Digital modeling method and system for physical field of complex shape component die forging forming process

Abstract

This invention discloses a method and system for digital modeling of the physical field during the forging process of complex-shaped components. The method includes: establishing a finite element model of the forging process and obtaining the physical field at any given moment through simulation calculation; dividing the forming process into multiple stages to obtain the transition points and end points of the forging stages; designing multiple different combinations of forging process parameters and performing finite element numerical simulation calculations for each set of parameters; establishing a coordinate system and selecting data sampling points on the forging based on the Poisson disk sampling method; for each stage transition point or end point, applying the finite element simulation calculation results to obtain the deformation physical quantities of the data sampling points and constructing a dataset; selecting a machine learning algorithm and applying the dataset to establish a prediction model; using the prediction model to obtain the predicted values ​​of the deformation physical quantities at the target stage transition point or end point; and using an interpolation algorithm to reconstruct the physical field and obtain the full-field digital results of the physical field, thereby achieving accurate prediction of the physical field during the forging process of complex-shaped components.

Description

Technical Field

[0001] This invention relates to the field of forging technology, and in particular to a method and system for digital modeling of the physical field in the die forging process of complex shaped components. Background Technology

[0002] The evolution of physical fields in typical hot working processes such as metal casting, forging, and additive manufacturing is a key factor determining the microstructure, properties, and quality of components. For a long time, because the physical fields such as temperature, stress, and strain fields during hot working processes cannot be directly observed, offline studies have typically been conducted using finite element numerical simulation methods. However, finite element simulation methods require long computation times, cannot establish real-time mapping relationships between hot working process parameters and physical fields, and are insufficient to meet the development needs of digital and intelligent manufacturing.

[0003] Scholars and technicians have already used deep learning methods to conduct research on digital modeling of physical fields in various scenarios. For example, patent CN 120280061 A discloses a method for predicting the microscopic physical fields of composite materials based on an improved convolutional neural network; patent CN 120145042A discloses a method for predicting the physical fields of satellites based on a deep neural network; and patent CN118332877 B discloses a method for calculating electromagnetic fields based on a physical information convolutional neural network. All of the above works are based on deep learning methods, which can only achieve digital modeling of physical fields with simple shapes and small gradient changes. This is because deep neural networks use multiple convolutional and pooling layers to extract and reduce the dimensionality of pixel matrix information in physical field image data, resulting in significant information loss during processing, especially when dealing with complex image data. To achieve accurate modeling of the entire physical field, a huge amount of image training data is required. Furthermore, the increasing complexity of the physical field's geometry and the increase in gradient changes significantly increase the difficulty of deep neural network modeling, making it difficult for the model accuracy to meet the requirements of engineering applications.

[0004] Die forging is a fundamental and crucial forming technology for complex components used in the manufacturing of high-end equipment such as aerospace and transportation. The physical fields involved in the die forging process are characterized by complex geometry and high non-uniformity of distribution. Traditional finite element simulation methods struggle to achieve real-time simulation, while deep learning methods fail to achieve accurate digital modeling across the entire field. Therefore, researching novel methods for digital modeling the evolution of the physical fields during the die forging process of complex-shaped components, and establishing rapid (sub-second) and accurate mapping relationships between process parameters and physical fields such as the temperature field, strain field, stress field, and strain rate field of the forging, is a significant requirement in the field of digital and intelligent forging technology for complex metal components. Summary of the Invention

[0005] To address the aforementioned problems, the present invention aims to provide a method and system for digital modeling of the physical field in the forging process of complex-shaped components. This method integrates multiple methods such as finite element simulation, Poisson disk sampling, machine learning, and thin plate spline interpolation. By discretizing field information extraction, constructing a dataset of feature points (data sampling points), machine learning modeling, and full-field digital reconstruction, the method achieves overall field prediction, solving the problem of accurate full-field digital modeling of physical fields with complex shape spaces and significant inhomogeneities. This enables accurate and rapid prediction of multiple physical fields in the forming process of complex-shaped components.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] On the one hand, a method for digital modeling of the physical field in the forging process of complex-shaped components is provided, the method comprising the following steps:

[0008] S1. Establish a finite element model of the forging process for finite element numerical simulation analysis of the physical field of the entire forming process, so as to obtain the physical field at any time.

[0009] S2. Based on the metal flow and cavity filling patterns during the forging process, the forming process is divided into n stages, with different forging speeds used in each stage to achieve uniform control of the physical field during the forging process. The end time of the first n-1 stages is called the transition point of the forging stage, resulting in a total of n-1 transition points. The end time of the last stage is called the end point of the forging process.

[0010] S3. Within a reasonable range of process parameters, design multiple sets of different forging process parameter combinations, and perform finite element numerical simulation calculations for each set of parameters to construct a physical field dataset.

[0011] S4. Taking into account the shape and structural characteristics of the forging, select an appropriate position on the forging as the origin of the coordinate system; based on the Poisson disk sampling method, select p feature points in the forging corresponding to each stage transition point or end point as data sampling points.

[0012] S5. Construct a finite element simulation dataset: For each transition point or end point of the forging process, apply the finite element simulation calculation results to obtain the deformation physical quantities of the data sampling points, including deformation temperature, strain rate, equivalent stress and equivalent strain, and construct a physical field dataset.

[0013] S6. Establish a machine learning prediction model: Select a transition point or end point of a certain stage in the forging process, take the forging process parameters and data sampling point coordinates as input, take the deformation physical quantity as output, select an appropriate machine learning algorithm, and apply the physical field dataset to establish a prediction model.

[0014] S7. Full-field prediction of physical field: Input real-time forging process parameters, use the established prediction model to obtain the predicted values ​​of deformation temperature, strain rate, equivalent stress and equivalent strain of p data sampling points in the forging corresponding to the transition point or end point of the target stage; based on the predicted values ​​of the data sampling points, use the interpolation algorithm to reconstruct the deformation temperature field, strain rate field, equivalent stress field and equivalent strain field of the transition point or end point of the target stage, and obtain the full-field digital result of the physical field.

[0015] Optionally, step S3 specifically includes the following steps:

[0016] S31. Design multiple parameter levels for different forging process parameters, wherein the forging process parameters include the friction factor λ between the billet and the die, the billet temperature T1, the die temperature T2, and the reduction speeds v1, v2, ..., v at n forging stages. n ;

[0017] S32. Using the orthogonal experimental design method, different combinations of forging process parameters were designed, and finite element numerical simulation calculations were performed.

[0018] Optionally, step S6 specifically includes the following steps:

[0019] S61. Determine the input variables for machine learning modeling: When predicting the deformation temperature, strain rate, equivalent stress, and equivalent strain at the transition point of the first stage, use λ, T1, T2, v1, and a i b i c i For input variables, where a i b i c i For data sampling point p i The coordinates; when predicting the deformation temperature, strain rate, equivalent stress, and equivalent strain at the transition point of the second stage, using λ, T1, T2, v1, v2, a i b i c i As input variables; and so on, when predicting the deformation temperature, strain rate, equivalent stress, and equivalent strain at the end point, use λ, T1, T2, v1, v2, ..., v n a i b i c i For input variables;

[0020] S62. Selecting a machine learning algorithm: The machine learning algorithm includes support vector regression, random forest, and gradient boosting regression, selected using the coefficient of determination R0. 2 The model performance was evaluated using the Mean Absolute Percentage Error (MAPE), and the algorithm with higher prediction accuracy was selected to build the machine learning prediction model.

[0021]

[0022] In the formula, m is the sample size, Y j For finite element simulation values, y j The model predicts the value; the coefficient of determination R0 2 The larger the value, the smaller the mean absolute percentage error (MAPE), and the higher the model prediction accuracy.

[0023] Optionally, step S7 specifically includes the following steps:

[0024] S71. Using the forging process parameters acquired in real time as input, the established prediction model is used to obtain the deformation physical quantities of p data sampling points in the forging at the transition point or end point of the target stage, including deformation temperature, strain rate, equivalent stress and equivalent strain.

[0025] S72. Using thin plate spline interpolation, calculate the interpolation function u of each physical field within the forging at the transition point or end point of the target stage according to the following formula, and then determine the full-field digital distribution of the deformation temperature field, strain rate field, equivalent stress field, and equivalent strain field:

[0026]

[0027] Among them, w i K is the weight of each data sampling point, representing the influence of the data sampling point on the interpolation point; K is the kernel function, which is the function that the data sampling point influences the interpolation point value; r is the weight of each data sampling point (a i ,b i ,c i The Euclidean distance between (a,b,c) and the interpolation point (a,b,c); k0, k1, k2, k3 are the coefficients of the linear term, all obtained through fitting;

[0028] During fitting, the smoothness and accuracy of the fitted data are balanced by minimizing the energy functional E.

[0029]

[0030] in, It is the data fitting term that makes the function fit value u(a) i ,b i ,c i Try to obtain the predicted value y of the machine learning prediction model using p data sampling points. i ; It is a smoothing term that controls the smoothness of the surface and avoids overfitting; D represents the entire interpolation region (a,b,c); These are the second-order partial derivatives of u with respect to a, b, and c, respectively. These are the mixed partial derivatives of u with respect to a and b, a and c, and b and c, respectively; α is the weighting factor, representing the trade-off between fitting data error and data smoothing.

[0031] On the other hand, a physical field digital modeling system for the forging process of complex shaped components is provided, for implementing the method described in any of the above claims, the system comprising:

[0032] The finite element model building module is used to build a finite element model of the forging process, and to perform finite element numerical simulation analysis of the physical field of the entire forming process to obtain the physical field at any time.

[0033] The stage division module is used to divide the forming process into n stages based on the metal flow and cavity filling pattern during the forging process. Each stage adopts a different forging speed to achieve uniform control of the physical field of the forging process. The end time of the first n-1 stages is called the forging stage transition point, and a total of n-1 stage transition points are obtained. The end time of the last stage is called the forging end point.

[0034] The parameter design module is used to design multiple different combinations of forging process parameters within a reasonable range of process parameters, and to perform finite element numerical simulation calculations for each set of parameters to build a physical field dataset.

[0035] The sampling point selection module is used to comprehensively consider the shape and structural characteristics of the forging, select an appropriate position on the forging as the origin of the coordinate system, and establish a coordinate system; based on the Poisson disk sampling method, p feature points are selected in the forging corresponding to the stage transition point or end point of each stage as data sampling points;

[0036] The dataset construction module is used to apply finite element simulation results to obtain the deformation physical quantities of the data sampling points for each transition point or end point of the forging process, including deformation temperature, strain rate, equivalent stress and equivalent strain, and to construct a physical field dataset.

[0037] The machine learning prediction model building module is used to select a transition point or end point of a certain stage in the forging process. It takes forging process parameters and data sampling point coordinates as input, deformation physical quantities as output, selects appropriate machine learning algorithms, and applies physical field datasets to build a prediction model.

[0038] The full-field physical field prediction module is used to input real-time forging process parameters and use the established prediction model to obtain the predicted values ​​of deformation temperature, strain rate, equivalent stress, and equivalent strain of p data sampling points in the forging corresponding to the transition point or end point of the target stage. Based on the predicted values ​​of the data sampling points, the deformation temperature field, strain rate field, equivalent stress field, and equivalent strain field of the transition point or end point of the target stage are reconstructed using an interpolation algorithm to obtain the full-field digital results of the physical field.

[0039] On the other hand, an electronic device is provided, the electronic device comprising:

[0040] processor;

[0041] The memory stores computer-readable instructions, which, when loaded and executed by the processor, implement the steps of the physical field digital modeling method for the forging process of complex-shaped components as described above.

[0042] On the other hand, a computer-readable storage medium is provided, wherein program code is stored in the computer-readable storage medium, and the program code can be called by a processor to execute the steps of the physical field digital modeling method for the forging process of complex shaped components described above.

[0043] The beneficial effects of the technical solution provided by this invention include at least the following:

[0044] (1) This invention can quickly predict the physical field of the forging process in less than a second, breaking through the bottleneck of traditional finite element simulation being time-consuming and unable to achieve online prediction, laying the foundation for digital and intelligent control of the die forging process;

[0045] (2) The present invention can accurately predict the physical field with complex shape space and significant non-uniformity (large gradient change), and meet the real-time analysis requirements of the physical field of the entire process of forging complex shaped components.

[0046] (3) This invention can be widely applied to the digital modeling of the entire physical field of the forging process of complex components in high-end manufacturing fields such as aerospace, rail transit, and new energy vehicles, and has good versatility. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 This is a flowchart of a physical field digital modeling method for a complex-shaped component forging process provided by an embodiment of the present invention;

[0049] Figure 2 These are schematic diagrams of the three-dimensional model and external dimensions of the forging provided in the embodiments of the present invention; wherein: (a) is a schematic diagram of the three-dimensional model of the forging, and (b) is a schematic diagram of the external dimensions of the forging;

[0050] Figure 3This is a schematic diagram of the forging stroke-load curve and forming process provided in an embodiment of the present invention;

[0051] Figure 4 This is a finite element mesh diagram and a schematic diagram of 100 data sampling points for the transition points and end points of the first to third stages of the forging process provided in this embodiment of the invention; wherein: (a), (b), and (c) are respectively the finite element mesh diagram (left side) and the schematic diagram of 100 data sampling points (black dots in the figure) (right side) for the transition points of the first to third stages of the forging process; (d) is the finite element mesh diagram (left side) and the schematic diagram of 100 data sampling points (black dots in the figure) (right side) for the end point of the forging process;

[0052] Figure 5 These are schematic diagrams illustrating the prediction effects of the deformation physical field prediction models at the transition points and end points of each stage of the die forging process provided in this embodiment of the invention; wherein, (a1) to (a4) are schematic diagrams illustrating the prediction effects of deformation temperature; (b1) to (b4) are schematic diagrams illustrating the prediction effects of strain rate; (c1) to (c4) are schematic diagrams illustrating the prediction effects of equivalent stress; and (d1) to (d4) are schematic diagrams illustrating the prediction effects of equivalent strain.

[0053] Figure 6 This is a schematic diagram showing the effect of rapid full-field prediction and reconstruction of the physical field under randomly generated process parameters provided in this embodiment of the invention, and its comparison with the finite element simulation results; wherein, (a1)~(a4) are schematic diagrams comparing deformation temperature; (b1)~(b4) are schematic diagrams comparing strain rate; (c1)~(c4) are schematic diagrams comparing equivalent stress; (d1)~(d4) are schematic diagrams comparing equivalent strain; in each figure, the left side of the center line is the finite element simulation result, and the right side is the model prediction and reconstruction result;

[0054] Figure 7 This is a schematic diagram of the physical field digital modeling system for the forging process of complex shaped components provided in an embodiment of the present invention. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0056] In embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Rather, the term "exemplary" is used to present concepts in a specific manner.

[0057] This invention provides a method for digital modeling of the physical field in the forging process of complex-shaped components, with reference to... Figure 1 As shown, the method includes the following steps:

[0058] S1. Establish a finite element model of the forging process for finite element numerical simulation analysis of the physical field of the entire forming process, so as to obtain the physical field at any time.

[0059] like Figure 2 The deep-cylinder thin-walled part shown is a specific forging provided in this embodiment. Figure 2 (a) is a schematic diagram of the three-dimensional model of the forging. Figure 2 Figure (b) shows a schematic diagram of the forging's external dimensions. The forging has a height of 34 mm and a minimum cylinder wall thickness of 3.6 mm, making it challenging to forge. The forging material is AA2014 aluminum alloy, and the billet size is φ50×19 mm. The forming process of the forging was simulated using Deform-3D software as the finite element numerical simulation platform, employing a rigid-plastic finite element model. The mesh was initially divided into tetrahedral shapes, with approximately 100,000 meshes, which were re-divided during the simulation based on mesh distortion. The die was considered a rigid body, and a shear friction model was selected as the friction boundary condition between the billet and the die. The heat exchange coefficient between the billet and the die was set to 11 N(s·mm·℃). -1 The heat transfer coefficient between the billet and air was set to 0.02 N(s·mm·℃). -1 The air temperature is set to 20℃.

[0060] S2. Based on the metal flow and cavity filling patterns during the forging process, the forming process is divided into n stages, with different forging speeds used in each stage to achieve uniform control of the physical field during the forging process. The end time of the first n-1 stages is called the transition point of the forging stage, resulting in a total of n-1 transition points. The end time of the last stage is called the end point of the forging process.

[0061] Specifically, such as Figure 3As shown, the forming process is divided into four stages: when the upper die stroke is 0–9 mm, it is the lower surface forming stage, with a forging speed of v1; when the upper die stroke is 9–12 mm, it is the upper surface forming stage, with a forging speed of v2; when the upper die stroke is 12–17 mm, it is the reverse extrusion filling stage, with a forging speed of v3; and when the upper die stroke exceeds 17 mm, it is the final filling stage, with a forging speed of v4. Based on the above division, the forging strokes of 9 mm, 12 mm, and 17 mm are respectively referred to as the transition points of the 1st, 2nd, and 3rd forging forming stages, and the forging stroke of 18 mm is referred to as the forging forming end point.

[0062] S3. Within a reasonable range of process parameters, design multiple different combinations of forging process parameters, and perform finite element numerical simulation calculations for each set of parameters to construct a physical field dataset.

[0063] This step specifically includes:

[0064] S31. Design multiple parameter levels for different forging process parameters, wherein the forging process parameters include the friction factor λ between the billet and the die, the billet temperature T1, the die temperature T2, and the reduction speeds v1, v2, ..., v at n forging stages. n .

[0065] As an optional embodiment of the present invention, the friction factor λ is taken at five levels: 0.1, 0.2, 0.3, 0.4, and 0.5; the mold temperature T1 is taken at seven levels: 180, 200, 220, 240, 260, 280, and 300℃; the billet temperature T2 is taken at ten levels: 300, 330, 360, 390, 400, 420, 440, 450, 460, and 480℃; and the pressing speeds v1, v2, v3, and v4 at the four stages are taken at eight levels: 1, 3, 5, 6, 7, 9, 12, and 15 mm / s.

[0066] S32. Using orthogonal experimental design, different combinations of forging process parameters are designed, and finite element numerical simulations are performed. The orthogonal experimental design method ensures the typicality of the designed process parameter combinations while minimizing the number of necessary combinations.

[0067] Specifically, the orthogonal method is used to design the pressing speed v in four stages, including the friction factor λ, billet temperature T1, die temperature T2, and the pressing speed v. n Finite element numerical simulation calculations were performed on 98 process schemes including seven factors (n = 1, 2, 3, 4).

[0068] S4. Taking into account the shape and structural characteristics of the forging, select an appropriate position on the forging as the origin of the coordinate system; based on the Poisson disk sampling method, select p feature points in the forging corresponding to each stage transition point or end point as data sampling points.

[0069] Since the forging in this embodiment is an axisymmetric part, a rectangular coordinate system is established on a plane perpendicular to the bottom surface, with the center of the circle on the bottom surface of the forging as the origin, and the x and z axes respectively. Based on the Poisson disk sampling method, 100 feature points are selected within the component corresponding to the transition point or end point of each stage as data sampling points, such as... Figure 4 As shown. Figure 4 In the middle: (a), (b), and (c) are the finite element mesh generation diagrams (left) and the schematic diagrams (right) of 100 data sampling points (black dots in the figure) at the transition points of the first to third stages of the die forging process, respectively; (d) is the finite element mesh generation diagram (left) and the schematic diagram (right) of 100 data sampling points (black dots in the figure) at the end point of the die forging process.

[0070] S5. Construct a finite element simulation dataset: For each transition point or end point of the forging process, apply the finite element simulation results to obtain the deformation physical quantities of the data sampling points, including deformation temperature, strain rate, equivalent stress, and equivalent strain, and construct a physical field dataset.

[0071] S6. Establish a machine learning prediction model: Select a transition point or end point in the forging process, take the forging process parameters and data sampling point coordinates as input, take the deformation physical quantities (deformation temperature, strain rate, equivalent stress and equivalent strain) as output, select a suitable machine learning algorithm, and apply the physical field dataset to establish a prediction model.

[0072] This step specifically includes:

[0073] S61. Determine the input variables for machine learning modeling: Taking the four stages divided above as an example, when predicting the deformation temperature, strain rate, equivalent stress, and equivalent strain at the transition point of the first stage, use λ, T1, T2, v1, x i z i Let x be the input variable, where x i z i For data sampling point p i The coordinates (the same below); when predicting the deformation temperature, strain rate, equivalent stress, and equivalent strain at the transition point of the second stage, λ, T1, T2, v1, v2, x i z i As input variables; when predicting the deformation temperature, strain rate, equivalent stress, and equivalent strain at the transition point of the third stage, use λ, T1, T2, v1, v2, v3, and x. i z i As input variables; when predicting the deformation temperature, strain rate, equivalent stress, and equivalent strain at the forming end point, use λ, T1, T2, v1, v2, v3, v4, and x. i zi For input variables.

[0074] S62. Selecting a machine learning algorithm: The machine learning algorithm includes support vector regression, random forest, and gradient boosting regression, selected using the coefficient of determination R0. 2 The model performance was evaluated using the Mean Absolute Percentage Error (MAPE), and the algorithm with higher prediction accuracy was selected to build the machine learning prediction model.

[0075]

[0076] In the formula, m is the sample size, Y j For finite element simulation values, y j The model predicts the value; the coefficient of determination R0 2 The larger the value, the smaller the mean absolute percentage error (MAPE), and the higher the model prediction accuracy.

[0077] In this embodiment of the invention, the gradient boosting regression algorithm was ultimately selected for machine learning modeling, and the model prediction effect is as follows: Figure 5 As shown. Figure 5 In the diagram, (a1) to (a4) are schematic diagrams of the deformation temperature prediction effect; (b1) to (b4) are schematic diagrams of the strain rate prediction effect; (c1) to (c4) are schematic diagrams of the equivalent stress prediction effect; and (d1) to (d4) are schematic diagrams of the equivalent strain prediction effect.

[0078] S7. Full-field prediction of physical field: Input real-time forging process parameters, use the established prediction model to obtain the predicted values ​​of deformation temperature, strain rate, equivalent stress and equivalent strain of p data sampling points in the forging corresponding to the transition point or end point of the target stage; based on the predicted values ​​of the data sampling points, use the interpolation algorithm to reconstruct the deformation temperature field, strain rate field, equivalent stress field and equivalent strain field of the transition point or end point of the target stage, and obtain the full-field digital result of the physical field.

[0079] This step specifically includes:

[0080] S71. Using the forging process parameters acquired in real time as input, the established prediction model is used to obtain the deformation physical quantities of 100 data sampling points in the forging at the transition point or end point of the target stage, including deformation temperature, strain rate, equivalent stress and equivalent strain.

[0081] S72. Using thin plate spline interpolation, calculate the interpolation function u of each physical field within the forging at the transition point or end point of the target stage according to the following formula, and then determine the full-field digital distribution of the deformation temperature field, strain rate field, equivalent stress field, and equivalent strain field:

[0082]

[0083] Among them, wi K is the weight of each data sampling point, representing the influence of the data sampling point on the interpolation point; K is the kernel function, which is the function that the data sampling point influences the interpolation point value; r is the weight of the data sampling point (x). i , z i The Euclidean distance between the point (x,z) and the interpolation point (x,z); k0, k1, and k2 are the coefficients of the linear term, all obtained through fitting;

[0084] During fitting, the smoothness and accuracy of the fitted data are balanced by minimizing the energy functional E.

[0085]

[0086] in, It is the data fitting term that makes the function fit value u(x) i ,z i Try to use a machine learning prediction model to predict the value y using 100 data sampling points. i ; It is a smoothing term that controls the smoothness of the surface and avoids overfitting; D represents the entire interpolation region (x,z); These are the second partial derivatives of u with respect to x, the mixed partial derivatives with respect to x and z, and the second partial derivative with respect to z, respectively; α is a weighting factor, representing the trade-off between fitting data error and data smoothing, which can be set to 10 in this embodiment.

[0087] The real-time prediction and reconstruction effect of the physical field under randomly generated process parameters (friction factor of 0.12, billet temperature of 435℃, die temperature of 277℃, and forging speeds of 11mm / s, 1mm / s, 6mm / s, and 7mm / s in four stages) and its comparison with the finite element simulation results are shown below. Figure 6 As shown. Figure 6 In the diagram, (a1) to (a4) are schematic diagrams comparing deformation temperatures; (b1) to (b4) are schematic diagrams comparing strain rates; (c1) to (c4) are schematic diagrams comparing equivalent stress; (d1) to (d4) are schematic diagrams comparing equivalent strain. In each diagram, the left side of the center line represents the finite element simulation results, and the right side represents the model prediction and reconstruction results.

[0088] At each stage transition point (or end point), 1000 location points were randomly selected, and the mean absolute percentage error (MAPE) between the physical field prediction results and the finite element numerical simulation results was calculated. The prediction errors of the four physical fields of the corresponding forging at each stage transition point (or end point) were all less than 10%.

[0089] In this embodiment of the invention, by integrating multiple methods such as finite element simulation, Poisson disk sampling, machine learning, and thin plate spline interpolation, and through discrete extraction of field information, construction of feature point (data sampling point) datasets, machine learning modeling, and full-field digital reconstruction, the overall physical field prediction is achieved. This solves the problem of accurate full-field digital modeling of physical fields with features such as complex shape space and significant inhomogeneity, and enables accurate prediction of the physical field during the die forging process of complex shaped components.

[0090] Accordingly, embodiments of the present invention also provide a physical field digital modeling system for the forging process of complex-shaped components, such as... Figure 7 As shown, the system includes:

[0091] The finite element model establishment module 201 is used to establish a finite element model of the forging process, and to perform finite element numerical simulation analysis of the physical field of the entire forming process to obtain the physical field at any time.

[0092] The stage division module 202 is used to divide the forming process into n stages based on the metal flow and cavity filling pattern during the forging process. Each stage adopts a different forging speed to achieve uniform control of the physical field of the forging process. The end time of the first n-1 stages is called the forging stage transition point, and a total of n-1 stage transition points are obtained. The end time of the last stage is called the forging end point.

[0093] The parameter design module 203 is used to design multiple different combinations of forging process parameters within a reasonable range of process parameters, and to perform finite element numerical simulation calculations for each set of parameters to construct a physical field dataset.

[0094] The sampling point selection module 204 is used to comprehensively consider the shape and structural characteristics of the forging, select an appropriate position on the forging as the origin of the coordinate system, and establish a coordinate system; based on the Poisson disk sampling method, p feature points are selected in the forging corresponding to each stage transition point or end point as data sampling points;

[0095] The dataset construction module 205 is used to apply finite element simulation results to obtain the deformation physical quantities of the data sampling points for each transition point or end point of the forging process, including deformation temperature, strain rate, equivalent stress and equivalent strain, and to construct a physical field dataset.

[0096] The machine learning prediction model building module 206 is used to select a transition point or end point of a certain stage in the forging process, take forging process parameters and data sampling point coordinates as input, take deformation physical quantities as output, select appropriate machine learning algorithms, and apply physical field datasets to build a prediction model.

[0097] The physical field full-field prediction module 207 is used to input real-time forging process parameters and obtain the predicted values ​​of deformation temperature, strain rate, equivalent stress, and equivalent strain of p data sampling points in the forging corresponding to the transition point or end point of the target stage using the established prediction model. Based on the predicted values ​​of the data sampling points, the deformation temperature field, strain rate field, equivalent stress field, and equivalent strain field of the transition point or end point of the target stage are reconstructed using an interpolation algorithm to obtain the full-field digital results of the physical field.

[0098] For ease of explanation, Figure 7 Only the main components of the system are shown. The system in this embodiment can be used to perform... Figure 1 The technical solutions of the method embodiments shown are similar in principle and in effect, and will not be described again here.

[0099] In an exemplary embodiment, the present invention also provides an electronic device, the electronic device comprising:

[0100] processor;

[0101] The memory stores computer-readable instructions, which, when loaded and executed by the processor, implement the steps of the physical field digital modeling method for the forging process of complex-shaped components as described above.

[0102] In an exemplary embodiment, the present invention also provides a computer-readable storage medium storing at least one instruction, which is loaded and executed by a processor to implement the steps of the physical field digital modeling method for the forging process of complex-shaped components described above. For example, the computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, or optical data storage device, etc.

[0103] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0104] The use of terms such as "an embodiment," "an embodiment," "an exemplary embodiment," and "some embodiments" in the specification indicates that the described embodiment may include a specific feature, structure, or characteristic, but not every embodiment necessarily includes that specific feature, structure, or characteristic. Furthermore, when a specific feature, structure, or characteristic is described in connection with an embodiment, implementing such a feature, structure, or characteristic in conjunction with other embodiments (whether explicitly described or not) should be within the knowledge of those skilled in the art.

[0105] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0106] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for digital modeling of the physical field in the forging process of complex-shaped components, characterized in that, Includes the following steps: S1. Establish a finite element model of the forging process for finite element numerical simulation analysis of the physical field of the entire forming process, so as to obtain the physical field at any time. S2. Based on the metal flow and cavity filling patterns during the forging process, the forming process is divided into n stages, with different forging speeds used in each stage to achieve uniform control of the physical field during the forging process. The end time of the first n-1 stages is called the transition point of the forging stage, resulting in a total of n-1 transition points. The end time of the last stage is called the end point of the forging process. S3. Within a reasonable range of process parameters, design multiple sets of different forging process parameter combinations, and perform finite element numerical simulation calculations for each set of parameters to construct a physical field dataset. S4. Taking into account the shape and structural characteristics of the forging, select an appropriate position on the forging as the origin of the coordinate system; based on the Poisson disk sampling method, select p feature points in the forging corresponding to each stage transition point or end point as data sampling points. S5. Construct a finite element simulation dataset: For each transition point or end point of the forging process, apply the finite element simulation calculation results to obtain the deformation physical quantities of the data sampling points, including deformation temperature, strain rate, equivalent stress and equivalent strain, and construct a physical field dataset. S6. Establish a machine learning prediction model: Select a transition point or end point of a certain stage in the forging process, take the forging process parameters and data sampling point coordinates as input, take the deformation physical quantity as output, select an appropriate machine learning algorithm, and apply the physical field dataset to establish a prediction model. S7. Full-field prediction of physical field: Input real-time forging process parameters, use the established prediction model to obtain the predicted values ​​of deformation temperature, strain rate, equivalent stress and equivalent strain of p data sampling points in the forging corresponding to the transition point or end point of the target stage; based on the predicted values ​​of the data sampling points, use the interpolation algorithm to reconstruct the deformation temperature field, strain rate field, equivalent stress field and equivalent strain field of the transition point or end point of the target stage, and obtain the full-field digital result of the physical field.

2. The physical field digital modeling method for the forging process of complex-shaped components according to claim 1, characterized in that, S3 specifically includes the following steps: S31. Design multiple parameter levels for different forging process parameters, wherein the forging process parameters include the friction factor λ between the billet and the die, the billet temperature T1, the die temperature T2, and the reduction speeds v1, v2, ..., v at n forging stages. n ; S32. Using the orthogonal experimental design method, different combinations of forging process parameters were designed, and finite element numerical simulation calculations were performed.

3. The physical field digital modeling method for the forging process of complex-shaped components according to claim 1, characterized in that, S6 specifically includes the following steps: S61. Determine the input variables for machine learning modeling: When predicting the deformation temperature, strain rate, equivalent stress, and equivalent strain at the transition point of the first stage, use λ, T1, T2, v1, and a i b i c i For input variables, where a i b i c i For data sampling point p i The coordinates; when predicting the deformation temperature, strain rate, equivalent stress, and equivalent strain at the transition point of the second stage, using λ, T1, T2, v1, v2, a i b i c i As input variables; and so on, when predicting the deformation temperature, strain rate, equivalent stress, and equivalent strain at the end point, use λ, T1, T2, v1, v2, ..., v n a i b i c i For input variables; S62. Selecting a machine learning algorithm: The machine learning algorithm includes support vector regression, random forest, and gradient boosting regression, selected using the coefficient of determination R0. 2 The model performance was evaluated using the Mean Absolute Percentage Error (MAPE), and the algorithm with higher prediction accuracy was selected to build the machine learning prediction model. In the formula, m is the sample size, Y j For finite element simulation values, y j The model predicts the value; the coefficient of determination R0 2 The larger the value, the smaller the mean absolute percentage error (MAPE), and the higher the model prediction accuracy.

4. The physical field digital modeling method for the forging process of complex-shaped components according to claim 1, characterized in that, S7 specifically includes the following steps: S71. Using the forging process parameters acquired in real time as input, the established prediction model is used to obtain the deformation physical quantities of p data sampling points in the forging at the transition point or end point of the target stage, including deformation temperature, strain rate, equivalent stress and equivalent strain. S72. Using thin plate spline interpolation, calculate the interpolation function u of each physical field within the forging at the transition point or end point of the target stage according to the following formula, and then determine the full-field digital distribution of the deformation temperature field, strain rate field, equivalent stress field, and equivalent strain field: Among them, w i K is the weight of each data sampling point, representing the influence of the data sampling point on the interpolation point; K is the kernel function, which is the function that the data sampling point influences the interpolation point value; r is the weight of each data sampling point (a i ,b i ,c i The Euclidean distance between (a,b,c) and the interpolation point (a,b,c); k0, k1, k2, k3 are the coefficients of the linear term, all obtained through fitting; During fitting, the smoothness and accuracy of the fitted data are balanced by minimizing the energy functional E. in, It is the data fitting term that makes the function fit value u(a) i ,b i ,c i Try to obtain the predicted value y of the machine learning prediction model using p data sampling points. i ; It is a smoothing term that controls the smoothness of the surface and avoids overfitting; D represents the entire interpolation region (a,b,c); These are the second-order partial derivatives of u with respect to a, b, and c, respectively. These are the mixed partial derivatives of u with respect to a and b, a and c, and b and c, respectively; α is the weighting factor, representing the trade-off between fitting data error and data smoothing.

5. A physical field digital modeling system for the forging process of complex-shaped components, the system being used to implement the method as described in any one of claims 1 to 4, characterized in that, The system includes: The finite element model building module is used to build a finite element model of the forging process, and to perform finite element numerical simulation analysis of the physical field of the entire forming process to obtain the physical field at any time. The stage division module is used to divide the forming process into n stages based on the metal flow and cavity filling pattern during the forging process. Each stage adopts a different forging speed to achieve uniform control of the physical field of the forging process. The end time of the first n-1 stages is called the forging stage transition point, and a total of n-1 stage transition points are obtained. The end time of the last stage is called the forging end point. The parameter design module is used to design multiple different combinations of forging process parameters within a reasonable range of process parameters, and to perform finite element numerical simulation calculations for each set of parameters to build a physical field dataset. The sampling point selection module is used to comprehensively consider the shape and structural characteristics of the forging, select an appropriate position on the forging as the origin of the coordinate system, and establish a coordinate system; based on the Poisson disk sampling method, p feature points are selected in the forging corresponding to each stage transition point or end point as data sampling points; The dataset construction module is used to apply finite element simulation results to obtain the deformation physical quantities of the data sampling points for each transition point or end point of the forging process, including deformation temperature, strain rate, equivalent stress and equivalent strain, and to construct a physical field dataset. The machine learning prediction model building module is used to select a transition point or end point of a certain stage in the forging process. It takes forging process parameters and data sampling point coordinates as input, deformation physical quantities as output, selects appropriate machine learning algorithms, and applies physical field datasets to build a prediction model. The full-field physical field prediction module is used to input real-time forging process parameters and use the established prediction model to obtain the predicted values ​​of deformation temperature, strain rate, equivalent stress, and equivalent strain of p data sampling points in the forging corresponding to the transition point or end point of the target stage. Based on the predicted values ​​of the data sampling points, the deformation temperature field, strain rate field, equivalent stress field, and equivalent strain field of the transition point or end point of the target stage are reconstructed using an interpolation algorithm to obtain the full-field digital results of the physical field.

6. An electronic device, characterized in that, The electronic device includes: processor; A memory storing computer-readable instructions that, when loaded and executed by the processor, implement the method as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores program code that can be invoked by a processor to execute the method as described in any one of claims 1 to 4.

Images