Optimization design method and device for mold structure
Through parameterized modeling, geometric feature extraction and multi-scale grid division, combined with multi-physics analysis and genetic algorithms, the problems of complexity and low efficiency of laser tool mold design in traditional design methods are solved, and the comprehensive improvement of laser tool mold performance is achieved. and flexibility in optimizing results.
Patent Information
- Application Number
- CN202411361272.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-27
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-09-27
AI Technical Summary
Traditional laser tool mold design methods lack systematization and intelligence, and are difficult to meet the multi-objective optimization needs in complex operating conditions, with low computing efficiency and insufficient accuracy.
Parametric modeling, geometric feature extraction and adaptive multi-scale meshing, combined with multi-physics field coupling analysis and variance analysis, a multi-objective optimization model was constructed, and a non-dominant sorting genetic algorithm was used to solve it to determine the target optimization structural parameters.
It improves the performance of the laser tool mold, achieves the improvement of calculation accuracy and efficiency, provides comprehensive performance evaluation and multiple trade-offs, and enhances the practicality and flexibility of the optimization results.
Smart Images

Figure CN118886281B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of mold structure design, and in particular to a method and device for optimizing the design of a mold structure. Background Art
[0002] The structural design of laser die cutting directly affects the processing accuracy and production efficiency. Traditional design methods can no longer meet the multi-objective optimization requirements under complex working conditions. In practical applications, laser die cutting often faces multi-physical field coupling problems such as high temperature, high stress and large deformation. The interaction between these factors makes the die cutting structure optimization extremely complicated.
[0003] At present, the design of laser cutting dies mainly relies on engineering experience and repeated experiments, lacking systematic and intelligent optimization methods. This design method is not only time-consuming and labor-intensive, but also difficult to fully consider the trade-offs between multiple performance indicators. In addition, traditional finite element analysis methods often face the dilemma of low computational efficiency and insufficient accuracy when dealing with complex geometric shapes and multi-scale problems. Summary of the invention
[0004] The present application provides a method and device for optimizing the design of a mold structure, thereby being able to improve optimization efficiency while ensuring calculation accuracy, thereby achieving a comprehensive improvement in laser mold performance.
[0005] In a first aspect, the present application provides a method for optimizing the design of a mold structure, and the method for optimizing the design of a mold structure comprises:
[0006] Parametric modeling is performed on the three-dimensional model of the laser cutting die to obtain an adjustable parameter set;
[0007] Extracting and classifying the geometric features of the laser cutting die and obtaining a feature classification result from the three-dimensional model of the laser cutting die;
[0008] Based on the adjustable parameter set and the feature classification result, adaptively multi-scale meshing is performed on the three-dimensional model of the laser cutting die to obtain a multi-scale finite element analysis model;
[0009] Performing multi-physical field coupling analysis on the multi-scale finite element analysis model to obtain stress distribution data, deformation data and temperature field distribution data;
[0010] Performing variance analysis on the stress distribution data, the deformation data and the temperature field distribution data to obtain target influencing parameters, and constructing a multi-objective optimization model according to the target influencing parameters;
[0011] The multi-objective optimization model is solved by a non-dominated sorting genetic algorithm to obtain a Pareto optimal solution set, and the target optimization structural parameters of the laser cutting die are determined from the Pareto optimal solution set.
[0012] A second aspect of the present application provides a mold structure optimization design device, the mold structure optimization design device comprising:
[0013] A modeling module is used to perform parameterized modeling on the three-dimensional model of the laser cutting die to obtain an adjustable parameter set;
[0014] A feature classification module is used to extract and classify the geometric features of the laser cutting die on the three-dimensional model of the laser cutting die to obtain a feature classification result;
[0015] A meshing module, used for performing adaptive multi-scale meshing on the three-dimensional model of the laser cutting die based on the adjustable parameter set and the feature classification result to obtain a multi-scale finite element analysis model;
[0016] A coupling analysis module, used to perform multi-physical field coupling analysis on the multi-scale finite element analysis model to obtain stress distribution data, deformation data and temperature field distribution data;
[0017] A construction module is used to perform variance analysis on the stress distribution data, the deformation data and the temperature field distribution data to obtain target influencing parameters, and to construct a multi-objective optimization model according to the target influencing parameters;
[0018] The solution module is used to solve the multi-objective optimization model by non-dominated sorting genetic algorithm to obtain a Pareto optimal solution set, and determine the target optimization structural parameters of the laser cutting die from the Pareto optimal solution set.
[0019] The third aspect of the present application provides an electronic device, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the electronic device executes the above-mentioned mold structure optimization design method.
[0020] A fourth aspect of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned method for optimizing the design of the mold structure.
[0021] Compared with the prior art, the present application has the following beneficial effects: by parametric modeling and geometric feature extraction of the laser knife die, an adjustable parameter set and feature classification results are established, which provides a comprehensive and accurate data basis for subsequent optimization, and is conducive to improving the pertinence and efficiency of optimization. Based on the feature classification results and the adjustable parameter set, adaptive multi-scale meshing is realized, and refinement processing can be performed in key areas, while taking into account the computational efficiency, thereby improving the accuracy and efficiency of finite element analysis. By constructing the thermal-mechanical coupling control equation, a comprehensive analysis of the laser knife die under complex working conditions is realized, and stress distribution, deformation and temperature field data can be obtained at the same time, providing a comprehensive performance evaluation for optimization. The variance analysis method is used to identify key influencing parameters and construct a multi-objective optimization model, which can effectively reduce the complexity of the optimization problem and improve the optimization efficiency while considering multiple performance indicators. Using the non-dominated sorting genetic algorithm to solve the multi-objective optimization problem, a series of Pareto optimal solutions can be obtained, providing decision makers with multiple trade-off solutions, and enhancing the practicality and flexibility of the optimization results. By comprehensively scoring and analyzing the Pareto optimal solution set, the best balance point can be found among multiple optimization objectives to ensure that the final selected optimization structural parameters meet the performance requirements. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0023] The structures, proportions, sizes, etc. illustrated in the drawings of this specification are only used to match the contents disclosed in the specification so as to facilitate understanding and reading by persons familiar with the technology. They are not used to limit the conditions under which the present invention can be implemented, and therefore have no substantive technical significance. Any structural modification, change in proportion or adjustment of size shall still fall within the scope of the technical contents disclosed in the present invention without affecting the effects and purposes that can be achieved by the present invention.
[0024] Figure 1 It is a schematic flow chart of a method for optimizing the design of a mold structure provided by an embodiment of the present invention;
[0025] Figure 2 It is a schematic block diagram of the structure of the optimization design device of the mold structure provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0026] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0027] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.
[0028] It should also be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.
[0029] It should be further understood that the term "and / or" used in the specification and appended claims of this application refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations. Figure 1 , an embodiment of the optimization design method of the mold structure in the embodiment of the present application includes:
[0030] Step 100, parametric modeling is performed on the three-dimensional model of the laser cutting die to obtain an adjustable parameter set;
[0031] It is understandable that the execution subject of the present application can be a mold structure optimization design device, or a terminal or a server, which is not limited here. The present application embodiment is described by taking a server as the execution subject as an example.
[0032] Specifically, the three-dimensional model of the laser die is geometrically decomposed, and the structure of the laser die is divided into three main parts: the die body, the blade, and the connection structure. Independent parameterization of these parts helps to better control and optimize the overall performance of the laser die. When the die body is parametrically modeled, the size of the die body is parameterized. The thickness, length, and width of the die body are key size parameters, and the adjustment of these parameters will directly affect the strength and stability of the die. At the same time, the parameterization of the blade focuses on its shape characteristics, including the angle of the blade, the radius of the blade, and the length of the blade. The parameterization of the connection structure focuses on the optimization of its layout. The diameter, position, and number of the connection holes are the main parameters. By reasonably configuring these parameters, the installation stability and service life of the die are ensured. According to the attribute database of the die material, the material parameters are extracted and defined. The material properties of the die directly affect its mechanical properties and durability. The elastic modulus, Poisson's ratio, and density are important material parameters that must be extracted and defined. The processing technology specifications of the die are analyzed to obtain the surface roughness and heat treatment parameters. According to the use environment specifications of the knife die, the environmental conditions are quantified to obtain the working temperature range and humidity range parameters. The setting of environmental parameters can ensure the reliability and stability of the knife die in actual use. At the same time, the mechanical performance indicators of the knife die are quantitatively analyzed to obtain the maximum allowable stress and maximum allowable deformation parameters to ensure that the knife die does not fail under high-load working conditions. Establish the constraint relationship matrix between each parameter, clarify the dependency and mutual influence between different parameters, and obtain the parameter constraint equation group. Integrate and normalize all parameters and parameter constraint equation groups. Unify parameters of different dimensions into a standardized framework to obtain the final adjustable parameter set.
[0033] Step 200, extracting and classifying the geometric features of the laser cutting die and the three-dimensional model of the laser cutting die to obtain a feature classification result;
[0034] Specifically, the three-dimensional model of the laser die is voxelized. The three-dimensional model of the laser die is converted into voxel representation data to achieve a discretized representation of the model. Voxelization decomposes the three-dimensional model into a series of small cubic units, so that the geometric structure of the model can be expressed in a format that is easier for computers to process. The voxel representation data is processed by a three-dimensional convolutional neural network to extract local features of the model. The three-dimensional convolutional neural network extracts local feature maps of the geometric structure of the die layer by layer through multi-layer convolution operations. These feature maps can capture local shape features in the model, such as important information such as edges, corners and surface curvature. The local feature map is subjected to a maximum pooling operation. The maximum pooling operation effectively reduces the dimension of the feature map by selecting the maximum value for each area of the feature map while retaining key feature information. A nonlinear activation function operation is performed on the feature map after dimensionality reduction to enhance the nonlinear expression ability of the feature. Activation functions such as ReLU can introduce nonlinear factors, so that the neural network can better learn complex geometric shape features and obtain an activated feature map. Based on the activated feature map, the curvature of the die surface is analyzed. Curvature analysis can identify the changing trend of the die surface and obtain curvature distribution data, which can reflect the curvature degree and shape characteristics of the die surface. Perform cluster analysis on the curvature distribution data, and classify similar curvature areas into the same category through clustering operations to form preliminary feature areas. Perform boundary detection on the preliminary feature areas to identify the precise boundaries of each feature area and obtain feature boundary data. By analyzing the feature boundary data, adjacent feature areas are merged, and similar or related areas are integrated into larger and more representative feature areas to obtain optimized feature areas. Shape descriptors are calculated for the optimized feature areas. Shape descriptors are a mathematical expression that quantitatively describes geometric shape features and can convert complex geometric shape information into feature description vectors. Feature description vectors are classified, and feature areas are classified according to preset classification criteria to obtain feature classification results.
[0035] The voxel representation data is input into the input layer of the three-dimensional convolutional neural network for four-dimensional tensor format conversion to meet the requirements of network input, and the network input data for subsequent convolution operations is obtained. The network input data is processed by the first convolution layer. 32 3x3x3 convolution kernels are used for three-dimensional convolution operations, the step size of the convolution is set to 1, and the padding is set to 1. The convolution operation calculates each position of the original data, and the spatial size of the input data is kept unchanged through the padding operation to obtain the first layer of feature maps. In order to ensure that the data distribution in the feature map is more uniform and accelerate the training of the network, the first layer of feature maps is batch normalized. Batch normalization can reduce internal covariate shift and improve the stability of the model. The ReLU activation function is applied to the normalized feature map for nonlinear activation, so that the model can better capture complex geometric features. The first layer of feature maps after activation are subjected to the maximum pooling operation, using a 2x2x2 pooling kernel and a step size of 2. After pooling, the spatial dimension of the feature map is halved to obtain the first layer of downsampled feature maps. Max pooling not only effectively reduces the size of the feature map, but also retains the significance of local features. The second convolution layer is used for the first layer of downsampled feature maps. In this stage, 64 3x3x3 convolution kernels are used for three-dimensional convolution operations, and the step size is also set to 1, and the padding is set to 1 to ensure that the feature map after convolution still maintains the same spatial size, and the second layer of feature maps is obtained. Similar to the first layer, the second layer of feature maps are batch normalized and the ReLU activation function is applied again to enhance the nonlinear feature expression ability. After activation, the second layer of feature maps are also subjected to the maximum pooling operation, using the same 2x2x2 pooling kernel with a step size of 2. After the pooling operation, the second layer of downsampled feature maps are obtained. After this layer of processing, the size of the feature map is further reduced, but the significance of the features is enhanced. The third convolution layer is used for the second layer of downsampled feature maps, and 128 3x3x3 convolution kernels are used for three-dimensional convolution operations. The step size is still 1 and the padding is 1. The size of the feature map after convolution remains unchanged, and the third layer of feature maps is obtained. The feature map is batch normalized to ensure the uniformity of data distribution, and the ReLU activation function is applied to improve the network's expressiveness. The third layer activated feature map is subjected to global average pooling. Through global average pooling, the spatial dimension of the feature map is compressed to 1x1x1 to obtain a local feature map. The local feature map integrates the results of multi-layer processing of the three-dimensional convolutional neural network and contains key information for deep feature extraction of the input data.
[0036] Step 300: Based on the adjustable parameter set and the feature classification result, adaptive multi-scale meshing is performed on the three-dimensional model of the laser cutting die to obtain a multi-scale finite element analysis model;
[0037] It should be noted that, according to the feature classification results, the three-dimensional model of the laser knife die is divided into regions, and the model is decomposed into multiple important regions with different characteristics. According to the division results, the initial grid size is assigned to each region to obtain the initial grid size distribution map. Combined with the blade parameters in the adjustable parameter set, the blade region is locally meshed. Since the geometric features of the blade region are complex and the accuracy requirements of the model are high, a finer mesh division is used in this region. By refining the mesh in the blade region, more accurate finite element analysis results are ensured in the high stress concentration area. Based on the initial mesh size distribution map and the refined mesh of the blade region, a multi-level mesh division framework is constructed. This framework uses different mesh densities for different regions to achieve a multi-scale description of the model. The multi-level mesh division framework is adaptively refined to improve the fineness of the mesh and obtain a preliminary refined mesh model. Adaptive refinement is the process of dynamically adjusting the mesh density according to the geometric features and stress distribution of the model to ensure that there is sufficient mesh density in the key area to capture the details of stress changes. The mesh of the preliminary refined mesh model is locally reconstructed. By optimizing the quality and structure of the mesh and eliminating possible mesh distortion, the optimized mesh model is obtained. The boundary layer mesh is generated for the optimized mesh model. Especially in fluid or heat conduction analysis, the boundary layer area usually has significant gradient changes, so it is necessary to generate a boundary layer mesh with high resolution to accurately capture these changes. Through this operation, a mesh model with a boundary layer is obtained to ensure that the results of the finite element analysis are more accurate in areas with complex boundary conditions. The transition zone mesh generation and node number optimization are performed on the mesh model with a boundary layer to obtain a multi-scale finite element analysis model. In the process of transition zone mesh generation, the fine boundary layer mesh and the relatively coarse internal mesh are smoothly transitioned to avoid numerical errors caused by mesh mutations. Node number optimization aims to improve the computational efficiency of finite element analysis. The optimized node number can reduce the randomness of memory access during the calculation process and improve the overall performance of the analysis.
[0038] Step 400: Perform multi-physical field coupling analysis on the multi-scale finite element analysis model to obtain stress distribution data, deformation data and temperature field distribution data;
[0039] Specifically, the material properties of the multi-scale finite element analysis model are distributed. According to the material properties of different regions, each finite element is assigned corresponding material parameters such as elastic modulus, Poisson's ratio, thermal conductivity and specific heat capacity to construct a material property distribution model. According to the working conditions of the laser die, the corresponding loads and boundary conditions are applied to the material property distribution model. The loads and boundary conditions include mechanical loads, thermal loads and fixed constraints, which jointly determine the stress and deformation behavior of the model in the actual working environment. By applying loads and boundary conditions, the initial condition model is constructed. The thermal-mechanical coupling control equation is constructed for the initial condition model. A heat conduction equation is established to describe the heat transfer process in the material; then a thermal stress equation is established to describe the relationship between stress and strain caused by temperature changes, and a coupling equation group is obtained. The coupling equation group is time discretized, and a discretized equation group is obtained by discretizing the continuous time domain into several time steps. These equation groups solve the changes in temperature field and stress field in each time step. The discretized equations are spatially discretized and projected onto the nodes and elements of the finite element mesh to obtain the global stiffness matrix and load vector. The global stiffness matrix reflects the stiffness characteristics of each part of the model, while the load vector represents the mechanical response under the external load. Based on the global stiffness matrix and load vector, numerical solutions are performed to obtain the displacement field and temperature field of the model in each time step. The displacement field is strained, and the stress field is calculated through the strain distribution data and the constitutive relationship of the material to obtain the stress distribution data. The stress distribution data can reveal the stress concentration and possible failure areas inside the laser die under different load and temperature conditions. At the same time, the displacement field is post-processed, and the deformation data is obtained by extracting the node displacement information. These data indicate the deformation state of the model at each time step. For the temperature field, interpolation processing is used to obtain smoother and continuous temperature field distribution data, reflecting the temperature distribution of the model under thermal load.
[0040] Step 500: Perform variance analysis on stress distribution data, deformation data and temperature field distribution data to obtain target influencing parameters, and construct a multi-objective optimization model based on the target influencing parameters;
[0041] Specifically, the stress distribution data, deformation data and temperature field distribution data are preprocessed, and the stress, deformation and temperature field distribution data are standardized. By converting the data into a unified scale, the dimensional differences between different physical quantities are eliminated to obtain a standardized data set. The standardized data set is subjected to dimensionality reduction processing. By principal component analysis or other dimensionality reduction techniques, the high-dimensional data is converted into a low-dimensional feature matrix, the most representative features in the data are retained, the influence of redundant information is reduced, and the efficiency and accuracy of subsequent analysis are improved. The reduced-dimensional feature matrix is subjected to variance analysis to calculate the contribution of each feature to the optimization target. Through variance analysis, the importance of each feature in key performance indicators such as stress, deformation and temperature is quantified, and the feature importance ranking is obtained accordingly. The feature importance ranking reflects the degree of influence of different parameters on mold performance. According to the obtained feature importance ranking, the top K features with the highest contribution are selected to obtain the target influencing parameters. A multivariate linear regression analysis is performed on the selected target influencing parameters to establish a linear relationship model between these parameters and the optimization target, and a preliminary regression equation is obtained. The preliminary regression equation reveals the influence of each target influencing parameter on the stress, deformation and temperature field distribution by fitting the data. The residual analysis of the preliminary regression equation is carried out. By analyzing the difference between the model prediction value and the actual data, the deviation in the regression model is corrected to obtain a more accurate corrected regression equation. Based on the corrected regression equation, the objective function of the multi-objective optimization problem is constructed. The objective function usually includes minimizing the maximum stress, minimizing the maximum deformation, and minimizing the maximum temperature. The multi-objective function expression is obtained by integrating the objective function. According to the mold design specifications and process restrictions, the value range and constraint conditions of each design variable are determined. By reasonably setting the design variables, the constraint equation group is obtained, and the multi-objective function expression is combined with the constraint equation group to form a comprehensive objective function. The comprehensive objective function not only considers the needs of multi-objective optimization, but also incorporates the actual design and process constraints, making the optimization results more practical and feasible. A sensitivity analysis is performed on the comprehensive objective function to evaluate the influence of each design variable on the optimization results. Sensitivity analysis can identify the variables that are most sensitive to the optimization results, helping the optimization model to more accurately reflect the actual situation. Finally, a scientific and reasonable multi-objective optimization model is constructed.
[0042] The feature matrix after dimensionality reduction is grouped. According to the different types of features, the data are divided into stress group, deformation group and temperature group to form three independent grouped data sets. Descriptive statistical analysis is performed on the grouped data sets. By calculating the statistical characteristics of each group of data, such as the mean, variance, and standard deviation, the statistical characteristics of each group are described. Based on the statistical characteristics, a normality test is performed on each group of data to determine whether the data conforms to the normal distribution. After obtaining the distribution characteristics of the data, data transformation is performed to adjust the data to a suitable range or distribution form to obtain the transformed data set. This may include logarithmic transformation, square root transformation or other forms of nonlinear transformation of the data to make the data closer to the normal distribution. The transformed data set is tested for homogeneity of variance to determine whether the variances between different data groups are equal and to obtain the inter-group variance results. According to the results of the homogeneity of variance test, if a group of data meets the homogeneity of variance, the data is processed using a one-way analysis of variance (ANOVA); for those data groups with unequal variances, Welch's ANOVA is used for analysis. The results of ANOVA give the F statistic and the corresponding p-value for each group of data. The F statistic measures the significance of the difference between groups, while the p-value provides a criterion for judging statistical significance. Multiple comparisons are made on the F statistic and p-value to determine the significance level of the difference between different groups. Based on the significance level, the effect size of each feature is calculated to reflect the explanatory power of each feature on the target variable. By processing the effect size, the explained variance ratio of each feature is obtained, which indicates the contribution of each feature to the overall variance. The explained variance ratio is normalized to obtain the relative contribution of each feature to the target. All features are arranged in descending order according to the relative contribution to obtain the feature importance ranking.
[0043] Step 600: Solve the multi-objective optimization model using a non-dominated sorting genetic algorithm to obtain a Pareto optimal solution set, and determine the target optimization structural parameters of the laser cutting die from the Pareto optimal solution set.
[0044] Specifically, the multi-objective optimization model is encoded. The design variables are converted into binary strings for processing in the genetic algorithm. The initial population is generated by encoding, and these initial individuals serve as the starting point of the optimization process. The initial population is evaluated for fitness, and the fitness value of each individual under each objective function is calculated to form a fitness matrix. According to the fitness matrix, the population is non-dominated sorted. The non-dominated sorting stratifies the individuals according to their performance in the multi-objective function to obtain the non-dominated level of each individual. The lower the non-dominated level, the better the performance of the individual in the optimization objective. On this basis, the crowding degree of individuals with the same non-dominated level is calculated, and the density of each individual in the solution space is calculated to obtain the crowding degree value. Individuals with larger crowding degree values indicate better diversity in the solution space, so they should be retained preferentially in the selection operation. The selection operation is performed based on the non-dominated level and the crowding degree value, and individuals with high fitness and low crowding degree are selected to form the parent population. After the parent population is selected, a crossover operation is performed, and the genetic information of the parent individuals is combined through the crossover operation to generate new offspring individuals, and the offspring population after the crossover is obtained. The crossover operation explores a better solution space region by reorganizing the existing solutions. The offspring population after the crossover is subjected to polynomial mutation operation to introduce random disturbances, increase the diversity of the population, and prevent the algorithm from falling into the local optimal solution. The mutated offspring population will be merged with the parent population to form a larger merged population. The merged population contains all the information of the parent and offspring, in this way, ensuring that excellent individuals will not be lost due to intergenerational replacement. The merged population is subjected to selection operation to gradually eliminate individuals with poor performance until the size of the population returns to the predetermined size. The selection process will continue until the maximum number of iterations or convergence conditions are reached. The convergence condition is usually that the algorithm is considered to have reached the optimal state when it fails to find a significantly better solution within several generations. After these conditions are met, the Pareto optimal solution set is obtained, which contains a set of non-dominated solutions, each of which represents the optimal compromise between different objectives. Decision analysis is performed on the Pareto optimal solution set, and its performance on different objectives is evaluated by calculating the comprehensive score of each solution. The calculation of the comprehensive score can be based on actual needs, and the importance of each goal can be weighted and averaged to obtain the comprehensive performance score of each solution. The solution with the highest comprehensive score is selected as the target optimization structural parameter of the laser cutting die.
[0045] In the embodiment of the present application, by parametric modeling and geometric feature extraction of the laser knife die, an adjustable parameter set and feature classification results are established, which provides a comprehensive and accurate data basis for subsequent optimization, which is conducive to improving the pertinence and efficiency of optimization. Based on the feature classification results and the adjustable parameter set, adaptive multi-scale grid division is realized, and refinement can be performed in key areas, while taking into account the computational efficiency, thereby improving the accuracy and efficiency of finite element analysis. By constructing the thermal-mechanical coupling control equation, a comprehensive analysis of the laser knife die under complex working conditions is realized, and stress distribution, deformation and temperature field data can be obtained at the same time, providing a comprehensive performance evaluation for optimization. The variance analysis method is used to identify key influencing parameters, and a multi-objective optimization model is constructed, which can effectively reduce the complexity of the optimization problem and improve the optimization efficiency while considering multiple performance indicators. Using the non-dominated sorting genetic algorithm to solve the multi-objective optimization problem, a series of Pareto optimal solutions can be obtained, providing decision makers with multiple trade-offs, and enhancing the practicality and flexibility of the optimization results. By comprehensively scoring and decision-making analysis of the Pareto optimal solution set, the best balance point can be found between multiple optimization objectives to ensure that the optimized structural parameters finally selected meet the performance requirements.
[0046] In a specific embodiment, the process of executing step 100 may specifically include the following steps:
[0047] Decompose the three-dimensional model of the laser cutting die into geometric structure to obtain the cutting die body, cutting edge and connecting structure;
[0048] The size of the die body is parameterized to obtain the thickness, length and width parameters of the body, the shape of the blade is parameterized to obtain the blade angle, blade radius and blade length parameters, and the layout of the connection structure is parameterized to obtain the connection hole diameter, position and quantity parameters;
[0049] According to the knife die material property database, the material parameters are extracted and defined to obtain the elastic modulus, Poisson's ratio and density parameters. The processing specifications of the knife die are analyzed to obtain the surface roughness and heat treatment parameters.
[0050] According to the use environment specifications of the die cutter, the environmental conditions are quantified to obtain the working temperature range and humidity range parameters, and the mechanical performance indicators of the die cutter are quantitatively analyzed to obtain the maximum allowable stress and maximum allowable deformation parameters;
[0051] A constraint relationship matrix between various parameters is established to obtain a set of parameter constraint equations, and all parameters and parameter constraint equations are integrated and normalized to obtain an adjustable parameter set.
[0052] Specifically, the overall geometric structure of the laser die is decomposed. The complex three-dimensional model is decomposed into multiple parts with clear functions, including the die body, the blade, and the connection structure. The die body is parameterized. The die body is the core part of the entire structure, and its geometric dimensions directly affect the overall performance and working effect of the die. In the parameterization process, the main parameters include the thickness, length, and width of the body. For example, if the thickness of the die body is recorded as , the length is recorded as , the width is recorded as , these parameters can be based on specific design requirements or standard specifications. For example, for a specific material or working condition, the body thickness may be required to be It can be changed within a certain range to meet different manufacturing requirements. becomes an adjustable parameter whose range can be expressed as ,in and are the minimum and maximum allowable values of the thickness, respectively. At the same time, the shape of the blade has a crucial influence on the cutting performance of the die. When parameterizing the blade shape, key parameters such as blade angle, blade radius, and blade length are considered. Assume that the blade angle is recorded as The blade radius is , the blade length is , the selection of these parameters needs to be determined according to the type and thickness of the cutting material. For example, a larger blade angle May be suitable for harder materials, and smaller blade angles It is suitable for softer materials. and blade length The choice should also be optimized according to the requirements of cutting accuracy and die life. For the connection structure, its main function is to ensure the stable connection between the die and other equipment or components. The layout parameterization of the connection structure needs to pay attention to parameters such as the diameter, position and number of the connection holes. Assume that the diameter of the connection hole is recorded as , the position is recorded as , the quantity is recorded as The determination of these parameters should take into account the strength requirements of the mechanical connection and the convenience of installation. Connection hole diameter The range is usually defined by the specifications of the bolts or pins used, and the position It is necessary to reasonably distribute the material according to the geometry of the die and the usage scenario to ensure the best connection effect. According to the die material property database, the material parameters are extracted and defined. The material properties of the die are important factors affecting its mechanical properties, usually including basic material parameters such as elastic modulus, Poisson's ratio and density. Assume that the elastic modulus is recorded as , Poisson's ratio is recorded as , the density is recorded as , the selection of these parameters needs to match the working environment and load conditions of the die. For example, for a high-strength, high-rigidity die, the elastic modulus Need to be higher, and Poisson's ratio It should be within a suitable range to ensure that the die is not prone to breakage or permanent deformation under high stress conditions. Analyze the die processing specifications. The processing technology directly affects the surface quality and service life of the die. Important processing parameters include surface roughness and heat treatment parameters. The optimization of these parameters can improve the wear resistance and fatigue resistance of the die. For high-precision cutting requirements, surface roughness It should be as small as possible to reduce friction loss and improve cutting accuracy. According to the use environment specifications of the die cutter, the environmental conditions are quantified. This includes determining the working temperature range and humidity range of the die cutter in actual use. Assume that the working temperature range is recorded as , humidity range is These environmental parameters need to be considered in combination with material parameters to ensure that the die can maintain stable performance under harsh environmental conditions. For example, the material of the die may need to maintain a low thermal expansion coefficient under high temperature conditions to prevent changes in dimensional accuracy. The mechanical properties of the die are quantitatively analyzed. Important mechanical properties include the maximum allowable stress and the maximum allowable deformation. Assume that the maximum allowable stress is recorded as The maximum allowable deformation is recorded as The determination of these indicators needs to consider the working conditions of the die under extreme load conditions. Maximum allowable stress It is usually limited by the yield strength or fracture strength of the material, and the maximum allowable deformation It is necessary to determine through experiments or simulations to ensure that the die does not deform excessively or fail during use. Establish the constraint relationship matrix between various parameters and construct the parameter constraint equation group. Assume that a parameter constraint equation is in the form of:
[0053] ;
[0054] in, represents the constraints, For each parameter, is the weight coefficient, is a constant term. The constraint relationship matrix and equation group help ensure that all parameters can work together while meeting the design specifications to avoid conflicts between parameters. In order to improve the optimization efficiency, all parameters and parameter constraint equation groups are integrated and normalized. Through normalization, parameters of different dimensions can be unified to the same scale to obtain an adjustable parameter set.
[0055] In a specific embodiment, the process of executing step 200 may specifically include the following steps:
[0056] voxelize the three-dimensional model of the laser cutting die to obtain voxel representation data;
[0057] Process the voxel representation data using a three-dimensional convolutional neural network to obtain a local feature map;
[0058] Perform a maximum pooling operation on the local feature map to obtain a feature map after dimensionality reduction, and perform a nonlinear activation function operation on the feature map after dimensionality reduction to obtain an activated feature map;
[0059] According to the activated feature map, the curvature analysis is performed on the surface of the die to obtain the curvature distribution data, and the curvature distribution data is clustered to obtain the preliminary feature area;
[0060] Perform boundary detection on the preliminary feature region to obtain feature boundary data, and merge adjacent feature regions according to the feature boundary data to obtain optimized feature regions;
[0061] The shape descriptor is calculated for the optimized feature region to obtain a feature description vector, and the feature description vector is classified to obtain a feature classification result.
[0062] Specifically, the 3D model is converted into voxel representation data. Voxelization is a process of discretizing continuous 3D geometric data by dividing the 3D space into multiple small cubic units, namely voxels. Each voxel represents the properties of a small volume in the model. Assume that the 3D model of the laser cutting die contains Voxels can be represented by a three-dimensional matrix, where each matrix element The attribute value corresponding to a voxel. Voxelization converts complex 3D geometric shapes into discrete data structures that are easy to calculate and analyze, which is convenient for subsequent deep learning processing. The voxel representation data is processed by a 3D convolutional neural network. The 3D convolutional neural network gradually extracts the local feature map of the model through multiple convolutional layers. The convolution operation can be regarded as a filter sliding in the 3D space. The filter extracts the features of the area by performing a dot product operation with the local area. Suppose a convolutional layer uses The size is The convolution kernel is operated, and the output feature map It can be expressed as:
[0063] ;
[0064] in, It is The weight parameters of the convolution kernels, is the value of the voxel representing the data, is the bias term, Then is the output feature map after the convolution operation. Through multi-layer convolution and gradually stacked feature maps, the 3D convolutional neural network can capture the multi-level local features of the model and extract important information about the geometric shape of the die. The local feature map is subjected to the maximum pooling operation to reduce the spatial dimension of the feature map. Maximum pooling is a dimensionality reduction technique that achieves data compression by selecting the maximum value in the pooling window to represent the features of the area. Assume that the pooling window size is 2, then the feature map after pooling It can be expressed as:
[0065] ;
[0066] The maximum pooling operation reduces the size of the feature map and retains the most significant feature information, improving the computational efficiency of the model. The nonlinear activation function is performed on the feature map after dimensionality reduction. Common activation functions such as RelU can enhance the nonlinear expression ability of the model. Based on the activated feature map, the curvature of the die surface is analyzed. Curvature is an important geometric property to measure the degree of surface curvature and can reflect the subtle shape changes of the die surface. Assume that the distribution of curvature in space is expressed as , by calculating the principal curvature of each point in the feature map to obtain the curvature distribution data. Curvature analysis can help identify high curvature areas on the die surface, which usually correspond to parts with cutting sharpness or sudden shape changes. Cluster analysis is performed on the curvature distribution data to classify similar curvature areas into the same category. Through the clustering algorithm, preliminary feature areas are identified, which usually represent parts with similar geometric characteristics on the die, such as blade tips, corners, etc. The result of cluster analysis is to divide the die surface into several feature areas, each of which corresponds to the geometric expression of a certain type of feature. Boundary detection is performed on the preliminary feature areas. The boundary of each feature area is determined by identifying the transition area between areas, and the feature boundary data is obtained. By analyzing the boundary data, adjacent feature areas are merged, and similar or related areas are integrated into larger and more representative optimized feature areas. For example, in the blade area of the die, boundary detection can help identify the edge and edge of the blade, and merge these edge areas into an overall feature area. Shape descriptors are calculated for the optimized feature areas. Shape descriptor is a mathematical tool that quantitatively describes geometric features and can extract complex geometric shape information into a vector. Assume that a shape descriptor vector is represented as , where each Represents a certain shape attribute of the feature area, such as area, mean curvature, or boundary length. The shape descriptor vector is classified to classify each feature area into different geometric feature categories. The classification process can be completed by machine learning algorithms, such as support vector machines or k-nearest neighbor classifiers. Through classification, feature classification results are obtained. For example, in the blade area, the shape descriptor may show high curvature and sharp edges in the area, which can be used to judge the sharpness and cutting performance of the blade.
[0067] In a specific embodiment, the process of performing a three-dimensional convolutional neural network processing on the voxel representation data to obtain a local feature map may specifically include the following steps:
[0068] Input the voxel representation data into the input layer of the three-dimensional convolutional neural network for four-dimensional tensor format conversion to obtain network input data;
[0069] Perform the first convolution layer on the network input data, use 32 3x3x3 convolution kernels to perform three-dimensional convolution operation, with a step size of 1 and a padding of 1 to obtain the first layer feature map;
[0070] The first layer feature map is batch normalized and the ReLU activation function is applied to obtain the first layer activation feature map. The first layer activation feature map is subjected to the maximum pooling operation, using a 2x2x2 pooling kernel with a step size of 2 to obtain the first layer downsampled feature map.
[0071] The second convolution layer is used to process the first layer downsampled feature map. 64 3x3x3 convolution kernels are used for three-dimensional convolution operation with a step size of 1 and a padding of 1 to obtain the second layer feature map.
[0072] The second layer feature map is batch normalized and the ReLU activation function is applied to obtain the second layer activation feature map. The second layer activation feature map is subjected to the maximum pooling operation, using a 2x2x2 pooling kernel with a step size of 2 to obtain the second layer downsampled feature map.
[0073] The third convolution layer is used to process the second layer downsampled feature map. 128 3x3x3 convolution kernels are used for three-dimensional convolution operation with a step size of 1 and a padding of 1 to obtain the third layer feature map.
[0074] The third layer feature map is batch normalized and the ReLU activation function is applied to obtain the third layer activation feature map. The third layer activation feature map is subjected to global average pooling operation to compress the spatial dimension of the third layer activation feature map to 1x1x1 to obtain the local feature map.
[0075] Specifically, the voxel representation data is input into the input layer of the 3D convolutional neural network for 4D tensor format conversion. The original 3D voxel data is expanded into a 4D tensor format with multiple channels to match the input requirements of the convolutional neural network. Assume that the voxel data is of size A three-dimensional matrix, where , and Represents depth, height and width respectively. In order to adapt to the input layer of the neural network, the three-dimensional data is converted to The four-dimensional tensor format of is the batch size (i.e. the number of samples processed at a time), is the number of channels (can be set to 1 for single voxel data), and , and The first convolutional layer is processed on the converted network input data. 32 The core of the convolution operation is to perform weighted summation of the input data through a sliding window to extract the features of the local area.
[0076] Assume the input data is ,in Indicates convolution kernels, , and Represent the depth, height and width of the convolution kernel, and the output feature map after convolution It can be expressed as:
[0077] ;
[0078] in, It is The bias term of the convolution kernel, , and is the index of the output feature map after convolution. The stride of the convolution kernel is set to 1, and the padding is set to 1. The convolution operation calculates each position of the input data, and the padding operation keeps the spatial size of the output feature map the same as the input. The first layer of feature maps is batch normalized. By standardizing the values of the feature maps to have the same mean and variance, the statistical distribution differences of different batches of data are eliminated, and the stability and convergence speed of network training are improved. The batch-normalized feature maps are then nonlinearly activated using the ReLU activation function. The expression of the ReLU function is:
[0079] ;
[0080] This operation truncates all negative values to 0, thereby introducing nonlinear factors and enabling the network to better fit complex data patterns. The activated feature map retains the important features of the input data while suppressing irrelevant information. The first layer of activated feature maps is subjected to the maximum pooling operation. Maximum pooling selects the maximum value within the pooling window to represent the features of the area, thereby achieving dimensionality reduction and data compression of the feature map. Assume that the pooling window size is , the step size is 2, then the feature map after pooling It can be expressed as:
[0081] ;
[0082] The maximum pooling operation reduces the size of the feature map by half, and the first layer of downsampled feature maps is obtained. The second convolutional layer is processed on the first layer of downsampled feature maps. The convolution kernel of the first convolution layer performs a three-dimensional convolution operation. The steps are similar to the first convolution layer. The step size of the convolution kernel is still set to 1 and the padding is set to 1. After the second convolution operation, the second layer feature map is obtained. The second layer feature map is batch normalized and the ReLU activation function is applied to ensure that the model maintains the ability to express nonlinear features. After activation, the feature map is max-pooled using the same The pooling kernel and stride of are used to reduce the spatial size of the feature map and obtain the second-layer downsampled feature map. The second-layer downsampled feature map is processed by the third convolution layer. In this layer, 128 The third layer of convolution can capture more complex geometric features and higher-level abstract representations through a larger number of convolution kernels and a deeper network layer. The output third layer feature map contains key information in the three-dimensional structure of the laser die. After batch normalization and ReLU activation of the third layer feature map, the third layer activation feature map is obtained. Global average pooling is performed on the third layer activation feature map. Global average pooling compresses the spatial dimension of the feature map to , that is, the average value of all values of each feature map is calculated and used as the output of the feature map. Assume that the size of the activation feature map of the third layer is , then the output after global average pooling is It is expressed as:
[0083]
[0084] Through global average pooling, the network compresses spatial information into representative local feature maps, providing a highly integrated feature expression for the final output of the model.
[0085] In a specific embodiment, the process of executing step 300 may specifically include the following steps:
[0086] According to the feature classification results, the three-dimensional model of the laser cutting die is divided into regions and the initial grid size is allocated to obtain the initial grid size distribution map;
[0087] According to the blade parameters in the adjustable parameter set, the blade area is locally meshed to obtain a blade area mesh, and a multi-level mesh division framework is constructed based on the initial mesh size distribution map and the blade area mesh.
[0088] Adaptively refine the multi-level grid division framework to obtain a preliminary refined grid model, and locally reconstruct the grid of the preliminary refined grid model to obtain an optimized grid model;
[0089] The optimized mesh model is subjected to boundary layer mesh generation to obtain a mesh model with a boundary layer, and the mesh model with a boundary layer is subjected to transition zone mesh generation and node number optimization to obtain a multi-scale finite element analysis model.
[0090] Specifically, the three-dimensional model of the laser cutting die is divided into regions according to the feature classification results. The three-dimensional model of the cutting die is divided into multiple regions, each of which may have different physical properties and geometric complexity. For example, the cutting die body, the blade and the connection area are divided into different regions, and the meshing strategy of each region may be different to meet specific computing requirements. After the region division is completed, the initial grid size is allocated to each region to form an initial grid size distribution map. The allocation of the initial grid size is usually based on the geometric characteristics of the region and the key parameters contained in the feature classification results. For example, areas with higher geometric complexity may require a smaller grid size to capture subtle geometric details. According to the blade parameters in the adjustable parameter set, the blade area is locally meshed. As the most critical part of the cutting die, the accuracy of the blade directly affects the cutting performance of the cutting die, so more refined meshing is required in this area. For example, if the geometric parameters of the blade include the blade angle , Blade radius and blade length The degree of mesh refinement is determined by these parameters. When it is smaller, it may mean that the curvature of the area is larger and a denser mesh is required to capture subtle changes. The mesh refinement process is achieved by dividing the initial mesh into smaller units, which can more accurately describe the shape and boundary of the blade. The refined mesh in the blade area can significantly improve the analysis accuracy without increasing the overall computational workload. The refined mesh in the blade area is combined with the initial mesh size distribution map to construct a multi-level mesh partitioning framework. The multi-level mesh partitioning framework uses different mesh densities in different areas to achieve a multi-scale description of the entire model. Larger mesh units can be retained in simpler geometric areas, while finer meshes are used in areas with complex geometry or stress concentration, which maintains high accuracy while reducing the overall computational cost. Based on the multi-level mesh partitioning framework, adaptive refinement is performed to obtain a preliminary refined mesh model. Adaptive refinement is the process of automatically adjusting the mesh density based on the preliminary calculation results. It refines the mesh in these areas by identifying areas with large stress, deformation or temperature gradients in the model. For example, in the preliminary calculation of finite element analysis, the stress concentration phenomenon in a certain area may cause an increase in local errors. Adaptive refinement can dynamically adjust the mesh density in this area to improve the accuracy of the calculation. The mesh of the preliminary refined mesh model is locally reconstructed to eliminate possible mesh distortion or low-quality units and obtain the optimized mesh model. The mesh reconstruction process adjusts the shape and distribution of the mesh units to improve the quality indicators of the units, such as the shape coefficient and area ratio of the units. The optimized mesh model can better adapt to complex geometric shapes and avoid the problem of error accumulation in numerical analysis. For example, by adjusting the position of certain mesh nodes, the shape of adjacent units is made more regular, improving the quality of the overall mesh and the stability of the analysis. The boundary layer mesh is generated for the optimized mesh model. The boundary layer mesh is suitable for problems such as fluid dynamics or heat conduction. In the area close to the solid surface, the gradient of physical quantities such as velocity or temperature is large. The boundary layer mesh can accurately capture these gradient changes by generating a denser mesh layer in the area close to the surface. For example, in the thermal analysis of laser cutting dies, the temperature gradient near the blade surface is large, so it is necessary to generate a dense boundary layer mesh in this area to ensure the accuracy of the heat conduction analysis. The transition zone mesh generation and node number optimization are performed on the mesh model with boundary layer to obtain the final multi-scale finite element analysis model. The purpose of transition zone mesh generation is to achieve a smooth transition between the dense boundary layer mesh and the coarse main body mesh, thereby avoiding numerical errors caused by sudden changes in mesh size. Node number optimization is to improve the computational efficiency of finite element analysis by optimizing the numbering order of mesh nodes. Assume that the stiffness matrix of the finite element equation system is and the load vector for:
[0091] ;
[0092] in, is the node displacement vector, the node number is optimized by reducing the matrix The non-zero element distribution of the matrix can improve the speed and efficiency of matrix solution. The multi-scale finite element analysis model finally obtained can meet the requirements of accuracy and efficiency at the same time.
[0093] In a specific embodiment, the process of executing step 400 may specifically include the following steps:
[0094] Material properties are assigned to the multi-scale finite element analysis model. According to the material characteristics of different regions, the corresponding elastic modulus, Poisson's ratio, thermal conductivity and specific heat capacity parameters are assigned to each unit to obtain the material property distribution model.
[0095] According to the working conditions of the laser cutting die, loads and boundary conditions, including mechanical loads, thermal loads and fixed constraints, are applied to the material property distribution model to obtain the initial condition model;
[0096] The thermal-mechanical coupling control equation is constructed for the initial condition model, the heat conduction equation and the thermal stress equation are established to obtain the coupling equation group, and the coupling equation group is time discretized to obtain the discretized equation group;
[0097] The discretized equations are spatially discretized to obtain the global stiffness matrix and load vector, and the displacement field and temperature field of each time step are solved based on the global stiffness matrix and load vector;
[0098] Perform strain calculation on the displacement field to obtain strain distribution data, and calculate the stress field based on the strain distribution data and the material constitutive relationship to obtain stress distribution data;
[0099] The displacement field is post-processed to extract the node displacement information to obtain the deformation data, and the temperature field is interpolated to obtain the temperature field distribution data.
[0100] Specifically, according to the material properties of different regions, each unit is assigned corresponding physical parameters, including elastic modulus, Poisson's ratio, thermal conductivity and specific heat capacity. The material properties of different regions will directly affect the mechanical and thermodynamic behavior of the model. By assigning appropriate material parameters to each finite element unit in the model, a material property distribution model that reflects the actual material behavior is constructed. According to the working conditions of the laser die, loads and boundary conditions are applied to the material property distribution model. Loading conditions usually include mechanical loads, thermal loads and fixed constraints. For example, in actual use, the laser die may be subjected to cutting force as a mechanical load, and at the same time, it will be affected by thermal loads due to the heat generated during the cutting process. Fixed constraints are applied to certain key parts to simulate the fixed support of the die under actual working conditions. By applying these conditions, the initial condition model is obtained. The thermal-mechanical coupling control equation is constructed for the initial condition model. The core of thermal-mechanical coupling analysis is to consider the mutual influence of heat conduction and thermal stress at the same time. The heat conduction equation describes the diffusion behavior of temperature in the material, and its mathematical expression is:
[0101] ;
[0102] in, is the density, is the specific heat capacity, is the temperature, is the thermal conductivity, is the internal heat source term. This equation shows that temperature change is not only affected by the thermal properties of the material itself, but also by the combined effect of external heat sources and thermal conductivity. At the same time, the thermal stress equation describes the stress distribution caused by temperature change, and its basic form is:
[0103] ;
[0104] in, is the stress tensor, is the volume force density. The thermal stress equation combines the thermal expansion effect with the mechanical response of the material, reflecting the influence of thermal load on the internal stress state of the material. By coupling the heat conduction equation and the thermal stress equation, a complete set of coupled equations is obtained, which together describe the behavior of the material under thermal-mechanical coupling. In order to solve these coupled equations on a computer, they are time discretized. The continuous time domain is divided into discrete time steps, so that the temperature and stress state of each time step can be solved step by step. Commonly used methods include the forward Euler method or the implicit backward Euler method, the time step The choice of is crucial to the stability and accuracy of the solution. Through time discretization, a discretized set of equations is obtained, and the temperature and stress at each time step can be solved step by step. Based on time discretization, the discretized set of equations is spatially discretized. Spatial discretization is to expand the coupling equations on the entire finite element grid to obtain the global stiffness matrix and the load vector
[0105] ;
[0106] in, is the displacement vector, represents the stiffness matrix of the material, integrating the elastic modulus, Poisson's ratio, and geometric information; and The external loads applied to the model are included. By solving the linear equations, the displacement field and temperature field at each time step are obtained. The strain is calculated for the displacement field. The strain calculation is based on the gradient of the displacement field, assuming that the strain tensor is , then it is related to the displacement gradient The relationship is:
[0107] ;
[0108] The strain tensor describes the geometric characteristics of material deformation and is combined with the material's constitutive relations (such as Hooke's law for linear elastic materials) to calculate the stress field.
[0109] ;
[0110] in, It is the elastic matrix of the material, which reflects the elastic modulus and Poisson's ratio of the material and other parameters. Through this process, the stress distribution data inside the model is obtained, revealing the stress state of the material under the load. The displacement field is post-processed to extract the displacement information of the node and obtain the deformation data of the entire model. The deformation data shows the displacement of each part of the laser die under the combined action of external load and thermal effect. For example, if the deformation of the die exceeds the design allowable range, it may cause it to fail in actual use, thereby affecting the processing quality. The temperature field is interpolated to obtain more accurate temperature field distribution data. Interpolation processing can obtain continuous temperature distribution by interpolating data between discrete calculation nodes. For example, the thermal fatigue of the laser die during operation can be evaluated through the temperature field distribution data, and the design can be optimized accordingly to extend its service life.
[0111] In a specific embodiment, the process of executing step 500 may specifically include the following steps:
[0112] Perform data preprocessing on stress distribution data, deformation data and temperature field distribution data to obtain a standardized data set, and perform dimensionality reduction processing on the standardized data set to obtain a feature matrix after dimensionality reduction;
[0113] Perform variance analysis on the feature matrix after dimensionality reduction, calculate the contribution of each feature to the target, and obtain the feature importance ranking;
[0114] According to the feature importance ranking, the top K features with the highest contribution are selected to obtain the target influencing parameters, and multiple linear regression analysis is performed on the target influencing parameters to establish a linear relationship model between the factors and the target, and obtain a preliminary regression equation;
[0115] Perform residual analysis on the preliminary regression equation to obtain a revised regression equation, and based on the revised regression equation, construct the objective function of the multi-objective optimization problem, including minimizing the maximum stress, minimizing the maximum deformation, and minimizing the maximum temperature, and obtain the multi-objective function expression;
[0116] According to the mold design specifications and process limitations, the value range and constraint conditions of each design variable are determined to obtain a set of constraint equations, and the multi-objective function expression and the set of constraint equations are combined to obtain a comprehensive objective function;
[0117] A sensitivity analysis of the comprehensive objective function is performed to obtain a multi-objective optimization model.
[0118] Specifically, the stress distribution data, deformation data, and temperature field distribution data are standardized. The mean normalization or z-score normalization method is used to subtract the mean of each data feature and divide it by the standard deviation to eliminate the dimensional differences between different features and obtain a standardized data set. The standardized data set is subjected to dimensionality reduction processing to reduce the dimension of the data and extract the most representative features. Common dimensionality reduction techniques include principal component analysis and linear discriminant analysis. Through dimensionality reduction processing, high-dimensional data is converted into a low-dimensional feature matrix, thereby retaining most of the effective information of the data and eliminating redundancy. Variance analysis is performed on the feature matrix after dimensionality reduction to calculate the contribution of each feature to the optimization objective. The purpose of variance analysis is to quantify the role of each feature in explaining the changes in the target variable and to rank these features according to their contribution. By analyzing the variance contribution of each feature, it is possible to identify which features play a key role in the optimization process. Assume that a certain feature The variance contribution of , it can be calculated by the following formula:
[0119] ;
[0120] in, Representation characteristics The variance of is the total number of features. By calculating the variance contribution of each feature and sorting them, we get the feature importance ranking. According to the feature importance ranking, we select the top feature with the highest contribution. Features are selected to obtain the target influencing parameters. Multiple linear regression analysis is performed on the target influencing parameters to establish a linear relationship model between the factors and the optimization target. Multiple linear regression analysis describes the relationship between the target variable and the selected features by fitting a linear model. Assume that the target variable is , the eigenvector is (Include selected features), then the preliminary regression equation can be expressed as:
[0121] ;
[0122] in, is the intercept term, is the regression coefficient, is the error term. Estimated by the least squares method value, and obtain a preliminary regression equation, which reveals the linear relationship between the target variable and each feature. Perform residual analysis on the preliminary regression equation to evaluate the fitting effect and accuracy of the model. Residual analysis includes calculating the mean and variance of the residuals, and detecting the normality and independence of the residuals. If the residual analysis shows that there are systematic deviations or non-normal distributions, the regression model needs to be corrected. By adjusting the parameters in the regression model or introducing nonlinear terms, the corrected regression equation is obtained. Based on the corrected regression equation, the objective function of the multi-objective optimization problem is constructed. For the optimal design of laser knife molds, the objective function usually includes multiple objectives such as minimizing the maximum stress, minimizing the maximum deformation, and minimizing the maximum temperature. Assume that these objectives are expressed as , and , then the expression of the multi-objective function can be expressed as:
[0123] ;
[0124] in, A vector representing design variables, including all design parameters related to the optimization objective. The multi-objective function expression reflects the need to optimize multiple objectives in the design space. In order to ensure that the optimization results meet the actual design specifications and process limitations, the value range and constraints of each design variable are determined to obtain a set of constraint equations. For example, a design variable The following constraints may need to be met:
[0125] ;
[0126] in, and Variables The minimum and maximum values of the multi-objective function are obtained by combining the multi-objective function expression with these constraints. The comprehensive objective function not only takes into account the optimization goal, but also incorporates the actual design and process constraints to ensure that the optimization results are practical. A sensitivity analysis is performed on the comprehensive objective function to evaluate the impact of different design variables on the optimization results. The results of the sensitivity analysis can help identify which variables are most critical to the final design solution and how to adjust these variables to optimize the design effect. Finally, an accurate multi-objective optimization model is constructed to guide the design optimization process of laser cutting molds.
[0127] In a specific embodiment, the execution step performs variance analysis on the feature matrix after dimension reduction, calculates the contribution of each feature to the target, and obtains the feature importance ranking process, which may specifically include the following steps:
[0128] The feature matrix after dimension reduction is grouped, and the data is divided into stress group, deformation group and temperature group according to the feature type to obtain the grouped data set;
[0129] Descriptive statistical analysis was performed on the grouped data sets to obtain statistical feature descriptions, and normality tests were performed on each group of data based on the statistical feature descriptions to obtain data distribution characteristics;
[0130] Based on the data distribution characteristics, the data is transformed to obtain the transformed data set, and the variance homogeneity test is performed on the transformed data set to obtain the inter-group variance results;
[0131] According to the inter-group variance results, one-way ANOVA was used for groups with homogeneous variances, and Welch's ANOVA was used for groups with unequal variances to obtain the F statistic and p value for each group;
[0132] Perform multiple comparisons on the F statistic and p value to obtain the significance level of the difference between the groups, and calculate the effect size of each feature based on the significance level to obtain the explained variance proportion of each feature;
[0133] The explained variance ratio is normalized to obtain the relative contribution of each feature to the target, and the features are arranged in descending order according to the relative contribution to obtain the feature importance ranking.
[0134] Specifically, the feature matrix after dimension reduction is grouped according to the feature type. According to the physical meaning represented by the feature, it is divided into three groups: stress group, deformation group and temperature group, and the grouped data set is obtained. Descriptive statistical analysis is performed on the grouped data set to obtain the basic statistical characteristics of the data. Descriptive statistical analysis includes calculating the mean, standard deviation, median, skewness and kurtosis of each group of data to help identify the central tendency, degree of dispersion and distribution form of the data. According to the description of statistical characteristics, a normality test is performed on each group of data to determine whether the data obeys the normal distribution. Common normality test methods include Shapiro-Wilk test or Kolmogorov-Smirnov test. The results of the normality test will provide a basis for the subsequent selection of statistical analysis methods. If the data deviates from the normal distribution, further data transformation may be required. Based on the data distribution characteristics, the data is transformed accordingly to obtain a transformed data set that better meets the analysis requirements. Common data transformation methods include logarithmic transformation, square root transformation or Box-Cox transformation, which can help the data be closer to the normal distribution. For example, if the stress group data exhibits a right-skewed distribution, it can be brought closer to a normal distribution by performing a logarithmic transformation:
[0135] ;
[0136] in, is the transformed data. After data transformation, a new transformed data set is obtained. A variance homogeneity test is performed on the transformed data set. The variance homogeneity test is used to determine whether the variances of different groups of data are equal. Commonly used variance homogeneity test methods include Levene's test and Bartlett's test. If the test result shows that the variances are homogeneous, that is, the variances of each group of data are equal, then one-way analysis of variance (ANOVA) can be used to analyze the differences between the data groups; if variance heterogeneity exists, Welch's ANOVA should be used, which is more robust to the case of unequal variance. Through one-way ANOVA or Welch's ANOVA, calculate the F statistic and p-value of each group of data. The F statistic reflects the ratio of the variation between groups to the variation within the group, and the p-value is used to judge the significance of the statistical results. For example, the F statistic It can be calculated by the following formula:
[0137] ;
[0138] If the p-value is lower than the preset significance level (usually 0.05), it can be considered that there are significant differences between different groups. Multiple comparison analysis is performed based on the F statistic and p-value to explore which features have significant differences between different data groups. On the basis of multiple comparison analysis, the effect size of each feature is calculated to obtain the proportion of explained variance of each feature. The effect size is used to measure the contribution of the feature to the change of the target variable and is a key indicator for ranking the importance of features. For example, a feature The effect size It can be expressed as:
[0139] ;
[0140] The sum of squares between groups reflects the variation of the feature between different groups, while the total sum of squares reflects the variation of the feature in the overall data. The larger the effect size, the stronger the explanatory power of the feature for the target variable. The explained variance ratio is normalized to obtain the relative contribution of each feature to the target. Normalization converts the effect size of all features into a value between 0 and 1 proportionally to ensure that they are compared on the same scale. Relative contribution It can be expressed as:
[0141] ;
[0142] in, is the total number of features. All features are sorted in descending order according to their relative contribution to obtain the importance ranking of the features, thereby identifying the features that have the greatest impact on the target variable.
[0143] In a specific embodiment, the process of executing step 600 may specifically include the following steps:
[0144] Encode the multi-objective optimization model, convert the design variables into binary strings, obtain the initial population, and evaluate the fitness of the initial population, calculate the fitness value of each individual under each objective function, and obtain the fitness matrix;
[0145] According to the fitness matrix, the non-dominated sorting of the population is performed to obtain the non-dominated level of each individual, and the crowding degree of individuals with the same non-dominated level is calculated to obtain the crowding value;
[0146] The parent population is selected based on the non-dominated level and the crowding value, and the parent population is crossovered to generate offspring individuals to obtain the offspring population after the crossover;
[0147] Perform polynomial mutation operation on the offspring population after crossover, introduce random disturbance, and obtain the mutated offspring population;
[0148] The mutated offspring population is merged with the parent population to form a merged population, and the merged population is selected until the maximum number of iterations or convergence conditions are reached to obtain the Pareto optimal solution set;
[0149] A decision analysis is performed on the Pareto optimal solution set, the comprehensive score of each solution is calculated, and the solution with the highest comprehensive score is selected as the target optimization structural parameter of the laser cutting die.
[0150] Specifically, the design variables are encoded and converted into binary strings to represent the solution to the optimization problem. The encoding method helps to discretize the continuous design variables, which is suitable for the genetic algorithm solution process. For example, suppose there are three design variables , and , each variable can take values within a specific range. By quantifying continuous variables into binary form, for example, using an 8-bit binary string to represent each variable, each design solution is represented as a 24-bit binary string. This string is an individual in the genetic algorithm, and the initial population consists of multiple such binary individuals. The initial population is evaluated for fitness, and the performance of each individual under each objective function is calculated to judge its quality. Assume there are two objective functions and ,in is the design variable vector obtained by decoding the binary string. The fitness evaluation is performed by decoding the binary individuals into continuous variable values, and then substituting them into the objective function to calculate the fitness value of each individual under these objective functions. For example, if the binary string of the individual is decoded into The value is 3.75, The value is 2.5, and after substituting it into the objective function, we may get and . The fitness values of all individuals form a fitness matrix. According to the fitness matrix, the individuals in the population are non-dominated. Non-dominated sorting is a key step in dealing with multi-objective optimization in genetic algorithms, and individuals are divided into different levels according to their dominance relationships. One individual dominates another individual, which means that it is no worse than another individual in all objective functions and performs better on at least one objective. Through non-dominated sorting, individuals are divided into multiple levels, and individuals in the first level are non-dominated, that is, no individual in the population dominates them. The non-dominated level of each individual is represented by an integer. The smaller the integer, the better the individual. The crowding degree is calculated for individuals with the same non-dominated level, and the adjacent distance of each individual in the target space is measured. The crowding degree reflects the distribution density of individuals in the target space. For example, if an individual is in the objective function and The performances on are 10 and 5 respectively, and its two nearest neighbors are The values on are 8 and 12, The values on are 4 and 6, then the crowding degree can be determined by calculating the sum of these two intervals. The smaller the crowding degree of an individual, the more crowded it is in a dense area in the target space. The genetic algorithm tends to retain those individuals with higher crowding degrees during the selection process to maintain the diversity of the population. Based on the non-dominated level and crowding value, the population is selected to obtain the parent population. In the selection process, the genetic algorithm first selects those individuals with lower non-dominated levels (i.e., individuals with smaller level values). If there are multiple individuals in the same level, individuals with higher crowding degrees are preferred. This selection strategy ensures that the next generation population not only retains the better solution, but also maintains the diversity of the solution. After the selection is completed, the parent population is crossovered to generate offspring individuals. The crossover operation generates new offspring individuals by exchanging the binary string parts of two parent individuals. For example, two binary strings 10101011 and 01010100 can generate two new individuals 10100100 and 01011011 through crossover. The offspring population after crossover is subjected to polynomial mutation operation. Mutation operation is an important step in genetic algorithms. By introducing random perturbations, the population can jump out of the local optimal solution and explore the solution space more comprehensively. Polynomial mutation is usually achieved by randomly flipping certain bits in the binary string. For example, the individual 101010111 may become 10111011 after mutation. Small-scale random mutation can enhance the global search ability of the algorithm. The mutated offspring population is merged with the parent population to form a larger merged population. By selecting the merged population, individuals with poor performance are gradually eliminated, and those with high fitness and strong diversity are retained until the size of the population is restored to the size of the initial population. This process is repeated until the maximum number of iterations is reached or the convergence condition is met. Convergence conditions usually include that the change in the objective function is less than a certain threshold, or the fitness of the optimal solution in the population is no longer significantly improved. The solution obtained through this process constitutes the Pareto optimal solution set, which represents the best compromise between different objectives in multi-objective optimization. Perform decision analysis on the Pareto optimal solution set to select the final optimal solution. In decision analysis, a weight is assigned to each objective function, and the comprehensive score of each solution is calculated based on the weight. For example, assuming the objective function and The weights are and , the comprehensive score of a solution It can be expressed as:
[0151] ;
[0152] By calculating the comprehensive score of all solutions, the solution with the highest score is selected as the final optimization solution. In the design of laser cutting dies, this optimization process can help determine the geometric parameters, material selection and process conditions of the cutting die, so that it can achieve the best balance in multiple performance indicators such as stress, deformation and temperature control. For example, through this optimization process, the best size and material combination of the cutting die under different working conditions can be found under the design constraints, thereby extending its service life and improving cutting accuracy.
[0153] The above describes the optimization design method of the mold structure in the embodiment of the present application. The following describes the optimization design device 10 of the mold structure in the embodiment of the present application. Figure 2 In one embodiment of the present application, an optimization design device 10 for the mold structure includes:
[0154] A modeling module 11 is used to perform parameterized modeling on the three-dimensional model of the laser cutting die to obtain an adjustable parameter set;
[0155] The feature classification module 12 is used to extract and classify the geometric features of the laser cutting die on the three-dimensional model of the laser cutting die to obtain the feature classification result;
[0156] A meshing module 13 is used to perform adaptive multi-scale meshing on the three-dimensional model of the laser cutting die based on the adjustable parameter set and the feature classification result to obtain a multi-scale finite element analysis model;
[0157] A coupling analysis module 14 is used to perform multi-physical field coupling analysis on the multi-scale finite element analysis model to obtain stress distribution data, deformation data and temperature field distribution data;
[0158] A construction module 15 is used to perform variance analysis on stress distribution data, deformation data and temperature field distribution data to obtain target influencing parameters, and to construct a multi-objective optimization model according to the target influencing parameters;
[0159] The solution module 16 is used to solve the multi-objective optimization model by using a non-dominated sorting genetic algorithm to obtain a Pareto optimal solution set, and determine the target optimization structural parameters of the laser cutting die from the Pareto optimal solution set.
[0160] Through the cooperation of the above components, the adjustable parameter set and feature classification results are established by parametric modeling and geometric feature extraction of the laser cutting die, which provides a comprehensive and accurate data basis for subsequent optimization and is conducive to improving the pertinence and efficiency of optimization. Based on the feature classification results and adjustable parameter set, adaptive multi-scale meshing is realized, which can be refined in key areas while taking into account the computational efficiency, thus improving the accuracy and efficiency of finite element analysis. By constructing the thermal-mechanical coupling control equation, a comprehensive analysis of the laser cutting die under complex working conditions is realized, and stress distribution, deformation and temperature field data can be obtained simultaneously, providing a comprehensive performance evaluation for optimization. The variance analysis method is used to identify key influencing parameters and construct a multi-objective optimization model, which can effectively reduce the complexity of the optimization problem and improve the optimization efficiency while considering multiple performance indicators. Using the non-dominated sorting genetic algorithm to solve the multi-objective optimization problem, a series of Pareto optimal solutions can be obtained, providing decision makers with multiple trade-off solutions, and enhancing the practicality and flexibility of the optimization results. By comprehensively scoring and decision-making analysis of the Pareto optimal solution set, the best balance point can be found between multiple optimization objectives to ensure that the final selected optimized structural parameters meet the performance requirements.
[0161] The present application also provides an electronic device, which includes a memory and a processor, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the processor executes the steps of the mold structure optimization design method in the above-mentioned embodiments.
[0162] The present application also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein instructions are stored in the computer, and when the instructions are executed on a computer, the computer executes the steps of the mold structure optimization design method.
[0163] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.
[0164] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including a number of instructions to enable an electronic device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.
[0165] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.
Claims
1. A method for optimizing the design of mold structure, It is characterized in that The optimization design method of the mold structure includes: The three-dimensional model of the laser knife die is parametrically modeled to obtain an adjustable parameter set; specifically, the method includes: geometrically decomposing the three-dimensional model of the laser knife die to obtain a knife die body, a blade and a connection structure; parameterizing the size of the knife die body to obtain body thickness, length and width parameters, and parameterizing the shape of the blade to obtain blade angle, blade radius and blade length parameters, and parameterizing the layout of the connection structure to obtain connection hole diameter, position and quantity parameters; extracting and defining material parameters according to a knife die material property database to obtain elastic modulus, Poisson's ratio and density parameters, and analyzing the processing technology specifications of the knife die to obtain surface roughness and heat treatment parameters; quantifying environmental conditions according to the use environment specifications of the knife die to obtain working temperature range and humidity range parameters, and quantitatively analyzing the mechanical performance indicators of the knife die to obtain maximum allowable stress and maximum allowable deformation parameters; establishing a constraint relationship matrix between various parameters to obtain a parameter constraint equation group, and integrating and normalizing all parameters and parameter constraint equation groups to obtain an adjustable parameter set; The three-dimensional model of the laser knife die is subjected to knife die geometric feature extraction and feature classification to obtain feature classification results; specifically comprising: voxelizing the three-dimensional model of the laser knife die to obtain voxel representation data; performing three-dimensional convolutional neural network processing on the voxel representation data to obtain a local feature map; performing maximum pooling operation on the local feature map to obtain a feature map after dimension reduction, and performing nonlinear activation function operation on the feature map after dimension reduction to obtain an activated feature map; performing curvature analysis on the knife die surface according to the activated feature map to obtain curvature distribution data, and performing cluster analysis on the curvature distribution data to obtain a preliminary feature region; performing boundary detection on the preliminary feature region to obtain feature boundary data, and merging adjacent feature regions according to the feature boundary data to obtain an optimized feature region; performing shape descriptor calculation on the optimized feature region to obtain a feature description vector, and classifying the feature description vector to obtain a feature classification result; Based on the adjustable parameter set and the feature classification result, adaptively multi-scale meshing is performed on the three-dimensional model of the laser cutting die to obtain a multi-scale finite element analysis model; Performing multi-physical field coupling analysis on the multi-scale finite element analysis model to obtain stress distribution data, deformation data and temperature field distribution data; Performing variance analysis on the stress distribution data, the deformation data and the temperature field distribution data to obtain target influencing parameters, and constructing a multi-objective optimization model according to the target influencing parameters; The multi-objective optimization model is solved by a non-dominated sorting genetic algorithm to obtain a Pareto optimal solution set, and the target optimization structural parameters of the laser cutting die are determined from the Pareto optimal solution set.
2. The method for optimizing the mold structure according to claim 1, It is characterized in that The step of performing three-dimensional convolutional neural network processing on the voxel representation data to obtain a local feature map includes: Inputting the voxel representation data into the input layer of the three-dimensional convolutional neural network for four-dimensional tensor format conversion to obtain network input data; Performing a first convolutional layer processing on the network input data, using 32 3x3x3 convolution kernels to perform a three-dimensional convolution operation with a step size of 1 and a padding of 1 to obtain a first layer feature map; Perform batch normalization on the first layer feature map and apply the ReLU activation function to obtain the first layer activation feature map, and perform a maximum pooling operation on the first layer activation feature map, using a 2x2x2 pooling kernel with a step size of 2 to obtain the first layer downsampled feature map; Perform a second convolutional layer on the first layer of downsampled feature maps, use 64 3x3x3 convolution kernels to perform a three-dimensional convolution operation, with a step size of 1 and a padding of 1, to obtain a second layer of feature maps; Perform batch normalization on the second layer feature map and apply the ReLU activation function to obtain the second layer activation feature map, and perform a maximum pooling operation on the second layer activation feature map, using a 2x2x2 pooling kernel with a step size of 2 to obtain the second layer down-sampled feature map; Perform the third convolutional layer processing on the second layer downsampled feature map, use 128 3x3x3 convolution kernels to perform three-dimensional convolution operation, with a step size of 1 and a padding of 1 to obtain the third layer feature map; The third layer feature map is batch normalized and a ReLU activation function is applied to obtain a third layer activation feature map, and a global average pooling operation is performed on the third layer activation feature map to compress the spatial dimension of the third layer activation feature map to 1x1x1 to obtain a local feature map.
3. The optimization design method of the mold structure according to claim 1, It is characterized in that Based on the adjustable parameter set and the feature classification result, the three-dimensional model of the laser cutting die is adaptively divided into multi-scale grids to obtain a multi-scale finite element analysis model, including: According to the feature classification result, the three-dimensional model of the laser cutting die is divided into regions and the initial grid size is allocated to obtain an initial grid size distribution map; According to the blade parameters in the adjustable parameter set, the blade region is locally meshed to obtain a blade region mesh, and a multi-level mesh division framework is constructed based on the initial mesh size distribution map and the blade region mesh; Adaptively refining the multi-level grid division framework to obtain a preliminary refined grid model, and locally reconstructing the grid of the preliminary refined grid model to obtain an optimized grid model; Boundary layer mesh generation is performed on the optimized mesh model to obtain a mesh model with a boundary layer, and transition zone mesh generation and node number optimization are performed on the mesh model with a boundary layer to obtain a multi-scale finite element analysis model.
4. The method for optimizing the design of the mold structure according to claim 1, It is characterized in that The multi-scale finite element analysis model is subjected to multi-physical field coupling analysis to obtain stress distribution data, deformation data and temperature field distribution data, including: Allocating material properties to the multi-scale finite element analysis model, assigning corresponding elastic modulus, Poisson's ratio, thermal conductivity and specific heat capacity parameters to each unit according to the material characteristics of different regions, and obtaining a material property distribution model; According to the working conditions of the laser cutting die, loads and boundary conditions are applied to the material property distribution model, including mechanical loads, thermal loads and fixed constraints, to obtain an initial condition model; Constructing a thermal-mechanical coupling control equation for the initial condition model, establishing a heat conduction equation and a thermal stress equation, obtaining a coupled equation group, and performing time discretization processing on the coupled equation group to obtain a discretized equation group; Performing spatial discretization processing on the discretized equation group to obtain a global stiffness matrix and a load vector, and solving the displacement field and the temperature field at each time step according to the global stiffness matrix and the load vector; Performing strain calculation on the displacement field to obtain strain distribution data, and calculating the stress field based on the strain distribution data and the material constitutive relationship to obtain stress distribution data; The displacement field is post-processed to extract node displacement information to obtain deformation data, and the temperature field is interpolated to obtain temperature field distribution data.
5. The method for optimizing the design of the mold structure according to claim 1, It is characterized in that The performing variance analysis on the stress distribution data, the deformation data and the temperature field distribution data to obtain target influencing parameters, and constructing a multi-objective optimization model according to the target influencing parameters, comprises: Performing data preprocessing on the stress distribution data, the deformation data and the temperature field distribution data to obtain a standardized data set, and performing dimensionality reduction processing on the standardized data set to obtain a feature matrix after dimensionality reduction; Performing variance analysis on the feature matrix after dimensionality reduction, calculating the contribution of each feature to the target, and obtaining a feature importance ranking; According to the feature importance ranking, the top K features with the highest contribution are selected to obtain the target influencing parameters, and a multivariate linear regression analysis is performed on the target influencing parameters to establish a linear relationship model between the factors and the target to obtain a preliminary regression equation; Performing residual analysis on the preliminary regression equation to obtain a revised regression equation, and constructing an objective function of a multi-objective optimization problem based on the revised regression equation, including minimizing maximum stress, minimizing maximum deformation, and minimizing maximum temperature, to obtain a multi-objective function expression; According to the mold design specifications and process restrictions, the value range and constraint conditions of each design variable are determined to obtain a set of constraint equations, and the multi-objective function expression is combined with the set of constraint equations to obtain a comprehensive objective function; A sensitivity analysis is performed on the comprehensive objective function to obtain a multi-objective optimization model.
6. The method for optimizing the design of the mold structure according to claim 5, It is characterized in that The variance analysis is performed on the feature matrix after dimension reduction, the contribution of each feature to the target is calculated, and the feature importance ranking is obtained, including: Grouping the feature matrix after dimension reduction, dividing the data into stress group, deformation group and temperature group according to feature type, to obtain a grouped data set; Performing descriptive statistical analysis on the grouped data sets to obtain statistical feature descriptions, and performing a normality test on each group of data based on the statistical feature descriptions to obtain data distribution characteristics; Performing data transformation based on the data distribution characteristics to obtain a transformed data set, and performing a variance homogeneity test on the transformed data set to obtain an inter-group variance result; According to the inter-group variance results, one-way ANOVA was used for groups with homogeneous variances, and Welch's ANOVA was used for groups with unequal variances to obtain the F statistic and p value of each group; Perform multiple comparisons on the F statistics and p-values to obtain the significance level of the differences between the groups, and calculate the effect size of each feature based on the significance level to obtain the explained variance ratio of each feature; The explained variance ratio is normalized to obtain the relative contribution of each feature to the target, and the features are arranged in descending order according to the relative contribution to obtain the feature importance ranking.
7. The method for optimizing the design of the mold structure according to claim 1, It is characterized in that The method of performing non-dominated sorting genetic algorithm to solve the multi-objective optimization model to obtain a Pareto optimal solution set, and determining target optimization structural parameters of the laser cutting die from the Pareto optimal solution set includes: Encoding the multi-objective optimization model, converting the design variables into binary strings, obtaining an initial population, and performing fitness evaluation on the initial population, calculating the fitness value of each individual under each objective function, and obtaining a fitness matrix; Performing a non-dominated sorting of the population according to the fitness matrix to obtain a non-dominated level of each individual, and calculating the crowding degree of individuals with the same non-dominated level to obtain a crowding degree value; Selecting based on the non-dominated level and the crowding value to obtain a parent population, performing a crossover operation on the parent population to generate offspring individuals, and obtaining a crossover offspring population; Performing a polynomial mutation operation on the offspring population after the crossover, introducing random disturbance, and obtaining a mutated offspring population; The mutated offspring population is merged with the parent population to form a merged population, and the merged population is selected until a maximum number of iterations or a convergence condition is reached to obtain a Pareto optimal solution set; A decision analysis is performed on the Pareto optimal solution set, the comprehensive score of each solution is calculated, and the solution with the highest comprehensive score is selected as the target optimized structural parameter of the laser cutting die.
8. An optimization design device for mold structure, It is characterized in that Used to perform the optimization design method of the mold structure according to any one of claims 1 to 7, the optimization design device of the mold structure comprises: A modeling module is used to perform parameterized modeling on the three-dimensional model of the laser cutting die to obtain an adjustable parameter set; A feature classification module is used to extract and classify the geometric features of the laser cutting die on the three-dimensional model of the laser cutting die to obtain a feature classification result; A meshing module, used for performing adaptive multi-scale meshing on the three-dimensional model of the laser cutting die based on the adjustable parameter set and the feature classification result to obtain a multi-scale finite element analysis model; A coupling analysis module, used to perform multi-physical field coupling analysis on the multi-scale finite element analysis model to obtain stress distribution data, deformation data and temperature field distribution data; A construction module is used to perform variance analysis on the stress distribution data, the deformation data and the temperature field distribution data to obtain target influencing parameters, and to construct a multi-objective optimization model according to the target influencing parameters; The solution module is used to solve the multi-objective optimization model by non-dominated sorting genetic algorithm to obtain a Pareto optimal solution set, and determine the target optimization structural parameters of the laser cutting die from the Pareto optimal solution set.
Citation Information
Patent Citations
Profile mold optimization method and system based on multi-objective optimization algorithm
CN117556677A
Parameterized simulation model design method and system for precision mold
CN118430720A
