application provides a contour accuracy compensation method for precision hot press forming die, comprising the following steps:

[0006] Grid division is performed on the three-dimensional model of the hot press forming die to obtain die surface grid data;

[0007] Curvature analysis is performed on the die surface grid data, and the die surface is divided into high curvature area and low curvature area by setting the curvature threshold;

[0008] High-density sampling is performed on the high curvature area, and low-density sampling is performed on the low curvature area to obtain a set of sampling points;

[0009] Three-coordinate measuring machine is used to measure the three-dimensional coordinates of the set of sampling points to obtain the surface contour data of the actual forming part;

[0010] The surface contour data of the actual forming part is compared and analyzed with the theoretical model to calculate the contour error of each sampling point to obtain the contour error distribution;

[0011] Based on the contour error distribution, a reverse distance weighted interpolation algorithm is used to perform interpolation operation on the entire die surface to obtain the die surface compensation amount distribution;

[0012] According to the mold surface compensation amount distribution, a three-dimensional model of the hot press forming mold is corrected to generate a compensated mold machining model.

[0013] The application also provides a profile precision compensation system of a precision hot press forming mold, comprising:

[0014] A mesh division module is configured to divide a three-dimensional model of the hot press forming mold into meshes to obtain mold surface mesh data.

[0015] A curvature analysis module is configured to analyze the curvature of the mold surface mesh data, divide the mold surface into a high-curvature area and a low-curvature area by setting a curvature threshold.

[0016] A sampling module is configured to sample the high-curvature area at a high density and sample the low-curvature area at a low density to obtain a set of sampling points.

[0017] A measurement module is configured to measure the three-dimensional coordinates of the set of sampling points by using a three-coordinate measuring machine to obtain surface profile data of an actual forming part.

[0018] A calculation module is configured to compare and analyze the surface profile data of the actual forming part with a theoretical model, calculate the profile error of each sampling point, and obtain a profile error distribution.

[0019] An interpolation operation module is configured to perform interpolation operation on the entire mold surface based on the profile error distribution by using a reverse distance weighted interpolation algorithm to obtain a mold surface compensation amount distribution.

[0020] A generation module is configured to correct the three-dimensional model of the hot press forming mold according to the mold surface compensation amount distribution to generate a compensated mold machining model.

[0021] The application also provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method according to any one of the above embodiments when executing the computer program.

[0022] The application also provides a computer readable storage medium storing a computer program, wherein the computer program is executed by a processor to implement the steps of the method according to any one of the above embodiments.

[0023] In summary, the technical scheme provided by the present application realizes more reasonable grid density distribution by surface feature extraction, region division and grid quality evaluation on the three-dimensional model of the mold, realizes more accurate high and low curvature region division by using Gaussian curvature and average curvature calculation, combining clustering analysis and boundary detection, uses different sampling densities for high and low curvature regions according to curvature gradient distribution, ensures the accuracy of key regions, and improves the sampling efficiency. Based on the spatial distribution characteristics of the sampling points, the measurement path is planned, the measurement system error model is combined, and the efficiency and accuracy of three-coordinate measurement are improved. Through vector difference calculation, principal curvature direction decomposition and spectral analysis methods, comprehensive analysis of the contour error is realized, the reverse distance weighted interpolation algorithm is used, the spatial index and local interpolation strategy are combined, and the efficiency and accuracy of the compensation amount distribution calculation are improved. Through parameterized modeling and topology optimization, the accurate reconstruction and adjustment of the compensation model are realized, the machining precision of the compensated mold is improved, and the intelligentization and automation of the contour precision compensation are realized. BRIEF DESCRIPTION OF DRAWINGS

[0024] Fig. 1 is a schematic diagram of the steps of the contour precision compensation method of the precision hot press forming mold in an embodiment of the present application;

[0025] Fig. 2 is a structural block diagram of the contour precision compensation system of the precision hot press forming mold in an embodiment of the present application;

[0026] Fig. 3 is a structural schematic block diagram of a computer device in an embodiment of the present application.

[0027] The implementation, functional characteristics and advantages of the present application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0028] In order to make the purpose, technical scheme and advantages of the present application more clear and explicit, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0029] Referring to Fig. 1, the present embodiment provides a contour precision compensation method for a precision hot press forming mold, comprising the following steps:

[0030] S1, grid division is performed on the three-dimensional model of the hot press forming mold to obtain mold surface grid data;

[0031] Specifically, surface feature extraction is performed on the three-dimensional model of the hot press forming die to extract geometric feature data of the surface of the model from the geometric shape of the model, including shape, curvature and boundary information of different regions on the surface of the die. According to the geometric feature data, the surface of the die is divided into regions, the surface of the die is divided into a plurality of sub-regions, and curvature calculation is performed for each sub-region to obtain curvature distribution data of each sub-region. The curvature distribution data can reflect the bending degree and complexity of each part of the die surface. According to the curvature distribution data, the grid density of each sub-region is reasonably allocated, and a grid density distribution scheme is developed to ensure that higher grid density is allocated in high curvature areas and lower grid density is allocated in low curvature areas, so as to optimize the use efficiency of computing resources. Based on the grid density distribution scheme, initial grid division is performed on the surface of the die to obtain initial grid data. The initial grid data is evaluated for grid quality to identify problems and deficiencies in the grid cells. According to the grid quality evaluation results, local refinement of the grid cells is performed for the areas with problems to improve the quality and accuracy of the grid and generate refined grid data. In order to ensure the smoothness and optimality of the grid, the refined grid data is subjected to grid smoothing processing to reduce the abrupt changes between grid cells and improve the consistency and smoothness of the grid. After the smoothing processing is completed, the smoothed grid data is subjected to topology optimization to ensure the overall structure of the grid cells is reasonable, and boundary adaptability adjustment is performed to ensure that the grid can well fit the actual boundary of the die surface, and finally the die surface grid data is obtained.

[0032] S2, curvature analysis is performed on the die surface grid data, the die surface is divided into high curvature regions and low curvature regions by setting a curvature threshold;

[0033] Specifically, curvature analysis is performed on the mesh data of the mold surface. By calculating the Gaussian curvature and mean curvature of each mesh node, the curvature values of all nodes on the mesh are obtained. The curvature values can reflect the bending degree of different parts of the mold surface. By clustering analysis on the mesh nodes, the curvature distribution areas of the mold surface are preliminarily divided, reflecting the differences in curvature of different areas. Boundary detection is performed on the preliminary curvature distribution areas to clearly define the boundary contours of each area. The boundaries of each curvature area are clearly defined, so that more accurate region division can be performed in subsequent processing. After the boundary contours are detected, the adjacent curvature areas are merged for processing, thereby avoiding unreasonable region division due to excessive subdivision, and obtaining optimized curvature distribution areas. Statistical analysis is performed on the optimized curvature distribution areas to reflect the overall distribution of the curvature of the mold surface. According to the curvature distribution histogram, a best curvature threshold, i.e. an initial curvature threshold, is calculated. Local sensitivity analysis is performed on the initial curvature threshold to obtain a threshold adjustment coefficient, which is used to correct the initial curvature threshold to obtain a more accurate target curvature threshold. Based on the target curvature threshold, the mold surface is binarized. The entire mold surface is preliminarily divided into high curvature areas and low curvature areas according to the curvature. Since the preliminary division result may have certain roughness or discontinuity, morphological operations are performed to further correct it. Morphological operations can optimize the smoothness and continuity of the region boundaries, ensuring that the finally divided high curvature areas and low curvature areas are more accurate and reasonable.

[0034] S3, high-density sampling is performed on the high curvature areas and low-density sampling is performed on the low curvature areas to obtain a set of sampling points;

[0035] Specifically, boundary extraction is performed on the high-curvature region and the low-curvature region respectively to obtain region boundary contour lines, and the boundaries of the regions are defined. The high-curvature region and the low-curvature region are mesh subdivided according to the boundary contour lines to obtain subdivided mesh data, ensuring higher resolution in the high-curvature region to capture finer curvature changes, and maintaining lower mesh density in the low-curvature region to improve computational efficiency. Curvature gradient analysis is performed on the subdivided mesh data, and the curvature gradient distribution map obtained by the analysis is used to understand the curvature changes on the mold surface. The curvature gradient distribution map can intuitively show the degree of curvature change in different regions, providing a basis for determining the sampling density of each region. According to the curvature gradient distribution map, sampling density coefficients are set in the high-curvature region and the low-curvature region respectively to generate a sampling density distribution scheme. The scheme ensures that a higher sampling density is set in the high-curvature region, while a lower sampling density is set in the low-curvature region. Based on the sampling density distribution scheme, initial sampling points are generated in the high-curvature region and the low-curvature region respectively to obtain a plurality of initial sampling points covering different regions of the mold surface and reflecting the curvature characteristics of each region. Spatial distribution uniformity analysis is performed on the plurality of initial sampling points to evaluate the distribution of the sampling points on the entire mold surface and identify regions that may be too dense or too sparse. According to the sampling point distribution evaluation result obtained by the spatial distribution uniformity analysis, necessary optimization and adjustment are performed on the initial sampling points. Through the adjustment, the distribution of the sampling points on the mold surface is improved to be more uniform and reasonable, and an optimized sampling point set is obtained.

[0036] S4, three-dimensional coordinate measurement of the sampling point set is performed using a coordinate measuring machine to obtain surface profile data of the actual molded part;

[0037] Specifically, spatial distribution analysis is performed on the sampling point set to determine the distribution characteristics of the sampling points in three-dimensional space. The characteristics include the distance between sampling points, the degree of density, and the distribution law of the sampling points on the mold surface. Through spatial distribution analysis, the concentrated and sparse areas of the sampling points are identified. According to the spatial distribution characteristics of the sampling points, the measurement path is planned, and an optimal measurement path is developed. The optimized measurement path can minimize the measurement time while ensuring the measurement accuracy and data integrity. The three-coordinate measuring machine is calibrated and error compensated. The calibration is performed to determine the reference state of the measuring machine in the actual measurement environment, and then an error model of the measurement system is constructed to accurately describe the errors that may occur in the measuring machine under different operating conditions. The error model can help identify and compensate for measurement errors caused by mechanical, temperature changes or other external factors. Based on the optimized measurement path and error model, the motion control parameters of the three-coordinate measuring machine are set, and accurate measurement control instructions are generated to guide the measuring machine to move according to the planned path in actual operation, ensuring the measurement accuracy of the sampling points. The three-coordinate measuring machine measures the sampling point set point by point according to the measurement control instructions to obtain the original measurement data. Noise filtering and outlier detection are performed on the original measurement data. Noise filtering can eliminate random errors introduced during measurement, while outlier detection can help identify and remove obviously unreasonable measurement data points to obtain preliminary processed measurement data. Weighted averaging and interpolation processing are performed on the preliminary processed measurement data to generate continuous surface profile data. Weighted averaging processing can smooth the data and reduce the influence of measurement errors, while interpolation processing is used to fill in the gaps between measurement data, making the surface profile data more complete and smooth. Data compression and format conversion are performed on the continuous surface profile data to meet the needs of subsequent analysis and use. Through data compression, the resource consumption of storage and processing is reduced, while format conversion ensures that the data can be smoothly transmitted and used between different systems and software, obtaining the surface profile data of the actual molded part.

[0038] S5, comparing the surface profile data of the actual molded part with the theoretical model, calculating the profile error of each sampling point, and obtaining the profile error distribution;

[0039] That is, the surface profile data of the actual molded part is aligned with the theoretical model in the coordinate system, and the error caused by the difference in the coordinate system is eliminated. Under the unified coordinate system, the nearest point projection algorithm is used to project the surface profile data of the actual molded part onto the surface of the theoretical model to generate a corresponding point set on the surface of the theoretical model. The corresponding point set directly reflects the difference in spatial position between the actual molded part and the theoretical model. The vector difference calculation is performed on the surface profile data of the actual molded part and the corresponding point set on the surface of the theoretical model to obtain an initial profile error vector field. Each vector of the vector field represents the profile error of a sampling point, including the direction and size of the error. According to the initial profile error vector field, the principal curvature direction decomposition algorithm is used for error vector decomposition to decompose the initial profile error vector field into normal error and tangential error components. The normal error reflects the deviation perpendicular to the surface, while the tangential error reflects the deviation along the surface direction. Statistical analysis is performed on the normal error and tangential error components to obtain the error distribution characteristic parameters. Through statistical analysis of the error distribution characteristic parameters, a probability density model of the error distribution is established, which can accurately describe the probability distribution of the error on the entire mold surface. The error distribution probability model is subjected to outlier detection and elimination to eliminate obviously unreasonable or abnormal error data points, and a corrected profile error data set is obtained. The corrected profile error data set is subjected to interpolation processing of discrete error points to generate a continuous profile error distribution field. Through interpolation processing, the gaps between the measurement data are filled, making the entire error field smoother and more coherent. After obtaining the continuous error distribution field, spectral analysis is performed to extract the spectral characteristics of the error distribution. The spectral characteristics reflect the spatial frequency components of the error distribution, which helps to understand the source and nature of the error. Based on the spectral characteristics, the profile error distribution field is subjected to multi-scale decomposition to obtain the profile error distribution at different scales.

[0040] S6, based on the profile error distribution, a reverse distance weighted interpolation algorithm is used to perform interpolation operation on the entire mold surface to obtain the mold surface compensation amount distribution;

[0041] Specifically, the profile error distribution is spatially discretized to convert the continuous error distribution data into a set of discrete points with spatial positions and error values, obtaining a discrete error point set. By incorporating the discrete error point set into a unified coordinate framework, the topological relationship between the error points is constructed, reflecting the proximity and connectivity of the error points in space. Boundary detection is performed on the topological relationship of the error points to obtain the boundary profile of the mold surface. According to the detected mold surface boundary profile, the entire mold surface is regridded to generate uniformly distributed target grid points. Spatial indexing is constructed for the target grid points and the discrete error point set to generate a fast retrieval structure, enabling rapid positioning of the spatial relationship between error points and grid points in subsequent interpolation processes, improving computational efficiency. Based on the fast retrieval structure, the error sampling points within the influence domain of each target grid point are determined to form a local interpolation data set containing all error information related to each target grid point. Reverse distance weight calculation is performed on each error sampling point in the local interpolation data set to obtain a weight coefficient matrix. The basic principle of reverse distance weight calculation is that the closer the error sampling point to the target grid point, the greater its influence on the interpolation result, and therefore the greater the weight coefficient. Based on the weight coefficient matrix, combined with the local interpolation data set, weighted average calculation is performed on each target grid point to obtain the preliminary interpolation result, which preliminarily reflects the error compensation amount distribution on the mold surface. Laplace smoothing is performed on the preliminary interpolation result to eliminate possible local mutations or discontinuities, obtaining the smoothed compensation amount distribution. Laplace smoothing adjusts the values between points to make the interpolation result smoother and more uniform. Based on the smoothed compensation amount distribution, the entire mold surface is processed continuously to make the compensation amount distribution more natural and accurate, ultimately obtaining the compensation amount distribution of the mold surface.

[0042] S7, according to the mold surface compensation amount distribution, the three-dimensional model of the hot press forming mold is corrected to generate a compensated mold processing model.

[0043] Specifically, the mold surface compensation amount distribution is gridded to generate compensation amount grid data. The continuous compensation amount distribution is converted into a set of discrete grid points, which represent the compensation values at different locations on the mold surface. The obtained compensation amount grid data is grid-matched with the three-dimensional model of the hot-press forming mold to establish the correspondence between the model grid and the compensation amount grid. Interpolation calculation is performed on the correspondence between the model grid and the compensation amount grid to smoothly transfer the discrete compensation amount data to the entire model surface, ensuring that each grid node can accurately obtain the corresponding compensation information and obtain the compensation vector of each node on the model surface. According to the calculated compensation vector, the nodes on the model surface are offset in coordinates to generate the preliminary corrected model data. The coordinate offset directly reflects the adjustment of the model surface profile, making the model surface closer to the design requirements. The preliminary corrected model data is subjected to surface reconstruction to reorganize the discrete node data into a smooth surface, generating a continuous compensation model surface. According to the continuous compensation model surface, feature recognition is performed to extract the key geometric features in the model. The geometric features may include boundaries, surface intersection points, regions with significant curvature changes, etc. The extracted key geometric features of the model are subjected to parametric modeling to generate a parametric compensation model. The advantage of parametric modeling is that it not only makes the model easier to adjust and optimize, but also can impose geometric constraints on the model to ensure the structural integrity of the model during the adjustment process. Geometric constraint analysis is performed on the parametric compensation model to determine the topological structure relationship between the parts inside the model. Geometric constraint analysis can identify possible geometric conflicts or unreasonable structures in the model. The topological structure relationship of the model is optimized and adjusted to eliminate possible geometric conflicts or improve unreasonable structures. Based on the optimized compensation model structure, a compensation mold machining model is generated.

[0044] The compensation amount distribution of the mold surface is equalized and parameterized, and the complex compensation amount distribution is converted into a uniform parameterized compensation amount distribution field. The initial compensation amount grid is obtained by grid division according to the parameterized compensation amount distribution field. The initial compensation amount grid is evaluated for grid quality, and defects and unreasonable places in the grid are identified. Through optimization, the distortion or discontinuity that may exist in the grid is eliminated, and the optimized compensation amount grid is obtained. Meanwhile, the three-dimensional model of the hot press forming die is extracted for features to obtain key geometric features in the model, such as boundaries, sharp points, curved surfaces and the like. The three-dimensional model of the hot press forming die is reconstructed by parameterization, and the three-dimensional model is converted into a parameterized model that is convenient for adjustment and matching. The curvature field distribution is obtained by analyzing the principal curvature direction of the parameterized model and the optimized compensation amount grid, reflecting the curvature characteristics of each point on the model surface. According to the curvature field distribution, the parameterized model and the compensation amount grid are coarsely matched, and the corresponding relationship between the model surface and the compensation amount grid is initially established. The initial corresponding relationship is optimized by using the iterative closest point algorithm (ICP algorithm). The iterative closest point algorithm gradually reduces the error by continuously adjusting the distance between the matching points, and obtains the accurate point-to-point mapping relationship. Based on the accurate point-to-point mapping relationship, a continuous mapping function is constructed, so that the relationship between the model grid and the compensation amount grid is more stable and continuous. The mapping function can ensure that the corresponding compensation amount can be accurately applied to each grid point of the three-dimensional model, realize the accurate compensation of the model surface, and finally obtain the corresponding relationship between the model grid and the compensation amount grid.

[0045] In one example, the three-dimensional model of the hot press forming die is divided into a grid to obtain grid data of the die surface, including:

[0046] The three-dimensional model of the hot press forming die is extracted for surface features to obtain geometric feature data of the die surface;

[0047] The die surface is divided into a plurality of sub-regions according to the geometric feature data, and the curvature distribution data of each sub-region is obtained by calculating the curvature of each sub-region;

[0048] The grid density distribution scheme is obtained by assigning grid densities to each sub-region according to the curvature distribution data, and the initial grid division is performed on the die surface based on the grid density distribution scheme to obtain initial grid data;

[0049] The grid quality evaluation result is obtained by evaluating the grid quality of the initial grid data, and the grid cells are locally refined according to the grid quality evaluation result to obtain refined grid data;

[0050] The refined grid data is subjected to grid smoothing processing to obtain smoothed grid data, and the smoothed grid data is subjected to topology optimization and boundary adaptability adjustment to obtain mold surface grid data.

[0051] In this example, surface feature extraction is performed on the three-dimensional model of the mold to identify and extract the geometric feature data of the mold surface, including curved surfaces, boundaries, sharp points, concave and convex parts, and other geometric characteristics. Through feature extraction, the complex three-dimensional model is decomposed into multiple regions with specific geometric characteristics. According to the extracted geometric feature data, the mold surface is divided into regions. The mold surface is divided into multiple sub-regions, each of which has certain homogeneity in geometric characteristics. For example, if a mold has a complex curved surface and some relatively flat areas, the curved surface part can be divided into a sub-region, and the flat part is divided into another sub-region. Curvature calculation is performed on each sub-region to obtain curvature distribution data of each sub-region. The curvature calculation can be realized by the following formula:

[0052] ;

[0053] Wherein, K represents the curvature, R is the radius of the curved surface. The curvature reflects the bending degree of the surface in different regions. A larger curvature value corresponds to a part of the surface with a more dramatic change, and vice versa, a smaller curvature value corresponds to a part of the surface with a more flat surface. According to the curvature distribution data, the grid density distribution scheme of each sub-region is obtained. Higher grid density is allocated in high curvature area, and lower grid density is allocated in low curvature area. The allocation strategy can be represented by the following formula:

[0054] ;

[0055] Wherein, K represents the curvature, K represents the curvature, K represents the curvature, K represents the curvature, is a monotonically increasing function, indicating that the greater the curvature value, the higher the grid density. Based on the grid density distribution scheme, the initial grid division is performed on the mold surface to obtain the initial grid data. The initial grid data is evaluated for grid quality to identify potential defects or unreasonable aspects in the grid, such as excessively stretched grid cells or irregular shapes. The grid quality can be measured by shape parameters of the grid, such as the angle of the grid cell, the area ratio, etc. If the evaluation result shows that the grid quality in some areas does not meet the standard, local grid refinement is needed for these areas. The principle of local refinement is to improve the quality of the grid cell with poor quality by increasing the number of grid cells or adjusting the grid shape to obtain refined grid data. For example, in a high-curvature area, if the initial grid cell shape is too stretched, it may lead to a decrease in calculation accuracy, so the grid can be refined by increasing the number of grid nodes to make the cell shape closer to the ideal state. The quality of the refined grid data is smoothed. By adjusting the position of the grid node, the transition between adjacent cells is more natural, eliminating possible mutations or discontinuities. Smoothing usually uses the Laplace smoothing algorithm, which is implemented by the following formula:

[0056] ;

[0057] where, is the new position of the th node after smoothing, is the position of the th node before smoothing, is the neighbor node set of the th node, is the smoothing coefficient, usually between 0 and 1. Through this process, the smoothness of the grid is significantly improved, and the smoothed grid data is obtained. For the smoothed grid data, topology optimization and boundary adaptability adjustment are performed to ensure the reasonableness of the grid structure and the precise alignment with the actual boundary of the mold, obtaining the mold surface grid data. Topology optimization includes adjusting the connection relationship between grid cells to make the entire grid structure more stable and reasonable, while boundary adaptability adjustment adjusts the position of grid nodes at the boundary to make the grid more consistent with the actual geometric boundary of the mold.

[0058] In one example, curvature analysis is performed on the mold surface grid data, and the mold surface is divided into high-curvature and low-curvature regions by setting a curvature threshold, including:

[0059] Gaussian curvature and mean curvature calculations are performed on the mold surface grid data to obtain the curvature value of each grid node, and clustering analysis is performed on the grid nodes according to the curvature value to obtain the preliminary curvature distribution region;

[0060] Boundary detection is performed on the preliminary curvature distribution region to obtain a boundary contour, and adjacent regions are merged according to the boundary contour to obtain an optimized curvature distribution region;

[0061] Statistical analysis is performed on the optimized curvature distribution region to obtain a curvature distribution histogram, and a best curvature threshold is calculated according to the curvature distribution histogram to obtain an initial curvature threshold;

[0062] Local sensitivity analysis is performed on the initial curvature threshold to obtain a threshold adjustment coefficient, and the initial curvature threshold is modified according to the threshold adjustment coefficient to obtain a target curvature threshold;

[0063] Based on the target curvature threshold, the mold surface is binarized to obtain a preliminary division result of high and low curvature regions, and morphological operations are performed on the preliminary division result to divide the mold surface into high curvature regions and low curvature regions.

[0064] In this example, Gaussian curvature and mean curvature are calculated for the mold surface grid data to obtain the curvature value of each grid node. Gaussian curvature and mean curvature are two important geometric quantities that describe the surface, which reflect the local bending degree and shape characteristics of the surface, respectively. Gaussian curvature ( ) can be calculated by the following formula:

[0065]

[0066] where and represent the principal curvatures, i.e., the maximum and minimum curvatures at a particular point. The sign and magnitude of the Gaussian curvature can reveal the concave-convex and curvature properties of the surface at that point: a positive value indicates an elliptic surface, a negative value indicates a hyperbolic surface, and a zero value indicates a flat or saddle-shaped surface. On the other hand, the mean curvature ( ) represents the overall bending degree of the surface at that point, which is defined as follows:

[0067]

[0068] ​​The grid nodes are clustered according to the curvature values. According to the curvature characteristics of the nodes, nodes with similar geometric properties are classified into the same group to obtain preliminary curvature distribution regions. Boundary detection is performed on the preliminary curvature distribution regions to accurately identify and divide the boundary contours of each region. According to the boundary contours, adjacent curvature regions are merged to effectively reduce over-segmentation of the regions, and more reasonable and optimized curvature distribution regions are obtained. Statistical analysis is performed on the optimized curvature distribution regions to generate a curvature distribution histogram. The curvature distribution histogram is a tool that reflects the distribution of curvature values on the entire mold surface. By analyzing the histogram, the frequency and distribution trend of different curvature values are reflected. The peak value and shape of the histogram can help determine the optimal curvature threshold to obtain an initial curvature threshold. Local sensitivity analysis is performed on the initial curvature threshold to evaluate the applicability of the initial curvature threshold in different regions and identify regions that may need adjustment. Through sensitivity analysis, a threshold adjustment coefficient is obtained, which is used to accurately correct the initial curvature threshold to obtain a final target curvature threshold. Based on the target curvature threshold, the entire mold surface is binarized. The curvature regions of the mold surface are divided into two categories: high curvature regions and low curvature regions. Morphological operations are performed on the preliminary division results, including dilation, erosion, opening operation, and closing operation, to further optimize the binarized regions, making the boundaries smoother and the interiors more coherent. For example, small isolated points can be removed by opening operation, and gaps in the boundary can be filled by closing operation, resulting in a more smooth and coherent high-low curvature region division result.

[0069] In one example, high-density sampling is performed on the high-curvature regions, and low-density sampling is performed on the low-curvature regions to obtain a set of sampling points, including:

[0070] Boundary extraction is performed on the high-curvature regions and the low-curvature regions respectively to obtain region boundary contour lines, and mesh subdivision is performed on the high-curvature regions and the low-curvature regions according to the region boundary contour lines to obtain subdivided mesh data;

[0071] Curvature gradient analysis is performed on the subdivided mesh data to obtain a curvature gradient distribution map, and sampling density coefficients are set for the high-curvature regions and the low-curvature regions according to the curvature gradient distribution map to obtain a sampling density distribution scheme;

[0072] Based on the sampling density distribution scheme, initial sampling points are generated for the high-curvature regions and the low-curvature regions to obtain a plurality of initial sampling points;

[0073] Spatial distribution uniformity analysis is performed on the plurality of initial sampling points to obtain a sampling point distribution evaluation result, and sampling point optimization adjustment is performed according to the sampling point distribution evaluation result to obtain a set of sampling points.

[0074] In this example, the curvature gradient refers to the rate of change of the curvature value in space. By performing boundary extraction on high curvature regions and low curvature regions respectively, the boundary contour lines of these regions are obtained. Based on the extracted region boundary contour lines, mesh subdivision is performed on the high curvature regions and low curvature regions respectively, and the mesh density inside the regions is improved to capture finer geometric features. For high curvature regions, due to the sharp change of the surface, a higher mesh density is needed to ensure accuracy, while for low curvature regions, a lower mesh density can be used to save computing resources. The curvature gradient analysis is performed on the subdivided mesh data to obtain the curvature gradient distribution map. The curvature gradient distribution map can reflect the change trend of the curvature in different regions. The region with a large curvature gradient means that the surface changes sharply at that place, and a higher sampling density is needed to capture the details. In the region with a small curvature gradient, the sampling density can be correspondingly reduced. This principle can be represented by the following formula:

[0075] ;

[0076] wherein, denotes the sampling density coefficient of the i-th mesh point, denotes the curvature gradient of the i-th point, denotes the sampling density coefficient of the i-th mesh point, denotes the curvature gradient of the i-th point, is an adjustment factor used to control the range of overall sampling density. Based on the sampling density coefficients calculated from the curvature gradient distribution map, a sampling density distribution scheme is formed, which ensures higher sampling density in high curvature gradient areas and lower sampling density in low curvature gradient areas, thus optimizing the distribution of sampling points. Based on the sampling density distribution scheme, initial sampling points are generated for high curvature areas and low curvature areas. A number of sampling points are randomly or regularly generated on the grid according to the sampling density coefficients of each grid point. Each sampling point represents a sampling location on the mold surface, which will be used for subsequent measurement and analysis. Spatial distribution uniformity analysis is performed on the sampling points to evaluate the distribution of initial sampling points on the entire mold surface. Uniform distribution can ensure that sampling points cover all important areas of the mold surface without excessive concentration or omission of certain areas. The uniformity of the distribution is evaluated by counting the distance between neighboring points of each sampling point, and if the sampling points in certain areas are too dense or sparse, optimization adjustment is needed. Based on the results of spatial distribution uniformity analysis, optimization adjustment of sampling points is performed to obtain a more reasonable set of sampling points. Optimization adjustment methods can include redistributing overly dense sampling points or increasing the number of sampling points in sparse areas to ensure uniform distribution of the final set of sampling points on the mold surface. For example, in high curvature areas, if the initial sampling points are too dense, it may lead to data redundancy, so the distribution can be optimized by reducing the number of sampling points; while in low curvature areas, if the number of sampling points in certain areas is too small, the distribution can be made more uniform by increasing the number of points.

[0077] In one example, a three-coordinate measuring machine is used to measure the set of sampling points in three-dimensional coordinates to obtain surface profile data of the actual molded part, including:

[0078] Spatial distribution analysis is performed on the set of sampling points to obtain sampling point spatial distribution characteristics, and measurement path planning is performed according to the sampling point spatial distribution characteristics to obtain an optimized measurement path;

[0079] The three-coordinate measuring machine is calibrated and error compensated to construct an error model of the measurement system, and the motion control parameters of the three-coordinate measuring machine are set according to the optimized measurement path and the error model to obtain measurement control instructions;

[0080] Based on the measurement control instructions, the set of sampling points is measured point by point to obtain raw measurement data, and noise filtering and outlier detection are performed on the raw measurement data to obtain preliminary processed measurement data;

[0081] Based on the preliminary processed measurement data, weighted average and interpolation processing are performed to obtain continuous surface profile data, and data compression and format conversion are performed on the continuous surface profile data to obtain surface profile data of the actual molded part.

[0082] In this example, a spatial distribution analysis is performed on the set of sampling points to obtain their distribution characteristics in three-dimensional space, including their density, uniformity, and coverage on the mold surface. Through analysis, regions of high or low sampling point density are identified. Path planning is a critical step in optimizing the efficiency of coordinate measuring machine measurements. Reasonable path planning can minimize measurement time while ensuring measurement accuracy and data integrity. To achieve optimized measurement path planning, the spatial distribution characteristics of the sampling points are used to design the measurement path using optimization algorithms. For example, the Traveling Salesman Problem algorithm or the Nearest Neighbor algorithm can be used for path optimization. The goal of the algorithm is to find the shortest path that allows the measuring machine to efficiently traverse all sampling points. Assuming the length of the measurement path is represented as , where represents the distance between the th sampling point and the th sampling point, the optimization objective can be represented as:

[0083] ;

[0084] Through the optimization process, the optimal solution of the measurement path is obtained, guiding the three-coordinate measuring machine to perform measurement operations in the shortest path order. The three-coordinate measuring machine is calibrated and error compensated. The purpose of calibration is to ensure the accuracy of the measuring machine in actual operation, while error compensation corrects the system errors that may occur during the measurement process by constructing an error model of the measurement system. The error model usually considers factors such as geometric errors of the measuring machine, errors caused by temperature, and nonlinear errors of the sensor. Through calibration and compensation, the accuracy consistency of the measuring machine under different conditions is ensured, and the influence of system errors on the measurement results is minimized. Based on the optimized measurement path and the constructed error model, the motion control parameters of the three-coordinate measuring machine are set, and the measurement control instructions are generated. The motion control parameter settings include the speed and acceleration of the measuring machine, the arrangement order of the measurement points, and the sampling frequency of the sensor during the measurement process, etc. The setting of these parameters needs to consider the complexity of the measurement path and the dynamic characteristics of the measuring machine to ensure that the data of each sampling point can be accurately and stably captured during the measurement process. After the motion control instruction setting is completed, the three-coordinate measuring machine will measure the sampling point set point by point according to these instructions. During the point-by-point measurement process, the measuring machine reaches each sampling point position in turn according to the optimized path and records the corresponding three-dimensional coordinate data, which constitutes the original measurement data. Noise filtering and outlier detection are performed on the original measurement data. The purpose of noise filtering is to eliminate random noise in the measurement data caused by equipment or environment, which can be processed using low-pass filter or Kalman filter technology. Outlier detection is used to identify and eliminate measurement data points that deviate significantly from the true value, which can be achieved through statistical analysis or model-based detection methods. For example, by calculating the residual of the measurement data points, if the residual of a certain data point exceeds the set threshold, it is considered as an outlier and is removed. After noise filtering and outlier detection, the preliminary processed measurement data is obtained. Weighted averaging and interpolation processing are performed on the preliminary processed measurement data. Weighted averaging processing smoothes the measurement data by weighted averaging of adjacent data points, reducing the influence of measurement error on the overall surface. Interpolation processing fills in the gaps between discrete points through mathematical methods to generate continuous surface data. For example, bilinear interpolation or cubic spline interpolation methods are used for interpolation processing. Assuming that in a two-dimensional plane, the coordinates of the known points are and , the corresponding measurement values are and , then the interpolation value of any point can be calculated by the following formula:

[0085] ;

[0086] The more continuous and accurate surface profile data is obtained through weighted average and interpolation processing. The continuous surface profile data is compressed and converted in format. The data compression can reduce the data storage, and the principal component analysis or Fourier transform technology can be used for compression, and the format conversion ensures the compatibility of the data in different systems or software. The surface profile data of the actual molded part is obtained.

[0087] In one example, the surface profile data of the actual molded part is compared with the theoretical model, the profile error of each sampling point is calculated, and the profile error distribution is obtained, including:

[0088] The surface profile data of the actual molded part and the theoretical model are aligned in the coordinate system, the data set in the unified coordinate system is obtained, and the nearest point projection algorithm is used to project the surface profile data of the actual molded part on the surface of the theoretical model according to the data set in the unified coordinate system, so as to obtain the corresponding point set on the surface of the theoretical model;

[0089] The vector difference calculation is performed on the surface profile data of the actual molded part and the corresponding point set on the surface of the theoretical model, the initial profile error vector field is obtained, and the principal curvature direction decomposition algorithm is used to decompose the error vector according to the initial profile error vector field, so as to obtain the normal error and tangential error components;

[0090] The normal error and tangential error components are statistically analyzed to obtain the error distribution characteristic parameters, and the error distribution probability model is obtained according to the error distribution characteristic parameters;

[0091] The error distribution probability model is detected and removed for abnormal values, and the modified profile error data set is obtained, and the discrete error point interpolation processing is performed according to the modified profile error data set, so as to obtain the continuous profile error distribution field;

[0092] The continuous profile error distribution field is analyzed to obtain the frequency spectrum characteristics of the error distribution, and the profile error distribution field is decomposed according to the frequency spectrum characteristics, so as to obtain the profile error distribution.

[0093] In this example, coordinate system alignment is performed between the surface contour data of the actual molded part and the theoretical model. The measurement data of the actual molded part and the data of the theoretical model are placed in the same coordinate system for direct comparison. Coordinate system alignment is achieved through a feature-point-based alignment method or the Iterative Closest Point (ICP) algorithm. The ICP algorithm adjusts the position and orientation of the two point sets by minimizing the distance between corresponding points, aligning the actual measurement data and the theoretical model in a unified coordinate system. The nearest point projection algorithm is used to project the surface contour data of the actual molded part. The actual measured point set is projected onto the surface of the theoretical model to find the closest point on the theoretical model for each measurement point. For each measurement point of the actual molded part, its closest point on the surface of the theoretical model is found, and the measurement point is projected onto that closest point. This is expressed by the following formula:

[0094] ;

[0095] in, It is the first on the actual molded part One point, It is the set of points on the surface of the theoretical model. This is the closest corresponding point on the theoretical model. The projection operation generates a set of corresponding points on the surface of the theoretical model, thus establishing a point-to-point correspondence between the actual molded part data and the theoretical model data. Vector difference calculation is performed on the surface contour data of the actual molded part and the corresponding point set on the surface of the theoretical model to obtain the initial contour error vector field. The purpose of the vector difference calculation is to determine the direction and magnitude of the deviation of each measurement point between the actual molded part and the theoretical model. Error vector It can be calculated using the following formula:

[0096] ;

[0097] in, It is the first Error vector at each point These are the measurement points on the actual molded part. These are the corresponding projection points on the theoretical model. The error vector field contains the deviation information of all measured points relative to the theoretical model. A matrix is ​​used to perform principal curvature direction decomposition on the error vector. The principal curvature direction decomposition algorithm is used to decompose the initial contour error vector field into normal and tangential error components. The normal error refers to the component of the error vector in the surface normal direction, while the tangential error refers to the component of the error vector in the surface tangential direction. The normal error typically reflects the deviation of the surface in the normal direction, while the tangential error reflects the offset of the surface in the tangential direction. This decomposition can be achieved using the following formula:

[0098] ;

[0099] where, denotes the normal error component, denotes the tangential error component. With this decomposition, the sources and nature of the errors are analyzed more clearly. Statistical analysis is performed on the normal and tangential error components to obtain distribution characteristic parameters of the errors, such as mean, variance, skewness, and kurtosis, etc. These parameters reflect the distribution law of the errors over the entire surface. Based on the distribution characteristic parameters, a probability density model of the error distribution is constructed. The error distribution probability density model describes the probability of the error occurring at different locations. Typically, this model can be achieved by estimating the probability density function of each error component. In order to improve the accuracy of error analysis, outlier detection and elimination are performed on the error distribution probability model. Through statistical methods or model-based methods, these outliers are identified and eliminated to obtain a corrected contour error data set. Based on the corrected contour error data set, interpolation processing of discrete error points is performed to generate a continuous contour error distribution field. The purpose of interpolation processing is to generate a smooth error distribution between the measurement points to more accurately describe the error situation on the entire surface. Spectral analysis is performed on the continuous contour error distribution field to extract the spectral features of the error distribution. Spectral analysis is used to identify the periodicity or frequency components in the error distribution, which may reflect specific features of the surface shape or periodic errors generated in the processing process. Through Fourier transform, the error distribution is converted from the spatial domain to the frequency domain to obtain the spectral representation of the error distribution. The spectral features help to understand the source and nature of the error. According to the spectral features, multi-scale decomposition is performed on the contour error distribution field to obtain error distributions at different scales. The process of multi-scale decomposition can reveal the variation law of the error at different scales, such as the relationship between detail errors and overall deviations. Through techniques such as wavelet transform, the error distribution is decomposed into multiple scale levels, each corresponding to different frequency components, to obtain a more comprehensive error analysis result.

[0100] In one example, based on the contour error distribution, a reverse distance weighted interpolation algorithm is used to perform interpolation operation on the entire mold surface to obtain the mold surface compensation amount distribution, including:

[0101] The contour error distribution is subjected to spatial discretization processing to obtain a discrete error point set, and a topological relationship of the error points is constructed according to the discrete error point set;

[0102] Boundary detection is performed on the topological relationship of the error points to obtain a mold surface boundary contour, and the mold surface is subjected to grid re-partitioning according to the mold surface boundary contour to obtain a uniformly distributed target grid point;

[0103] The target grid points and the discrete error point set are spatially indexed to construct a fast retrieval structure, and according to the fast retrieval structure, the error sampling points in the influence domain of each target grid point are determined to obtain a local interpolation data set;

[0104] The inverse distance weight calculation is performed on each error sampling point in the local interpolation data set to obtain a weight coefficient matrix, and according to the weight coefficient matrix and the local interpolation data set, the weighted average calculation is performed on each target grid point to obtain a preliminary interpolation result;

[0105] The Laplace smoothing processing is performed on the preliminary interpolation result to obtain a smoothed compensation amount distribution, and according to the smoothed compensation amount distribution, the mold surface is continuously processed to obtain a mold surface compensation amount distribution.

[0106] In this example, the continuous error distribution is transformed into a discrete set of error points. The continuous error information in the error distribution field is represented as a set of discrete points, each representing the error value at a specific location on the mold surface. A topological relationship is constructed based on this discrete error point set. This topological relationship reflects the proximity and connectivity of the error points in space. By constructing this topological relationship, we determine which error points are adjacent to each other and their geometric relationships. For example, algorithms such as Delaunay triangulation or Voronoi diagrams are used to connect the error point set into a network structure, describing the relative positional relationships between the error points. Boundary detection is performed on the topological relationship of the error points to identify the boundary contours on the mold surface. The goal of boundary detection is to extract the boundary contour lines of the concentrated error point regions on the mold surface, thus clearly distinguishing the inner and outer boundaries of the mold surface. Based on the boundary contours of the mold surface obtained from the boundary detection, the mold surface is re-meshed. Through homogenization, a set of uniformly distributed target mesh points is generated across the entire mold surface. The uniform distribution of mesh points improves the accuracy of interpolation calculations, ensuring the accuracy and continuity of the compensation distribution. Mesh re-partitioning typically employs quadrilateral or triangular mesh generation algorithms to generate a new mesh structure based on boundary conditions and internal node density. A spatial index is constructed for the target mesh points and the set of discrete error points, establishing a fast retrieval structure. This spatial index structure makes the search and matching process among a large number of data points more efficient. Common spatial indexing methods include kd-trees and R-trees, which quickly locate and retrieve error sampling points adjacent to each target mesh point by constructing a hierarchical index structure. Through spatial indexing, the set of error sampling points within the influence domain of each target mesh point is determined, forming a local interpolation dataset. The local interpolation dataset includes all error sampling points associated with the target mesh point. When calculating the inverse distance weight for each error sampling point in the local interpolation dataset, the inverse distance weighted interpolation method is typically used. The basic idea of ​​inverse distance weight is that the closer an error sampling point is to the target mesh point, the greater its influence and the greater its weight. The formula for calculating the inverse distance weight is as follows:

[0107] ;

[0108] in, Indicates the first The weights of each error sampling point This indicates the distance from the sampling point to the target grid point. is a distance attenuation exponent, usually taking a value of 2 or more. By calculating the weight coefficient matrix, the error sampling points in the local interpolation data set are weighted and averaged to obtain the preliminary interpolation results of each target grid point. The preliminary interpolation results are subjected to Laplace smoothing processing. The Laplace smoothing algorithm is usually iteratively calculated by the following formula:

[0109] ;

[0110] wherein, represents the position of a certain grid point in the preliminary interpolation results, is a smoothing coefficient, usually taking a value between 0 and 1, is the Laplace operator result of the point, representing the average position difference of the point and its neighbor points. Through multiple iterations, the smoothed compensation distribution is obtained, eliminating surface mutations or discontinuities caused by local errors. Based on the smoothed compensation distribution, the entire mold surface is subjected to continuous processing to generate a complete mold surface compensation distribution. Continuous processing achieves smooth transition of the mold surface compensation by interpolation calculation between discrete grid points, ensuring that the compensation result accurately reflects the shape requirements of the mold and provides accurate data for processing.

[0111] In one example, according to the mold surface compensation distribution, the three-dimensional model of the thermoforming mold is corrected to generate a compensated mold processing model, including:

[0112] The mold surface compensation distribution is subjected to grid processing to obtain compensation grid data, and the three-dimensional model of the thermoforming mold is subjected to grid matching according to the compensation grid data to obtain the corresponding relationship between the model grid and the compensation grid;

[0113] The corresponding relationship between the model grid and the compensation grid is subjected to interpolation calculation to obtain the compensation vector of each node on the model surface, and the model surface nodes are subjected to coordinate offset according to the compensation vector to obtain the preliminary corrected model data;

[0114] The preliminary corrected model data is subjected to surface reconstruction to obtain a continuous compensation model surface, and the model key geometric features are obtained according to the continuous compensation model surface;

[0115] The model key geometric features are subjected to parametric modeling to obtain a parametric compensation model, and the model topological structure relationship is obtained according to the parametric compensation model;

[0116] Optimize the topological structure of the model, and obtain the optimized compensation model structure. According to the optimized compensation model structure, a compensation mold machining model is generated.

[0117] In this example, the compensation amount distribution of the mold surface is gridded. The continuous compensation amount distribution data is discretized into a set of regular grid point data, obtaining the compensation amount grid data. The gridding process usually involves dividing the mold surface into a series of small grid units, each unit corresponding to a compensation amount value. The compensation amount grid data is matched with the three-dimensional model of the hot press forming mold. A one-to-one correspondence is established between the compensation amount grid and the model grid, so that each model grid point can obtain the corresponding compensation amount. The matching process can be realized by nearest neighbor search or mapping function. For each grid point of the model, find its nearest point in the compensation amount grid data, and assign the compensation amount value of the point to the model grid point. This process can be represented by the following formula:

[0118] ;

[0119] wherein, is the coordinate of the initial model grid point, is the corresponding compensation amount vector, is the compensated and revised grid point coordinate. Through this formula, the compensation vector in the compensation grid data is applied to each grid point of the model to obtain the preliminary revised model data. The preliminary revised model data is subjected to surface reconstruction to reorganize the discrete revised grid points into a continuous surface model, and this process usually uses mathematical tools such as NURBS (Non-Uniform Rational B-Spline) or Bezier surface. Through surface reconstruction, a smooth compensation model surface is generated to avoid the phenomenon of surface unevenness or discontinuity caused by the deviation of discrete points. The continuous compensation model surface is subjected to feature recognition. Key geometric features are extracted from the compensation model surface, which usually include boundaries, surface intersection points, protruding and recessed parts, etc. Feature recognition can be achieved by analyzing curvature changes, gradient information and geometric discontinuities. After completing feature recognition, the key geometric features of the model are subjected to parametric modeling, and the compensated geometric features are represented by mathematical functions or parameters, making the model more flexible and easy to operate in subsequent design and adjustment. Through parametric modeling, the geometric shape of the model is directly related to the design parameters, facilitating design changes or optimization adjustment. For example, a surface is represented by control points and weight parameters, so that when the position of the control point or the weight is adjusted, the shape of the surface will change, realizing flexible model optimization. After the parametric model is generated, geometric constraint analysis is performed according to the parametric model. The purpose of geometric constraint analysis is to ensure that the model always meets the geometric conditions and constraint relationships of the design during modification or optimization. By analyzing the relationship between different geometric features in the model, geometric conflicts or unreasonable places that may exist are identified and adjusted in the optimization process. Geometric constraint analysis usually involves topology analysis and optimization to ensure that the parts of the model have correct connection relationships and relative positions. The topology relationship of the model is optimized and adjusted. The goal of topology optimization is to eliminate geometric conflicts, simplify the model structure, and improve the machinability and stability of the model. Through optimization, a more reasonable and concise compensation model structure is generated, for example, when optimizing a complex surface, the topology structure of the surface is simplified by adjusting the grid density and control point position, making it easier to manufacture and process. A compensation mold processing model is generated according to the optimized compensation model structure. The generation of the processing model usually needs to consider the actual processing technology and equipment constraints to ensure that the model can be smoothly implemented in actual production. For example, when generating a processing model of an automobile body mold, the casting process, cooling system arrangement and assembly tolerance of the mold need to be considered to ensure the manufacturing precision and functionality of the mold.

[0120] In one example, the mold surface compensation amount distribution is subjected to grid processing to obtain compensation grid data, and a three-dimensional model of the hot press forming mold is subjected to grid matching according to the compensation grid data to obtain the correspondence relationship between the model grid and the compensation grid, including:

[0121] The mold surface compensation amount distribution is subjected to equal parameterization processing to obtain a parameterized compensation amount distribution field, and the compensation amount distribution is subjected to grid division according to the parameterized compensation amount distribution field to obtain an initial compensation amount grid;

[0122] The initial compensation amount grid is subjected to grid quality evaluation and optimization to obtain an optimized compensation amount grid, and the three-dimensional model of the hot press forming die is subjected to feature extraction to obtain a target model feature;

[0123] The three-dimensional model of the hot press forming die is subjected to parameterized reconstruction according to the target model feature to obtain a parameterized model, and the parameterized model and the optimized compensation amount grid are subjected to principal curvature direction analysis to obtain a curvature field distribution;

[0124] The parameterized model and the compensation amount grid are subjected to coarse matching according to the curvature field distribution to obtain a preliminary corresponding relationship;

[0125] The preliminary corresponding relationship is subjected to iterative closest point algorithm optimization to obtain an accurate point-to-point mapping relationship, and a continuous mapping function is constructed based on the point-to-point mapping relationship to obtain the corresponding relationship between the model grid and the compensation amount grid.

[0126] In this example, the mold surface compensation distribution is isoparametrically processed, transforming the complex and irregular compensation distribution field into a regular and uniform parameterized space to facilitate subsequent mesh generation and model matching. In this process, techniques such as B-spline or NURBS (Non-Uniform Rational B-Spline) are used to represent the compensation distribution field, resulting in a parameterized compensation distribution field. The parameterized compensation distribution field is meshed to generate an initial compensation mesh. The parameterized compensation distribution field is divided into a series of small grid cells, each representing the compensation value of a local area. Generally, mesh generation can be achieved through standard mesh generation algorithms, such as quadrilateral mesh generation or triangular mesh generation. The initial compensation mesh is evaluated and optimized for quality, checking whether the shape and distribution of the mesh meet the expected requirements, mainly checking whether the shape of the mesh is regular, whether there are excessive size differences between mesh cells, and whether there are stretching or distortion conditions. Quality evaluation can be achieved by calculating the shape parameters of the mesh cells (such as angle, area ratio, edge length ratio, etc.). If the mesh quality is found to be substandard, it needs to be improved through mesh optimization algorithms, such as local refinement or redivision to eliminate irregular mesh cells, resulting in an optimized compensation mesh. Feature extraction is performed on the three-dimensional model of the thermoforming mold to identify and extract key geometric features on the mold surface, such as boundary lines, surface intersection points, protrusions and recesses, etc. By analyzing the curvature, normal vector and other geometric properties of the model, the positions and shapes of these key features are determined. Based on the extracted target model features, the three-dimensional model of the mold is parameterized and reconstructed. The parameterized reconstruction process is to convert the original three-dimensional model into a parameterized model, so that its geometric shape can be described and controlled by a set of parameters. The advantage of parameterized model is that it can achieve precise control of model shape by adjusting parameters, which is of great significance in model optimization and matching. In this process, NURBS surface or other parameterized modeling techniques are used to construct the model, so that the complex shape of the model surface can be accurately described by parameterized representation. The principal curvature direction analysis is performed on the parameterized model and the optimized compensation mesh to determine the curvature direction and size of each point on the model surface. The principal curvature direction is usually determined by calculating the directions of the maximum and minimum curvature on the model surface, which describe the bending degree of the surface in different directions. Through principal curvature direction analysis, a curvature field distribution map is generated, which reflects the curvature variation of each part of the model surface. According to the curvature field distribution map, the parameterized model and the compensation mesh are coarsely matched to preliminarily determine the correspondence between them. This process can be achieved through nearest neighbor search or other geometric matching algorithms, pairing each grid point on the parameterized model with the point in the compensation mesh with the closest curvature, to obtain the preliminary point-to-point correspondence. The preliminary correspondence is optimized through the Iterative Closest Point (ICP) algorithm.ICP algorithm gradually reduces the error between the two by repeatedly adjusting the point-to-point correspondence, and obtains an accurate point-to-point mapping relationship. The basic principle of ICP algorithm is to minimize the distance between the corresponding points, and its objective function can be expressed as:

[0127] ;

[0128] wherein, represents the grid points of the parameterized model, represents the corresponding points in the compensation grid, is a rotation matrix, is a translation vector, is the distance error between the points. By iteratively optimizing and , the error function can be minimized to obtain an accurate mapping relationship. Based on the accurate point-to-point mapping relationship, a continuous mapping function is constructed. The continuous mapping function describes the spatial transformation relationship between the parameterized model grid and the compensation grid, so that they can correspond continuously and smoothly on the entire surface. Through the mapping function, accurate adjustment and compensation of the model surface are realized, so that the model can accurately reflect the design requirements and meet the actual processing needs.

[0129] Referring to FIG. 2, the embodiment provides a profile accuracy compensation system for a precision hot press forming die, comprising:

[0130] a grid division module 1 for dividing a three-dimensional model of the hot press forming die into a grid to obtain die surface grid data;

[0131] a curvature analysis module 2 for analyzing the curvature of the die surface grid data, dividing the die surface into a high-curvature area and a low-curvature area by setting a curvature threshold;

[0132] a sampling module 3 for high-density sampling of the high-curvature area and low-density sampling of the low-curvature area to obtain a set of sampling points;

[0133] a measurement module 4 for three-dimensional coordinate measurement of the set of sampling points using a three-coordinate measuring machine to obtain surface profile data of an actual formed part;

[0134] a calculation module 5 for comparing and analyzing the surface profile data of the actual formed part with a theoretical model, calculating the profile error of each sampling point, and obtaining a profile error distribution;

[0135] an interpolation operation module 6 for performing interpolation operation on the entire die surface based on the profile error distribution using a reverse distance weighted interpolation algorithm to obtain a die surface compensation amount distribution;

[0136] The generating module 7 is configured to correct the three-dimensional model of the hot press forming die according to the die surface compensation amount distribution, and generate a compensated die machining model.

[0137] In the embodiment, the specific implementation of each unit in the system embodiment is described above in the method embodiment, and will not be repeated here.

[0138] Referring to FIG. 3, the computer device in the embodiment of the application can be a server, and the internal structure thereof can be as shown in FIG. 3. The computer device comprises a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium. The database of the computer device is configured to store corresponding data in the embodiment. The network interface of the computer device is configured to communicate with an external terminal through a network connection. The computer program is executed by the processor to implement the above method.

[0139] Those skilled in the art can understand that the structure shown in FIG. 3 is only a block diagram of part of the structure related to the scheme of the application, and does not constitute a limitation on the computer device to which the scheme of the application is applied.

[0140] The embodiment of the application further provides a computer readable storage medium, which stores a computer program. The computer program is executed by the processor to implement the above method. It can be understood that the computer readable storage medium in the embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0141] In summary, the embodiment of the application provides a profile precision compensation method for a precision hot press forming die,

[0142] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above-mentioned embodiment methods. Any reference to memory, storage, database or other medium provided by the present application and used in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM, etc.

[0143] It should be noted that in this document, the terms "comprise", "comprising", or any other variant thereof are intended to cover non-exclusive inclusions, so that processes, systems, articles or methods that include a series of elements not only include those elements, but also include other elements not explicitly listed, or inherent to such processes, systems, articles or methods. Without more limitations, the element defined by the statement "comprises a" does not exclude the presence of another identical element in the process, system, article or method that includes the element.

[0144] The above description is only the preferred embodiment of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation, or direct or indirect application in other related technical fields, based on the content of the present application specification and drawings, are also included in the patent protection scope of the present application.

Claims

1. A profile accuracy compensation method of a precision hot press forming die, characterized by, The method comprises the following steps: grid division is performed on the three-dimensional model of the hot press forming die to obtain die surface grid data; curvature analysis is performed on the die surface grid data, the die surface is divided into a high-curvature region and a low-curvature region by setting a curvature threshold value; high-density sampling is performed on the high-curvature region, and low-density sampling is performed on the low-curvature region to obtain a set of sampling points; three-coordinate measurement is performed on the set of sampling points by using a three-coordinate measuring machine to obtain surface profile data of an actual forming part; the surface profile data of the actual forming part is compared and analyzed with a theoretical model, profile error of each sampling point is calculated, and profile error distribution is obtained; based on the profile error distribution, an inverse distance weighted interpolation algorithm is used to perform interpolation operation on the entire die surface to obtain die surface compensation amount distribution; the three-dimensional model of the hot press forming die is corrected according to the die surface compensation amount distribution to generate a die machining model after compensation.

2. The method of claim 1, wherein The grid division on the three-dimensional model of the hot press forming die to obtain die surface grid data comprises: surface feature extraction is performed on the three-dimensional model of the hot press forming die to obtain geometric feature data of the die surface; region division is performed on the die surface according to the geometric feature data to obtain a plurality of sub-regions, curvature calculation is performed on each sub-region, and curvature distribution data of each sub-region is obtained; grid density distribution scheme is obtained according to the curvature distribution data of each sub-region, initial grid division is performed on the die surface based on the grid density distribution scheme, and initial grid data is obtained; grid quality evaluation is performed on the initial grid data to obtain grid quality evaluation results, and local refinement of grid units is performed according to the grid quality evaluation results to obtain refined grid data; grid smoothing processing is performed on the refined grid data to obtain smoothed grid data, and topological structure optimization and boundary adaptability adjustment are performed on the smoothed grid data to obtain die surface grid data.

3. The method of claim 2, wherein the compensation is performed by a computer. The curvature analysis on the die surface grid data, the die surface being divided into a high-curvature region and a low-curvature region by setting a curvature threshold value, comprises: Gaussian curvature and average curvature calculation is performed on the die surface grid data to obtain curvature values of each grid node, and cluster analysis is performed on the grid nodes according to the curvature values to obtain preliminary curvature distribution regions; boundary detection is performed on the preliminary curvature distribution regions to obtain boundary contours, and adjacent regions are merged according to the boundary contours to obtain optimized curvature distribution regions; statistical analysis is performed on the optimized curvature distribution regions to obtain a curvature distribution histogram, and the best curvature threshold value is calculated according to the curvature distribution histogram to obtain an initial curvature threshold value; local sensitivity analysis is performed on the initial curvature threshold value to obtain a threshold adjustment coefficient, and the initial curvature threshold value is corrected according to the threshold adjustment coefficient to obtain a target curvature threshold value; Binaryzation is performed on the mold surface based on the target curvature threshold value to obtain a preliminary division result of high and low curvature regions, and morphological operation is performed on the preliminary division result to divide the mold surface into high and low curvature regions.

4. The method of claim 3, wherein The high curvature region is sampled at a high density and the low curvature region is sampled at a low density to obtain a set of sampling points, including: Boundary extraction is performed on the high and low curvature regions respectively to obtain region boundary contour lines, and mesh subdivision is performed on the high and low curvature regions according to the region boundary contour lines to obtain subdivided mesh data; Curvature gradient analysis is performed on the subdivided mesh data to obtain a curvature gradient distribution map, and sampling density coefficients are set for the high and low curvature regions according to the curvature gradient distribution map to obtain a sampling density distribution scheme; Based on the sampling density distribution scheme, initial sampling points are generated for the high and low curvature regions to obtain a plurality of initial sampling points; Spatial distribution uniformity analysis is performed on the plurality of initial sampling points to obtain a sampling point distribution evaluation result, and sampling point optimization adjustment is performed according to the sampling point distribution evaluation result to obtain a set of sampling points.

5. The method of compensation for profile accuracy of a precision hot press forming die according to claim 4, characterized in that, Three-coordinate measurement is performed on the set of sampling points by a three-coordinate measuring machine to obtain surface profile data of an actual molded part, including: Spatial distribution analysis is performed on the set of sampling points to obtain sampling point spatial distribution characteristics, and measurement path planning is performed according to the sampling point spatial distribution characteristics to obtain an optimized measurement path; Calibration and error compensation are performed on the three-coordinate measuring machine to construct an error model of the measurement system, and motion control parameter setting is performed on the three-coordinate measuring machine according to the optimized measurement path and the error model to obtain measurement control instructions; Point-by-point measurement is performed on the set of sampling points based on the measurement control instructions to obtain raw measurement data, and noise filtering and outlier detection are performed on the raw measurement data to obtain preliminary processed measurement data; Weighted average and interpolation processing are performed according to the preliminary processed measurement data to obtain continuous surface profile data, and data compression and format conversion are performed on the continuous surface profile data to obtain surface profile data of an actual molded part.

6. The method of compensation for profile accuracy of a precision hot press forming die according to claim 5, characterized in that, The surface profile data of the actual molded part is compared and analyzed with a theoretical model to calculate the profile error of each sampling point to obtain a profile error distribution, including: Coordinate system alignment is performed on the surface profile data of the actual molded part and the theoretical model to obtain a data set in a unified coordinate system, and nearest point projection algorithm is used to perform projection processing on the surface profile data of the actual molded part according to the data set in the unified coordinate system to obtain a corresponding point set on the surface of the theoretical model; Vector difference calculation is performed on the surface profile data of the actual molded part and the corresponding point set on the surface of the theoretical model to obtain an initial profile error vector field, and error vector decomposition is performed using principal curvature direction decomposition algorithm according to the initial profile error vector field to obtain normal error and tangential error components; Performing statistical analysis on the normal error and the tangential error component to obtain error distribution characteristic parameters, and performing error distribution probability density modeling according to the error distribution characteristic parameters to obtain an error distribution probability model; Performing outlier detection and elimination on the error distribution probability model to obtain a modified contour error data set, and performing discrete error point interpolation processing on the modified contour error data set to obtain a continuous contour error distribution field; Performing spectral analysis on the continuous contour error distribution field to obtain error distribution frequency spectrum characteristics, and performing multi-scale decomposition on the contour error distribution field according to the frequency spectrum characteristics to obtain a contour error distribution.

7. The method of compensation for profile accuracy of a precision hot press forming die according to claim 6, characterized in that, Based on the contour error distribution, performing interpolation operation on the entire mold surface by using a reverse distance weighted interpolation algorithm to obtain a mold surface compensation amount distribution, including: Performing spatial discretization processing on the contour error distribution to obtain a discrete error point set, and constructing a topological relationship of error points according to the discrete error point set; Performing boundary detection on the topological relationship of the error points to obtain a mold surface boundary contour, and performing grid re-partitioning on the mold surface according to the mold surface boundary contour to obtain a uniformly distributed target grid point; Performing spatial index construction on the target grid point and the discrete error point set to obtain a fast retrieval structure, and determining error sampling points in an influence domain for each target grid point according to the fast retrieval structure to obtain a local interpolation data set; Performing reverse distance weight calculation on each error sampling point in the local interpolation data set to obtain a weight coefficient matrix, and performing weighted average calculation on each target grid point according to the weight coefficient matrix and the local interpolation data set to obtain a preliminary interpolation result; Performing Laplace smoothing processing on the preliminary interpolation result to obtain a smoothed compensation amount distribution, and performing continuous processing on the mold surface according to the smoothed compensation amount distribution to obtain a mold surface compensation amount distribution.

8. The method of claim 1, wherein, According to the mold surface compensation amount distribution, performing modification on the three-dimensional model of the hot press forming mold to generate a compensated mold processing model, including: Performing grid processing on the mold surface compensation amount distribution to obtain compensation amount grid data, and performing grid matching on the three-dimensional model of the hot press forming mold according to the compensation amount grid data to obtain a correspondence relationship between model grid and compensation amount grid; Performing interpolation calculation on the correspondence relationship between model grid and compensation amount grid to obtain a compensation vector of each node of the model surface, and performing coordinate offset on the model surface node according to the compensation vector to obtain a preliminary modified model data; Performing surface reconstruction on the preliminary modified model data to obtain a continuous compensation model surface, and performing feature recognition on the continuous compensation model surface to obtain a model key geometric feature; Performing parameterized modeling on the model key geometric feature to obtain a parameterized compensation model, and performing geometric constraint analysis on the parameterized compensation model to obtain a model topological structure relationship; The model topological structure relationship is optimized and adjusted to obtain an optimized compensation model structure, and a compensated mold machining model is generated according to the optimized compensation model structure.

9. The method of compensation for profile accuracy of a precision hot press forming die according to claim 8, characterized in that, The mold surface compensation amount distribution is subjected to grid processing to obtain compensation amount grid data, and the three-dimensional model of the hot press forming mold is subjected to grid matching according to the compensation amount grid data to obtain a corresponding relationship between the model grid and the compensation amount grid, including: The mold surface compensation amount distribution is subjected to isoparametric processing to obtain a parameterized compensation amount distribution field, and the compensation amount distribution is subjected to grid division according to the parameterized compensation amount distribution field to obtain an initial compensation amount grid; The initial compensation amount grid is subjected to grid quality evaluation and optimization to obtain an optimized compensation amount grid, and the three-dimensional model of the hot press forming mold is subjected to feature extraction to obtain target model features; According to the target model features, the three-dimensional model of the hot press forming mold is subjected to parameterized reconstruction to obtain a parameterized model, and the parameterized model and the optimized compensation amount grid are subjected to principal curvature direction analysis to obtain a curvature field distribution; According to the curvature field distribution, the parameterized model and the compensation amount grid are subjected to rough matching to obtain a preliminary corresponding relationship; The preliminary corresponding relationship is subjected to iterative closest point algorithm optimization to obtain an accurate point-to-point mapping relationship, and a continuous mapping function is constructed based on the point-to-point mapping relationship to obtain the corresponding relationship between the model grid and the compensation amount grid.

10. A profile accuracy compensation system for a precision hot press forming die, characterized by, A system for performing the profile accuracy compensation method of the precision hot press forming mold according to any one of claims 1-9, the system comprising: a grid division module for dividing the three-dimensional model of the hot press forming mold into grids to obtain mold surface grid data; a curvature analysis module for analyzing the curvature of the mold surface grid data, dividing the mold surface into high-curvature regions and low-curvature regions by setting a curvature threshold; a sampling module for high-density sampling of the high-curvature regions and low-density sampling of the low-curvature regions to obtain a set of sampling points; a measurement module for measuring the three-dimensional coordinates of the set of sampling points using a three-coordinate measuring machine to obtain the surface profile data of the actual formed part; a calculation module for comparing and analyzing the surface profile data of the actual formed part with the theoretical model, calculating the profile error of each sampling point to obtain a profile error distribution; an interpolation operation module for performing interpolation operation on the entire mold surface based on the profile error distribution using a reverse distance weighted interpolation algorithm to obtain the mold surface compensation amount distribution; a generation module for correcting the three-dimensional model of the hot press forming mold according to the mold surface compensation amount distribution to generate a compensated mold machining model.

Images