Extrusion process parameter optimization method and related device

Abstract

The invention discloses an extrusion process parameter optimization method and a related device, and relates to the technical field of metal forming process optimization and intelligent manufacturing, firstly, a deep neural network is adopted to process to-be-optimized process parameters to obtain an appearance prediction result of a tubular part, and a simulation result and the appearance prediction result of the tubular part are input into an interpolation model to obtain an interpolation model; the method comprises the following steps: firstly, interpolating a uniform regular grid to obtain an interpolated uniform regular grid, then, inputting the interpolated uniform regular grid into an intrinsic field quantity prediction model to obtain a field quantity prediction result, and finally, inputting to-be-optimized process parameters and the field quantity prediction result into a rapid simulation agent model to obtain an optimized process parameter set of the tubular part. According to the method, the optimization process parameters of the tubular part can be rapidly calculated through the rapid simulation agent model, and the efficiency of optimizing the extrusion process is greatly improved.

Description

Technical Field

[0001] The present application relates to the technical field of metal forming process optimization and intelligent manufacturing, and in particular to an extrusion process parameter optimization method and related devices. Background Art

[0002] The hot extrusion process is a manufacturing process that processes metal materials to above the recrystallization temperature and passes them through the orifice of the extrusion die by applying pressure to obtain long or hollow profiles with specific cross-sectional shapes and sizes. It is understandable that the hot extrusion process is a key technology for metal forming, and its product quality is significantly affected by process parameters. However, the traditional way to optimize the extrusion process is based on empiricism: trial and error. There are the following problems:

[0003] The efficiency of relying on empirical trial and error is low: Traditional optimization is mostly based on empirical parameter adjustment, and parameter combinations are verified through repeated mold trials, resulting in long mold trial cycles, high costs, but low efficiency. In addition, the simulation of hot extrusion process involves hot elastic-plastic deformation, as well as calculations of materials, geometry, contact nonlinearity, etc., which generally require reliance on process simulation software for calculation. However, the calculation and simulation time of accurate metal flow behavior is long, and the efficiency of process tuning is low, which makes it difficult to meet the needs of rapid optimization design of forging process parameters, especially when the number of parameters involved is very large and the range of parameter changes is very wide, the efficiency is even lower.

[0004] It can be seen that the efficiency of optimizing the extrusion process in the prior art is low. Summary of the invention

[0005] In view of the above problems, the present application provides an extrusion process parameter optimization method and related devices to achieve the purpose of improving the efficiency of the extrusion process. The specific scheme is as follows:

[0006] The first aspect of the present application provides a method for optimizing extrusion process parameters, comprising:

[0007] The process parameter set to be optimized of the tubular part is input into the deep neural network model to obtain the shape prediction result of the tubular part; the process parameter set to be optimized is: a set of key sensitive process parameters obtained after sampling the simulation results of the tubular part extrusion process and performing parameter sensitivity analysis;

[0008] The simulation results and the shape prediction results are input into the interpolation model to obtain the interpolated uniform regular grid; the interpolation model is used to interpolate the field quantity data of different unstructured grids with different process parameters to the uniform regular grid to be interpolated to obtain the interpolated uniform regular grid; the different unstructured grids with different process parameters are obtained by discretizing the simulation results;

[0009] The interpolated uniform regular grid is input into the intrinsic field quantity prediction model to obtain the field quantity prediction result;

[0010] The process parameter set to be optimized and the field quantity prediction results are input into the fast simulation agent model to obtain the optimized process parameter set of the tubular part; the fast simulation agent model is used to process the process parameters to be optimized in the process parameter set to be optimized through the nonlinear mapping relationship between the process parameters and the objective function to obtain the optimized process parameter set.

[0011] In a possible implementation, the shape prediction result includes the displacement result of the punch end of the blank and the displacement result of the free end of the tubular part; the field quantity data of different unstructured grids with different process parameters are correspondingly interpolated to the uniform regular grid to be interpolated, and the uniform regular grid after interpolation is obtained, including:

[0012] The displacement results of the punch end and the free end are used to update the grid coordinates of the uniform regular grid to be interpolated;

[0013] The field quantity data of each grid of different unstructured grids with different process parameters are correspondingly interpolated to each grid of the uniform regular grid to be interpolated.

[0014] In a possible implementation, the intrinsic field quantity prediction model is a prediction model that integrates the intrinsic orthogonal decomposition model and the modal participation coefficient solution model. The interpolated uniform regular grid is input into the intrinsic field quantity prediction model to obtain the field quantity prediction results, including:

[0015] The intrinsic orthogonal decomposition model is used to process the field quantity data in the interpolated uniform regular grid into low-order linear modal field quantity data;

[0016] The modal participation coefficient solution model is used to calculate the modal participation coefficient required for the field quantity data prediction of the intrinsic orthogonal decomposition model;

[0017] The intrinsic orthogonal decomposition model uses the modal participation coefficient calculated by the modal participation coefficient solution model to solve the modal participation coefficient and obtain the field quantity prediction result.

[0018] In one possible implementation, the modal participation coefficient solution model includes a back-propagation neural network model;

[0019] The back propagation neural network model is used to update the modal participation coefficients in the proper orthogonal decomposition model and is also used to predict the scalar data of the tubular component.

[0020] In a possible implementation, the fast simulation proxy model is a nonlinear proxy model between process parameters and objective functions constructed based on a Kriging model;

[0021] The process parameter set to be optimized and the field quantity prediction results are input into the fast simulation agent model to obtain the optimized process parameter set of the tubular part, including:

[0022] The field quantity prediction result is input into the Kriging model to obtain a nonlinear mapping relationship between the process parameters to be optimized and the objective function in the process parameter set to be optimized;

[0023] The objective function value of the process parameter to be optimized in the process parameter set to be optimized is calculated in combination with the expected improvement plus point criterion, and the objective function value is the optimized process parameter.

[0024] In a possible implementation, before inputting the process parameter set to be optimized of the tubular part into the deep neural network model, the method further includes:

[0025] A 2D axisymmetric equivalent model of the tubular component is established using numerical simulation tools;

[0026] Sampling the input process parameters of the 2D axisymmetric equivalent model establishment process to obtain a process parameter set;

[0027] Parameter sensitivity analysis is performed on each process parameter in the process parameter set to obtain the process parameter set to be optimized.

[0028] A second aspect of the present application provides an extrusion process parameter optimization device, comprising:

[0029] The shape prediction unit is used to input the process parameter set to be optimized of the tubular part into the deep neural network model to obtain the shape prediction result of the tubular part; the process parameter set to be optimized is: a set of key sensitive process parameters obtained after sampling the simulation results of the tubular part extrusion process and performing parameter sensitivity analysis;

[0030] The interpolation unit is used to input the simulation results and the shape prediction results into the interpolation model to obtain the interpolated uniform regular grid; the interpolation model is used to interpolate the field quantity data of different unstructured grids with different process parameters to the uniform regular grid to be interpolated to obtain the interpolated uniform regular grid; the different unstructured grids with different process parameters are obtained by discretizing the simulation results;

[0031] A field quantity prediction unit is used to input the interpolated uniform regular grid into the intrinsic field quantity prediction model to obtain the field quantity prediction result;

[0032] The process parameter optimization unit is used to input the process parameter set to be optimized and the field quantity prediction results into the fast simulation agent model to obtain the optimized process parameter set of the tubular part; the fast simulation agent model is used to process the process parameters to be optimized in the process parameter set to be optimized through the nonlinear mapping relationship between the process parameters and the objective function to obtain the optimized process parameter set.

[0033] The third aspect of the present application provides an extrusion process parameter optimization device, comprising at least one processor and a memory connected to the processor, wherein:

[0034] The memory is used to store computer programs;

[0035] The processor is used to execute the computer program so as to enable the extrusion process parameter optimization device to implement any of the extrusion process parameter optimization methods described above.

[0036] A fourth aspect of the present application provides a computer program product, comprising computer-readable instructions. When the computer-readable instructions are executed on an electronic device, the electronic device implements any of the extrusion process parameter optimization methods described above.

[0037] A fifth aspect of the present application provides a computer storage medium, which carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement any of the extrusion process parameter optimization methods described above.

[0038] By means of the above technical scheme, the extrusion process parameter optimization method and related device provided by the present application firstly use a deep neural network to process the process parameters to be optimized, obtain the shape prediction result of the tubular part, input the simulation result and the shape prediction result of the tubular part into the interpolation model, obtain the uniform regular grid after interpolation, then input the uniform regular grid after interpolation into the intrinsic field quantity prediction model, obtain the field quantity prediction result, and finally input the process parameters to be optimized and the field quantity prediction result into the fast simulation proxy model to obtain the optimized process parameter set of the tubular part. The present application can quickly calculate the optimized process parameters of the tubular part through the fast simulation proxy model, which greatly improves the efficiency of optimizing the extrusion process. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the accompanying drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and the originals and elements are not necessarily drawn to scale.

[0040] Figure 1 A schematic diagram of a process flow of the extrusion process parameter optimization method provided in this application;

[0041] Figure 2 An example diagram of the tubular member extrusion process provided for this application;

[0042] Figure 3 An example diagram of displacement of a tubular part extrusion process provided in this application;

[0043] Figure 4An example diagram of the structure of the deep neural network model provided for this application;

[0044] Figure 5 An example diagram of the interpolation effect provided for this application;

[0045] Figure 6 A schematic diagram of the structure of an extrusion process parameter optimization device provided in this application;

[0046] Figure 7 This is a schematic diagram of the structure of the extrusion process parameter optimization equipment provided in this application. DETAILED DESCRIPTION

[0047] The following describes the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. The terms used in the implementation method section of the present application are only used to explain the specific embodiments of the present application, and are not intended to limit the present application.

[0048] The embodiments of the present application are described below in conjunction with the accompanying drawings. Those skilled in the art will appreciate that, with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0049] The terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and need not be used to describe a specific order or sequential order. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, which is merely a way of distinguishing the objects of the same attributes when describing them in the embodiments of the present application. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, so that the process, method, system, product or equipment comprising a series of units need not be limited to those units, but may include other units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0050] In the prior art, the traditional experimental scheme is generally used to optimize the extrusion process parameters. This method has the following problems:

[0051] High time and cost: Taking the aluminum extrusion process as an example, it is necessary to establish a relationship model between process parameters and forming quality through multiple rounds of trial and error. A single test takes several hours to several days, and the material loss is as high as 10%-15%. The length of the billet is limited during extrusion, resulting in a low yield rate.

[0052] Limited prediction ability: Taking nickel-based high-temperature alloy extrusion as an example, the influence of extrusion and die angle on the temperature field needs to be fitted through a large amount of experimental data, but it is impossible to accurately predict the behavior of dynamic recrystallization. Defects such as shrinkage holes and dead zone folding are prone to occur during the extrusion process. It can be seen that traditional experimental methods are difficult to quantify and model.

[0053] Environmental interference is significant: Taking the extrusion of 5A06 aluminum alloy ribbed box as an example, mold temperature fluctuations will lead to uneven filling of the bottom ribs, which requires repeated debugging to optimize.

[0054] In addition, there are many problems with the optimization solutions based on simulation in the prior art:

[0055] Taking the simulation solution based on commercial software as an example, the main problems are as follows:

[0056] High consumption of computing resources: The simulation of cladding extrusion of nickel-based powder high-temperature alloys requires the division of millions of grid units. A single full-coupled field simulation can take several hours and requires reliance on high-performance computing clusters. The simulation of aluminum profile extrusion requires repeated adjustment of structural parameters such as diverter holes and feed troughs, extending the development cycle by 3-5 times.

[0057] Parameter boundary dependence experience: The optimization algorithms built into commercial software, such as the gradient descent method, require manual preset of the parameter search space such as extrusion speed and die gap, which is easy to fall into the local optimal solution. For example, too high an extrusion speed will lead to coarse grains, but traditional methods have difficulty in automatically identifying the critical threshold.

[0058] Inefficient iterative process: In the optimization of cold extrusion process parameters, orthogonal experiments require the design of 9 sets of solutions, each of which requires multiple simulation verifications, which takes several weeks. The difference in metal flow speed in the extrusion barrel will cause the coarse grain area in the shrinkage sensitive area to be incorrectly merged, affecting the effect of process parameter adjustment.

[0059] Another optimization method for extrusion process parameters used in the prior art is the forging optimization area segmentation method, which has the following problems:

[0060] The segmentation method based on mass / volume equivalence has the following problems:

[0061] Lack of field-volume correlation: The temporal and spatial distribution characteristics of the material flow velocity field and stress field are not considered. For example, the difference in metal flow velocity in the extrusion barrel will cause the shrinkage sensitive area and the coarse grain to be incorrectly merged, affecting the effect of process parameter adjustment.

[0062] Insufficient dynamic adaptability: Fixed segmentation areas are difficult to match the transient characteristics of extrusion deformation. For example, in a multi-stage extrusion process, the stress distribution of the previous process will significantly change the optimization target area of ​​the subsequent process, and traditional methods cannot be adjusted dynamically.

[0063] Low adjustment accuracy of process parameters: Traditional methods can only allow process parameters to fluctuate 5% up or down based on the baseline value. However, field-quantity coupling analysis shows that a 0.1mm fine-tuning of key parameters can lead to significant changes in forming quality.

[0064] In summary, there are many problems in the prior art for optimizing extrusion process parameters. Traditional experimental schemes are difficult to cope with complex coupling scenarios, simulation schemes are limited by computing resources and experience, and the optimization region segmentation method does not integrate the temporal and spatial characteristics of field quantities, resulting in defects such as shrinkage, folding, and uneven grains. The common problem in various process parameter optimization methods in the prior art is low efficiency.

[0065] In order to solve the above problems, the present application provides an extrusion process parameter optimization method and related devices.

[0066] Optional, see Figure 1 , a flow chart of the extrusion process parameter optimization method provided in this application.

[0067] like Figure 1 As shown, the extrusion process parameter optimization method includes the following steps:

[0068] Step 101: Input the set of process parameters to be optimized for the tubular part into a deep neural network model to obtain a shape prediction result of the tubular part.

[0069] The core focus of the hot extrusion process parameters of tubular parts lies in the precise control of the material rheological law and the multi-physical field coupling mechanism. The process optimization needs to comprehensively consider the influence of the following key parameters:

[0070] Material mechanical parameters: including the elastic modulus, yield strength, tensile strength, etc. of the material, will directly affect the deformation resistance and flow stress. For example, the yield strength of 7075 aluminum alloy is 465-500MPa, and the tensile strength is 500-600MPa. Its mechanical properties change significantly with deformation temperature and extrusion ratio.

[0071] Fracture criterion parameters: used to predict the fracture behavior of materials during deformation, such as maximum shear stress theory, dynamic recrystallization model, etc. For example, when GH3625 alloy tubes are extruded, the Laves phase melting critical temperature is greater than 1150°C, and the crack growth threshold is directly related to the stress concentration factor.

[0072] Constitutive relationship parameters: mathematical models that describe the stress-strain-temperature coupling behavior of materials, such as the Arrhenius equation. For example, in the constitutive equation of ZK60 magnesium alloy, the stress exponent is 9.45 and the activation energy is 120KJ / mol, and the parameters are determined by hot compression tests.

[0073] Temperature and boundary condition parameters: including billet temperature, mold preheating temperature, lubrication conditions, etc., affect heat conduction and friction. For example, when TA2 titanium alloy is extruded, the temperature difference needs to be controlled below 904.6 degrees Celsius when the phase change temperature is below 904.6 degrees Celsius, and the mold temperature of 300-600 degrees Celsius can reduce the extrusion pressure.

[0074] Punch stroke parameters: determine the metal filling capacity and deformation degree, and are strongly related to the extrusion ratio. For example, in two-stage extrusion, the punch stroke distribution needs to balance the preceding filling and the subsequent stress distribution, and the shrinkage suppression efficiency can be increased by 40%.

[0075] The above parameters will jointly affect the forming quality through multi-field coupling mechanisms, such as temperature field-velocity field-stress field. For example, when TA2 alloy is extruded at 870℃, the combination of temperature gradient <15℃ / mm, mold temperature 550℃, and extrusion speed 10mm / s can make the grain size 35-55μm and the tensile strength reach 520MPa.

[0076] The parameters mentioned above are the process parameters to be optimized. It can be understood that the various process parameters to be optimized included in the process parameter set to be optimized include but are not limited to the various parameters mentioned above.

[0077] Next, the process of obtaining the process parameter set to be optimized is explained.

[0078] Step 1: Use numerical simulation tools to establish a 2D axisymmetric equivalent model of the tubular component.

[0079] This process is mainly a simulation process of the hot extrusion process of the tubular part, which can be specifically a CAE (Computer Aided Engineering) simulation, to establish a 2D axisymmetric equivalent model of the hot extrusion process of the tubular part and obtain a CAE result file.

[0080] The reasons for establishing a 2D axisymmetric equivalent model of the hot extrusion process of tubular parts during simulation are as follows:

[0081] Plastic forming is a nonlinear process, and the simulation requires a lot of calculations and takes a long time. Considering that most forging parts have axisymmetric characteristics, the 2D axisymmetric model can accurately reflect the geometric characteristics of forging parts, reduce the number of calculation nodes and units, and thus reduce the demand for computing resources and simulation time. It is suitable for rapid iteration, preliminary design, and application environments that require the generation of a large number of simulation databases.

[0082] Among them, numerical simulation tools can also be called finite element analysis software, which can be Deform or self-developed solver.

[0083] Specifically, Deform or a self-developed solver is used to establish a 2D axisymmetric model simulation model of the tubular part, and the hot extrusion process of the tubular part is simulated. The size change of the tubular part in the hot extrusion process, the automatic drawing network and the automatic batch calculation of the solver calculation are realized through Python scripts to obtain the CAE result file.

[0084] Step 2: Sampling the input process parameters of the 2D axisymmetric equivalent model establishment process to obtain a process parameter set.

[0085] In order to obtain the original data or data set required for training the proxy model for optimizing the extrusion process parameters in the present application, it is necessary to sample the 2D axisymmetric model simulation model of the tubular hot extrusion process.

[0086] The sampling methods include but are not limited to the Sobol sequence method and LHS (Latin Hypercube Sampling).

[0087] Specifically, based on the actual design and production requirements of tubular parts, we first parametrically model key parameters such as blank size and boundary conditions, and scientifically allocate sampling intervals. Then, the Sobol sequence method or LHS method is used to efficiently sample the simulation input parameters to ensure that the sample space coverage is complete and evenly distributed, thereby improving the statistical reliability of the simulation results and the accuracy of sensitivity analysis, and obtaining a set of process parameters after sampling.

[0088] The simulation input parameters may specifically include punch speed, initial temperature field, billet size, extrusion ratio, distance between die core rod and billet, etc. The simulation output results include temperature field, strain field, grain size and other data.

[0089] Step 3: Perform parameter sensitivity analysis on each process parameter in the process parameter set to obtain the process parameter set to be optimized.

[0090] Next, it is necessary to conduct DOE (Design of Experiments) analysis on each process parameter in the process parameter set. Specifically, it is necessary to conduct parameter sensitivity analysis on each process parameter in the process parameter set, analyze the sensitivity between input and output, so as to determine which parameters in the process parameter set are key influencing parameters, and take the key influencing parameters in the process parameter set as the process parameter set to be optimized.

[0091] Specifically, DOE analysis is quickly performed on the process parameter set through the cloud platform, and CAE result files are batch processed using scripts. The data of physical fields such as equivalent strain and stress of all nodes in all time steps are processed and analyzed to identify sensitive parameters, and select key parameters to form the process parameter set to be optimized.

[0092] In summary, the process parameter set to be optimized is: a set of key sensitive process parameters obtained after sampling the simulation results of the tubular part extrusion process and performing parameter sensitivity analysis.

[0093] After obtaining the process parameter set to be optimized, it is input into a pre-trained deep neural network model (DNN) to obtain the shape prediction result of the tubular part.

[0094] It should be noted that the purpose of using the DNN proxy model in this application is to predict the final shape parameters under different process parameters, mainly including the punch end displacement results and free end displacement results of the tubular part.

[0095] For example, see Figure 2 , an example diagram of the tubular part extrusion process provided in this application.

[0096] After the metal billet is heated to a suitable temperature, it is placed in an extrusion barrel, which can be Figure 2 The moving mold shown in the figure is located in the space, and the moving mold pushes the blank into the mold cavity through the coordinated movement of the extrusion shaft and the extrusion mandrel.

[0097] The extrusion shaft applies pressure to push the blank through the mold cavity. The punch end is the end part of the extrusion shaft that applies pressure, directly pushing the blank into the mold cavity; the free end is the end part of the core rod away from the extrusion shaft, which can move freely or separate after extrusion is completed. The relative movement between the punch end and the free end will directly affect the metal flow direction and molding quality.

[0098] Correspondingly, see Figure 3 , an example diagram of displacement of a tubular part extrusion process provided in this application.

[0099] like Figure 3 As shown, the displacement change of the tubular part before and after extrusion mainly shows the displacement of the punch end of the blank and the free end of the tubular part.

[0100] The DNN proxy model used in this application adopts a fully connected structure and may include only one hidden layer in the middle.

[0101] For example, see Figure 4 , a structural example diagram of the deep neural network model provided in this application.

[0102] like Figure 4 As shown, the deep neural network model used in the present application can be a deep neural network proxy model including an input layer, a hidden layer and an output layer.

[0103] Optionally, the process parameter set to be optimized is input into a deep neural network model to obtain the displacement of the punch end of the blank and the displacement of the free end of the tubular part.

[0104] Among them, the displacement of the punch end of the blank and the displacement of the free end of the tubular part are essentially the coordinate values ​​of the space points, which can be specifically replaced by the grid under the steady-state results in the entire extrusion deformation area of ​​the tubular part and the maximum y coordinate value and the minimum y coordinate value of the data.

[0105] It is understandable that the shape prediction results output by the deep neural network model are essentially the x, y coordinates of spatial points, which can be specifically expressed as: (x1, y1), (x2, y2), (x3, y3), (x4, y4), (x5, y5) and so on spatial coordinate points.

[0106] It should also be noted that when using a deep neural network model for shape prediction, the line boundaries can be extracted from the 2D grid data, and the line boundaries are discretized by describing the control point parameters using NURBS (Non-Uniform Rational B-Spline) or B-spline curves. The control point parameters are introduced into the above-mentioned punch end displacement and free end displacement output parameters for DNN model prediction.

[0107] Step 102: Input the simulation results and the shape prediction results into the interpolation model to obtain a uniform regular grid after interpolation.

[0108] The simulation result is the simulation result of the tubular part extrusion process mentioned above, which can be a CAE result file. The shape of the tubular part will change during extrusion. In this application, a deep neural network model is mainly used to predict the shape prediction result of the tubular part after extrusion. When the shape of the tubular part changes during the extrusion process, the field quantity data of the space where the entire tubular part area is located is also changing, such as displacement value, temperature value, etc.

[0109] Specifically, for finite element simulation in the structural field, that is, the CAE simulation mentioned above, an unstructured network is generally used to discretize the entire physical domain, and the physical space where the tubular parts are located is discretized into small unstructured grids. The mesh distortion of the tubular parts under extrusion simulation is very large. The original unstructured grid needs to be re-divided due to the large deformation during the forging process. This leads to inconsistent basic grids of the final simulation results under different input parameters, and the grid points and numbers are inconsistent. Specifically, the original grid before distortion and the grid after distortion are not the same.

[0110] It is understandable that in the original simulation data, the unstructured field quantity results under different process parameters correspond to different unstructured grids, which leads to the inconsistency of the dimensions of the field quantity results under different process parameters.

[0111] The solution proposed in this application is to use a uniform structured grid to unify the dimensions of the field quantity results under different process parameters. Specifically, the field quantity data of different unstructured grids with different process parameters are interpolated into the uniform grid to be interpolated to obtain the interpolated uniform grid.

[0112] Specifically, the unstructured grid is interpolated onto the structured grid, specifically, the field quantity data on the unstructured grid points are interpolated to other positions in the space, and the other positions in the space are predetermined relatively regular uniform structured grids, that is, the uniform regular grid to be interpolated.

[0113] The interpolation process can be implemented based on the interpolation model. The specific operation process of the interpolation model can be:

[0114] Firstly, based on the displacement of the punch end of the blank and the displacement of the free end of the tubular part predicted by the deep neural network model mentioned above, the grid coordinates of the interpolated uniform regular grid are updated.

[0115] Then, the field quantity data of each grid point of different unstructured grids with different process parameters in the simulation results of the tubular parts are interpolated to the corresponding grids in the uniform regular grid to be interpolated. The initial unstructured temperature field and displacement structure are interpolated to the corresponding grids. The initial unstructured temperature field and displacement structure mentioned here are obtained by simulation of the 2D axisymmetric equivalent model, which is the simulation result of the tubular parts extrusion process mentioned above.

[0116] It should also be noted that interpolation can specifically be performed by using regional interpolation to improve the overall interpolation accuracy.

[0117] For example, see 5, an example diagram of the interpolation effect provided in this application.

[0118] like Figure 5 As shown, the left side shows an unstructured grid, and the right side shows the interpolation effect of a structured grid. The grid topology and number of points of the structured grid are fixed, and the coordinates of its network nodes change with the change of shape.

[0119] The interpolated structured grid is obtained by interpolating the variable values ​​on the original grid to a new uniform and regular grid.

[0120] It should be noted that the interpolated uniform grid is the basis for the intrinsic field quantity prediction model to predict the field quantity. Specifically, the basis for the intrinsic field quantity prediction model to predict the field quantity is the node value on the interpolated uniform grid.

[0121] Step 103: input the interpolated uniform regular grid into the intrinsic field quantity prediction model to obtain the field quantity prediction result.

[0122] The intrinsic field quantity prediction model is mainly used to predict the field quantity data of the tubular parts after extrusion molding, which may specifically include the displacement value and temperature value at each grid point.

[0123] The intrinsic field quantity prediction model provided in this application is a prediction model that integrates the intrinsic orthogonal decomposition model and the modal participation coefficient solution model. Among them, the intrinsic orthogonal decomposition model is a POD (Print Orthogonal Decomposition) model, which is a physical field proxy model established using the POD method. POD is an intrinsic decomposition calculated using SVD (Singular Value Decomposition) to obtain eigenvalues ​​and eigenvectors. Then, the field quantity results in the physical space are reduced to the modal space, which is a linearization process; the essence is to extract the main features in the high-dimensional physical field data, establish a low-dimensional model to approximate the original system, and achieve order reduction by retaining the main mode and ignoring the secondary mode, and achieve the purpose of reducing the degree of freedom through linear modal reduction.

[0124] The calculation process of the intrinsic orthogonal decomposition model can be as follows:

[0125] The snapshot matrix S is obtained based on the sampling points in the uniform regular grid after interpolation:

[0126] ;

[0127] Among them, N is the number of sampling points, p is the dimension of the field quantity, u (x i ) is the field result of the i-th sampling point, which can be temperature field or strain field, etc.;

[0128] Then, the snapshot matrix is ​​decomposed by SVD, and the calculation formula can be as follows:

[0129] ;

[0130] Among them, U is the left singular vector matrix, is the singular value matrix, is the transpose of the right singular vector matrix.

[0131] Take the first few columns of U to form a POD basis:

[0132] ;

[0133] in, For the POD base, is the ith POD mode, M is the order of the modal space, and M can be determined according to a tolerance of 0.99 and the minimum value in p, which needs to be determined based on the efficiency required by the user.

[0134] The formula of the intrinsic orthogonal decomposition model provided in this application can be specifically:

[0135] Where x is the spatial position of the field quantity data, is the field quantity data, M is the order of the modal space, is the predicted value of the field quantity data, is the average value of the field quantity data, is the scalar coefficient for each mode, also known as the modal participation coefficient, is the spatial mode basis function.

[0136] It can be seen from the formula of the eigenorthogonal decomposition model that the predicted value of the field quantity data can also be calculated by this formula. The unknown quantity in this formula is mainly the modal participation coefficient. Therefore, it is necessary to use the modal participation coefficient solution model to calculate the modal participation coefficient.

[0137] The modal participation coefficient solution model provided in this application can be a BPNN (Back Propagation Neural Network) model, which has strong nonlinear mapping capabilities and high-intensity self-learning and self-adaptive capabilities. The input layer is the process parameters to be optimized, the output layer is the modal participation coefficient, and the Loss function is the MRE deviation value of the temperature field and displacement field results. The calculation formula is as follows:

[0138]

[0139] Where N is the number of samples, is the predicted value of the ith mode participation coefficient, is the actual value of the ith mode participation coefficient, is a very small positive number, The maximum absolute value between the predicted value of the i-th modal participation coefficient and the actual value of the i-th modal participation coefficient is taken.

[0140] In addition, the BPNN model can also be used to predict scalars, such as grain size, grain uniformity index, etc.

[0141] The training steps of the BPNN model can be as follows:

[0142] Step 1: Initialize the weights. The weights are generally given randomly by setting the weight range.

[0143] Step 2: Calculate the output values ​​of the hidden layer and the output layer.

[0144] Step 3: Calculate the reverse error of each layer.

[0145] Step 4: Determine whether the training requirements are met. If so, the model training is completed and the model is output; if not, update the weights and repeat steps 2 and 3.

[0146] In short, the modal participation coefficient in the POD model is updated according to the output of BPNN, so as to calculate the field quantity prediction results.

[0147] The formula for the field quantity prediction result can be expressed as:

[0148]

[0149] in, is the predicted value of the field quantity data, is the average value of the field quantity data, M is the order of the modal space, is the predicted value of the scalar coefficient for each mode, is the spatial mode basis function.

[0150] Optionally, an intrinsic orthogonal decomposition model is used to process the field quantity data in the interpolated uniform regular grid into low-order linear modal field quantity data, and a modal participation coefficient solution model is used to calculate the modal parameter coefficients required for the intrinsic orthogonal decomposition model to predict the field quantity data. The intrinsic orthogonal decomposition model calls the modal parameter coefficients calculated by the modal participation coefficient solution model to calculate the field quantity prediction results, thereby obtaining the field quantity prediction results.

[0151] Step 104: input the process parameter set to be optimized and the field quantity prediction result into the fast simulation agent model to obtain the optimized process parameter set of the tubular part.

[0152] It should be noted that the fast simulation proxy model is a nonlinear mapping proxy model between process parameters and objective functions. The model constructs the nonlinear relationship between process parameters and objective functions based on the Kriging model, and constructs the liberalization problem through the expected improvement (EI) plus point criterion.

[0153] The specific calculation process can be as follows:

[0154] When it is known that the minimum value of the objective function in the current sample is , for the samples to be updated , whose fast simulation proxy model estimates ,and , the expectation of objective function improvement can be expressed as follows:

[0155] ;

[0156] in, is an estimated value, is the predicted mean of the objective function value, is the variance term, is the minimum value of the objective function, For samples to be updated, Update samples for the Kriging model The prediction variance of is the cumulative distribution function of the standard normal distribution, is the probability density function of the standard normal distribution, Estimated values ​​of process parameters.

[0157] The fast simulation agent model adopts the cross-validation method, and the accuracy evaluation is to be evaluated by the determination coefficient. The expression formula of the determination coefficient can be:

[0158] ;

[0159] in, is the coefficient of determination, is the actual value of the i-th observation point, is the predicted value of the ith observation point, is the average of all actual values, i is the sample number, and p is the total number of samples.

[0160] It should also be noted that the optimization model and algorithm used in the fast simulation agent model can also use other intelligent optimization algorithms such as NSGA-II (Non-dominated Sorting Genetic Algorithm II) or discrete optimization algorithms to solve the Pareto frontier.

[0161] Optionally, the field quantity prediction results are input into the Kriging model to obtain the nonlinear mapping relationship between the process parameters to be optimized in the process parameter set to be optimized and the objective function, and the objective function value of the process parameters to be optimized in the process parameter set to be optimized is calculated in combination with the expected improvement plus point criterion, and the objective function value is the optimized process parameter.

[0162] In the extrusion process parameter optimization method provided in the present application, linear order reduction and nonlinear mapping are taken into consideration when establishing a fast simulation proxy model. The spatial degrees of freedom are reduced by linear modal order reduction, and the neglect of nonlinearity in the intrinsic orthogonal decomposition process is compensated by BPNN.

[0163] In summary, the extrusion process parameter optimization method provided by the present application first uses a deep neural network to process the process parameters to be optimized, obtains the shape prediction results of the tubular parts, inputs the simulation results and shape prediction results of the tubular parts into the interpolation model, obtains the interpolated uniform regular grid, then inputs the interpolated uniform regular grid into the intrinsic field quantity prediction model to obtain the field quantity prediction results, and finally inputs the process parameters to be optimized and the field quantity prediction results into the fast simulation proxy model to obtain the optimized process parameter set of the tubular parts. The present application can quickly calculate the optimized process parameters of the tubular parts through the fast simulation proxy model, which greatly improves the efficiency of optimizing the extrusion process.

[0164] It should be noted that the present application adopts the intrinsic field quantity prediction model to predict the field quantity data. For the scalar results, in addition to the prediction using the BPNN model mentioned above, the RNN (Recurrent Neural Network) model or LSTM (Long Short-Term Memory) and other time series-based models can also be used for prediction.

[0165] In addition, an SBO optimization framework can be established, under which the fast simulation agent model and the optimization algorithm model can be two independent modules, which can be subsequently expanded and combined with multiple algorithms for different problems.

[0166] A method for optimizing extrusion process parameters provided in an embodiment of the present application is introduced above, and a device for executing the above-mentioned method for optimizing extrusion process parameters will be introduced below.

[0167] See also Figure 6 , Figure 6 This is a schematic diagram of the structure of an extrusion process parameter optimization device provided in this application. Figure 6 As shown, the device comprises:

[0168] Shape prediction unit 10, interpolation unit 20, field quantity prediction unit 30 and process parameter optimization unit 40; wherein:

[0169] The shape prediction unit 10 is used to input the process parameter set to be optimized of the tubular part into the deep neural network model to obtain the shape prediction result of the tubular part; the process parameter set to be optimized is: a set of key sensitive process parameters obtained after sampling the simulation results of the tubular part extrusion process and performing parameter sensitivity analysis;

[0170] The interpolation unit 20 is used to input the simulation results and the shape prediction results into the interpolation model to obtain the interpolated uniform regular grid; the interpolation model is used to interpolate the field quantity data of different unstructured grids with different process parameters to the uniform regular grid to be interpolated to obtain the interpolated uniform regular grid; the different unstructured grids with different process parameters are obtained by discretizing the simulation results;

[0171] A field quantity prediction unit 30 is used to input the interpolated uniform regular grid into the intrinsic field quantity prediction model to obtain a field quantity prediction result;

[0172] The process parameter optimization unit 40 is used to input the process parameter set to be optimized and the field quantity prediction results into the fast simulation agent model to obtain the optimized process parameter set of the tubular part; the fast simulation agent model is used to process the process parameters to be optimized in the process parameter set to be optimized through the nonlinear mapping relationship between the process parameters and the objective function to obtain the optimized process parameter set.

[0173] In one embodiment, the shape prediction result in the interpolation unit 20 includes the punch end displacement result of the blank and the free end displacement result of the tubular member; the interpolation unit 20 is specifically used for:

[0174] The displacement results of the punch end and the free end are used to update the grid coordinates of the uniform regular grid to be interpolated;

[0175] The field quantity data of each grid of different unstructured grids with different process parameters are correspondingly interpolated to each grid of the uniform regular grid to be interpolated.

[0176] In one embodiment, the intrinsic field quantity prediction model in the field quantity prediction unit 30 is a prediction model that integrates the intrinsic orthogonal decomposition model and the modal participation coefficient solution model. The field quantity prediction unit 30 is specifically used for:

[0177] The intrinsic orthogonal decomposition model is used to process the field quantity data in the interpolated uniform regular grid into low-order linear modal field quantity data;

[0178] The modal participation coefficient solution model is used to calculate the modal participation coefficient required for the field quantity data prediction of the intrinsic orthogonal decomposition model;

[0179] The intrinsic orthogonal decomposition model uses the modal participation coefficient calculated by the modal participation coefficient solution model to solve the modal participation coefficient and obtain the field quantity prediction result.

[0180] In one embodiment, the modal participation coefficient solution model in the field quantity prediction unit 30 includes a back propagation neural network model, which is used to update the modal participation coefficients in the intrinsic orthogonal decomposition model and also used to predict the scalar data of the tubular component.

[0181] In one embodiment, the fast simulation proxy model in the process parameter optimization unit 40 is a nonlinear proxy model between the process parameters and the objective function based on the Kriging model; the process parameter optimization unit 40 is specifically used for:

[0182] The field quantity prediction result is input into the Kriging model to obtain a nonlinear mapping relationship between the process parameters to be optimized and the objective function in the process parameter set to be optimized;

[0183] The objective function value of the process parameter to be optimized in the process parameter set to be optimized is calculated in combination with the expected improvement plus point criterion, and the objective function value is the optimized process parameter.

[0184] In one embodiment, the extrusion process parameter optimization device further includes a process parameter set determination unit;

[0185] A process parameter set determination unit, used for establishing a 2D axisymmetric equivalent model of the tubular component using a numerical simulation tool;

[0186] Sampling the input process parameters of the 2D axisymmetric equivalent model establishment process to obtain a process parameter set;

[0187] Parameter sensitivity analysis is performed on each process parameter in the process parameter set to obtain the process parameter set to be optimized.

[0188] The present application also provides an extrusion process parameter optimization device. Figure 7 As shown, it shows a schematic diagram of the structure of the extrusion process parameter optimization device suitable for implementing the extrusion process parameter optimization device provided in the present application. The extrusion process parameter optimization device in the embodiment of the present application may include but is not limited to fixed terminals such as mobile phones, laptop computers, PDAs (personal digital assistants), PADs (tablet computers), desktop computers, etc. Figure 7 The extrusion process parameter optimization device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0189] like Figure 7 As shown, the extrusion process parameter optimization device may include a processing device (such as a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage device 608 to a random access memory (RAM) 603. When the extrusion process parameter optimization device is powered on, various programs and data required for the operation of the extrusion process parameter optimization device are also stored in the RAM 603. The processing device 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0190] Typically, the following devices may be connected to the I / O interface 605: input devices 606 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; output devices 607 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 608 including, for example, a memory card, a hard disk, etc.; and communication devices 609. The communication device 609 may allow the extrusion process parameter optimization device to communicate with other devices wirelessly or by wire to exchange data. Although Figure 7 The extrusion process parameter optimization apparatus having various devices is shown, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0191] Also provided in an embodiment of the present application is a computer program product, including computer-readable instructions. When the computer-readable instructions are executed on an electronic device, the electronic device implements any one of the extrusion process parameter optimization methods provided in the embodiments of the present application.

[0192] A computer storage medium is also provided in an embodiment of the present application. The storage medium carries one or more computer programs. When one or more computer programs are executed by an electronic device, the electronic device can implement any extrusion process parameter optimization method provided in the embodiment of the present application.

[0193] It should also be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the drawings of the device embodiments provided by the present application, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines.

[0194] Through the description of the above implementation mode, the technicians in the field can clearly understand that the present application can be implemented by means of software plus necessary general hardware, and of course, it can also be implemented by special hardware including special integrated circuits, special CPUs, special memories, special components, etc. In general, all functions completed by computer programs can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be various, such as analog circuits, digital circuits or special circuits. However, for the present application, software program implementation is a better implementation mode in more cases. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a readable storage medium, such as a computer floppy disk, a U disk, a mobile hard disk, a ROM, a RAM, a disk or an optical disk, etc., including a number of instructions to enable a computer device (which can be a personal computer, a training device, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0195] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product.

[0196] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on the computer, the process or function according to the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, a computer, a training device or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, training device or data center. The computer-readable storage medium can be any available medium that a computer can store or a data storage device such as a training device, a data center, etc. that includes one or more available media integration. Available media can be magnetic media, (e.g., floppy disk, hard disk, tape), optical media (e.g., DVD), or semiconductor media (e.g., solid-state hard disk (SSD)), etc.

Claims

1. A method for optimizing extrusion process parameters, characterized in that: include: Inputting a set of process parameters to be optimized for the tubular part into a deep neural network model to obtain a shape prediction result of the tubular part; The process parameter set to be optimized is: a set of key sensitive process parameters obtained after sampling the simulation results of the tubular part extrusion process and performing parameter sensitivity analysis; The simulation result and the shape prediction result are input into an interpolation model to obtain an interpolated uniform regular grid; the interpolation model is used to: interpolate the field quantity data of different unstructured grids with different process parameters to the uniform regular grid to be interpolated to obtain the interpolated uniform regular grid; the different unstructured grids with different process parameters are obtained by discretizing the simulation result; Inputting the interpolated uniform regular grid into an intrinsic field quantity prediction model to obtain a field quantity prediction result; Inputting the process parameter set to be optimized and the field quantity prediction result into a fast simulation agent model to obtain an optimized process parameter set for the tubular part; The fast simulation agent model is used to process the process parameters to be optimized in the process parameter set to be optimized through the nonlinear mapping relationship between the process parameters and the objective function to obtain the optimized process parameter set.

2. The extrusion process parameter optimization method according to claim 1, characterized in that: The shape prediction result includes the displacement result of the punch end of the blank and the displacement result of the free end of the tubular part; the field quantity data of different unstructured grids with different process parameters are interpolated to the uniform regular grid to be interpolated to obtain the uniform regular grid after interpolation, including: Using the punch end displacement result and the free end displacement result to update the grid coordinates of the uniform regular grid to be interpolated; The field quantity data of each grid of the different unstructured grids with different process parameters are correspondingly interpolated to each grid of the uniform regular grid to be interpolated.

3. The extrusion process parameter optimization method according to claim 1, characterized in that: The intrinsic field quantity prediction model is a prediction model that integrates the intrinsic orthogonal decomposition model and the modal participation coefficient solution model. The interpolated uniform regular grid is input into the intrinsic field quantity prediction model to obtain the field quantity prediction result, including: The intrinsic orthogonal decomposition model is used to process the field quantity data in the interpolated uniform regular grid into low-order linear modal field quantity data; The modal participation coefficient solution model is used to calculate the modal participation coefficient required for the intrinsic orthogonal decomposition model to predict field quantity data; The intrinsic orthogonal decomposition model uses the modal participation coefficient calculated by the modal participation coefficient solution model to solve the modal participation coefficient to obtain the field quantity prediction result.

4. The extrusion process parameter optimization method according to claim 3, characterized in that: The modal participation coefficient solution model includes a back propagation neural network model; The back propagation neural network model is used to update the modal participation coefficients in the intrinsic orthogonal decomposition model and is also used to predict the scalar data of the tubular member.

5. The extrusion process parameter optimization method according to claim 1, characterized in that: The fast simulation proxy model is a nonlinear proxy model between the process parameters and the objective function constructed based on the Kriging model; The step of inputting the process parameter set to be optimized and the field quantity prediction result into the fast simulation agent model to obtain the optimized process parameter set of the tubular part comprises: Inputting the field quantity prediction result into the Kriging model to obtain a nonlinear mapping relationship between the process parameters to be optimized in the process parameter set to be optimized and the objective function; The objective function value of the process parameter to be optimized in the process parameter set to be optimized is calculated in combination with the expected improvement plus point criterion, and the objective function value is the optimized process parameter.

6. The extrusion process parameter optimization method according to claim 1, characterized in that: Before the set of process parameters to be optimized for the tubular part is input into the deep neural network model, it also includes: A 2D axisymmetric equivalent model of the tubular component is established by using a numerical simulation tool; Sampling the input process parameters of the 2D axisymmetric equivalent model establishment process to obtain a process parameter set; A parameter sensitivity analysis is performed on each process parameter in the process parameter set to obtain the process parameter set to be optimized.

7. An extrusion process parameter optimization device, characterized in that: include: A shape prediction unit, used for inputting a set of process parameters to be optimized for the tubular part into a deep neural network model to obtain a shape prediction result of the tubular part; The process parameter set to be optimized is: a set of key sensitive process parameters obtained after sampling the simulation results of the tubular part extrusion process and performing parameter sensitivity analysis; An interpolation unit is used to input the simulation result and the shape prediction result into an interpolation model to obtain an interpolated uniform regular grid; the interpolation model is used to interpolate the field quantity data of different unstructured grids with different process parameters to the uniform regular grid to be interpolated to obtain the interpolated uniform regular grid; the different unstructured grids with different process parameters are obtained by discretizing the simulation result; A field quantity prediction unit, used for inputting the interpolated uniform regular grid into an intrinsic field quantity prediction model to obtain a field quantity prediction result; A process parameter optimization unit, used for inputting the process parameter set to be optimized and the field quantity prediction result into a fast simulation agent model to obtain an optimized process parameter set of the tubular part; The fast simulation agent model is used to process the process parameters to be optimized in the process parameter set to be optimized through the nonlinear mapping relationship between the process parameters and the objective function to obtain the optimized process parameter set.

8. An extrusion process parameter optimization device, characterized in that: The method comprises at least one processor and a memory connected to the processor, wherein: The memory is used to store computer programs; The processor is used to execute the computer program so that the extrusion process parameter optimization device can implement the extrusion process parameter optimization method according to any one of claims 1 to 6.

9. A computer program product, characterized in that It comprises computer-readable instructions, and when the computer-readable instructions are executed on an electronic device, the electronic device implements the extrusion process parameter optimization method as claimed in any one of claims 1 to 6.

10. A computer storage medium, characterized in that: The storage medium carries one or more computer programs, and when the one or more computer programs are executed by an electronic device, the electronic device can implement the extrusion process parameter optimization method as described in any one of claims 1 to 6.

Images