Forging and pressing prediction model training method and device and forging and pressing simulation method and device

Abstract

The invention discloses a forging and pressing prediction model training method, and relates to the technical field of equipment manufacturing, and the specific implementation scheme is as follows: obtaining simulation data of a forging and pressing simulation model corresponding to a forging and pressing workpiece; dividing the forging simulation model and the space where the forging simulation model is located into multiple layers of space units, and enabling each space unit to have two attributes of geometrical shape and stress; on the basis of the simulation data, matrixes of different channels are adopted to represent the geometrical shape and the stress attribute of each space unit, and a sample data set is obtained; and based on the sample data set, training the initial forging and pressing prediction model to obtain a trained forging and pressing prediction model. And a reliable implementation means is provided for predicting forging and pressing by machine learning.

Description

Technical Field

[0001] The present disclosure belongs to the field of equipment manufacturing technology, and in particular relates to a forging prediction model training method and device, a forging simulation method and device, an electronic device, and a computer-readable storage medium. Background Art

[0002] The forging process is a forming process that uses the hammer, anvil, punch or die of a forging machine to apply pressure to the blank to cause plastic deformation, thereby obtaining a part of the desired shape and size. The effects of forging include increasing the cross-section of the blank and smoothing the end face, improving the lateral mechanical properties of the forging and reducing anisotropy.

[0003] Due to the different shapes of the initial workpieces, the shapes obtained after forging are also different, and the stress distribution is closely related to the shape. In addition, since the mechanical analysis of the forging process involves nonlinear and transient calculations, the simulation calculation of the forging process is more time-consuming and consumes computing resources.

[0004] The information disclosed in this background technology section is only intended to enhance the understanding of the overall background of the disclosure and should not be regarded as an acknowledgement or any form of suggestion that the information constitutes the prior art already known to a person skilled in the art. Summary of the invention

[0005] The purpose of the present invention is to solve the technical problem that forging simulation of workpieces of various shapes cannot be quickly predicted, and provides a forging prediction model training method and device, a forging simulation method and device, an electronic device, and a computer-readable storage medium.

[0006] A first aspect of the present disclosure provides a forging prediction model training method, the method comprising: acquiring simulation data of a forging simulation model corresponding to a forging workpiece; dividing the forging simulation model and the space where the forging simulation model is located into multiple layers of spatial units, and making each spatial unit have two attributes, namely, geometric shape and stress; based on the simulation data, using matrices of different channels to respectively characterize the geometric shape and stress attributes of each spatial unit to obtain a sample data set; based on the sample data set, training an initial forging prediction model to obtain a trained forging prediction model.

[0007] The second aspect of the present disclosure provides a forging simulation method, which includes: obtaining a forging simulation model of a part to be forged; obtaining a single-channel forging matrix based on the forging simulation model; inputting the forging matrix into a forging prediction model to obtain a prediction matrix output by the forging prediction model; the forging prediction model is a model trained using the forging prediction model training method of the first aspect; based on the prediction matrix, obtaining the shape and stress of the part to be forged after forging.

[0008] The third aspect of the present disclosure provides a forging prediction model training device, which includes: a data acquisition unit, configured to acquire simulation data of a forging simulation model corresponding to a forging workpiece; a division unit, configured to divide the forging simulation model and the space where the forging simulation model is located into multiple layers of spatial units, and to enable each spatial unit to have two properties: geometric shape and stress; a characterization unit, based on the simulation data, using matrices of different channels to respectively characterize the geometric shape and stress properties of each spatial unit to obtain a sample data set; a training unit, configured to train the initial forging prediction model based on the sample data set to obtain a trained forging prediction model.

[0009] The fourth aspect of the present disclosure provides a forging simulation device, which includes: a model acquisition unit, configured to acquire a forging simulation model of a part to be forged; a matrix acquisition unit, configured to obtain a single-channel forging matrix based on the forging simulation model; a prediction unit, configured to input the forging matrix into a forging prediction model to obtain a prediction matrix output by the forging prediction model; the forging prediction model is a model trained by the forging prediction model training device as in the third aspect; and a result acquisition unit, configured to obtain the shape and stress of the part to be forged after forging based on the prediction matrix.

[0010] A fifth aspect of the present disclosure provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method described in any implementation of the first aspect or the second aspect.

[0011] A sixth aspect of the present disclosure provides a non-transitory computer-readable storage medium storing computer instructions, where the computer instructions are used to cause a computer to execute the method described in any implementation of the first aspect or the second aspect.

[0012] The forging prediction model training method and device provided by the embodiments of the present disclosure first obtain simulation data of the forging simulation model corresponding to the forging workpiece; secondly, divide the forging simulation model and the space where the forging simulation model is located into multi-layer spatial units, and make each spatial unit have two attributes of geometric shape and stress; thirdly, based on the simulation data, use matrices of different channels to respectively characterize the geometric shape and stress attributes of each spatial unit to obtain a sample data set; finally, based on the sample data set, train the initial forging prediction model to obtain a trained forging prediction model, thereby, after dividing the spatial units, set the geometric shape and stress attributes for the spatial units, and use matrices of different channels to respectively characterize the geometric shape attributes and stress attributes of the spatial units, so that the forging prediction model can predict the stress distribution of fixed-shape forgings under these two attributes, which provides a reliable means for using the model to predict the stress of forgings and improves the reliability of forging prediction.

[0013] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 is a flow chart of an embodiment of a forging prediction model training method disclosed herein;

[0015] Figure 2 is a schematic representation of the geometry and stress properties of the present disclosure;

[0016] Figure 3a It is a structural schematic diagram of the forging simulation model disclosed in the present invention;

[0017] Figure 3b Yes Figure 3a The schematic diagram of the structure obtained after taking one eighth;

[0018] Figure 4 is a flow chart of an embodiment of a forging simulation method according to the present disclosure;

[0019] Figure 5 It is a schematic diagram of the relationship between the forging matrix, the forging prediction model and the prediction matrix disclosed in the present invention;

[0020] Figure 6 is a structural schematic diagram of an embodiment of a forging prediction model training device disclosed in the present invention;

[0021] Figure 7 is a structural schematic diagram of an embodiment of a forging simulation device disclosed in the present invention;

[0022] Figure 8It is a block diagram of an electronic device used to implement the forging prediction model training method of the embodiment of the present disclosure. DETAILED DESCRIPTION

[0023] Unless explicitly stated otherwise, throughout the specification and claims, the term “comprise” or variations such as “include” or “comprising”, etc., will be understood to include the stated elements or components but not to exclude other elements or components.

[0024] The technical solution of the present disclosure is described below by means of specific embodiments. It should be understood that one or more steps mentioned in the present disclosure do not exclude the existence of other methods and steps before and after the combination step, or other methods and steps may be inserted between these explicitly mentioned steps. It should also be understood that these examples are only used to illustrate the present disclosure and are not used to limit the scope of the present disclosure. Unless otherwise specified, the numbering of each method step is only for the purpose of identifying each method step, and does not limit the order of arrangement of each method or limit the scope of implementation of the present disclosure. Changes or adjustments in their relative relationships can also be regarded as the scope of implementation of the present disclosure without substantial changes in the technical content.

[0025] The sources of the raw materials and instruments used in the examples are not particularly limited and can be purchased from the market or prepared according to conventional methods known to those skilled in the art.

[0026] Forging is a very common process that uses the hammer, anvil, punch or die of a forging machine to apply pressure to the blank to cause plastic deformation, thereby obtaining a forming method of the desired shape and size of the product. The effects of forging include increasing the cross section of the blank and flattening the end face, improving the transverse mechanical properties of the forging and reducing anisotropy.

[0027] Since the initial workpiece has a different shape, the shape obtained after forging is also different, and the stress distribution is closely related to the shape. In addition, since the mechanical analysis of the simulated forging process involves nonlinear and transient calculations, the simulation calculation of the forging process is more time-consuming and consumes computing resources. The simulation analysis is time-consuming and labor-intensive, which makes the need for rapid simulation using machine learning more urgent. The shape after forging will change, and the three-dimensional shape needs to be predicted. The existing three-dimensional shape is mainly predicted by the three-dimensional coordinates of fixed grid nodes, and there is no precedent for the rapid prediction of three-dimensional shapes. The existing rapid simulation mainly predicts the stress distribution of forgings with fixed shapes, and changes the pressure or pressing speed to obtain the corresponding stress. The above methods cannot be used for rapid simulation of forgings with changing shapes.

[0028] In view of the defects in the prior art, the present invention provides a forging prediction model training method, which is based on the three-dimensional geometry of convolutional neural networks and the stress distribution prediction method thereon. By discretizing the space into a two-dimensional matrix, the geometric shape and stress are characterized in different matrix channels, and then predicted and restored. In this way, it can be used for workpieces of various shapes. Unlike the grid division in simulation, this method uses the same discretization method for any geometric structure. That is, the process of cutting the rectangular space into a finite number of small rectangular blocks, each of which has two attributes, including object attributes and stress value attributes. Figure 1 A process 100 of an embodiment of a forging prediction model training method is shown. The forging prediction model training method includes the following steps:

[0029] Step 101, obtaining simulation data of a forging simulation model corresponding to a forging workpiece.

[0030] In this embodiment, the forging simulation model is a model established to simulate the forging process of a forging workpiece. Specifically, the process of obtaining the forging simulation model is as follows: use conventional modeling software for forging workpieces to model the forging workpiece, set material parameters in the modeling software, set the upper and lower dies of the forging workpiece, set the rigidity properties, load the forging load and boundary conditions, set the load step, and obtain the forging simulation model.

[0031] In this embodiment, the simulation data simulates the data of the forging state of the forged workpiece. Specifically, the simulation data includes: the model data of the forged workpiece before forging, and the shape and stress of the forged workpiece after forging obtained by solving the model data. After the forging simulation model is constructed, the shape and stress distribution can be obtained by solving the forging results.

[0032] In this embodiment, in order to obtain a sample data set of a forging prediction model, the forging workpiece may be a variety of workpieces, and corresponding sample data may be generated for each workpiece. The sample data set is obtained by combining the sample data of all workpieces in the geometric forging workpiece.

[0033] Step 102: divide the forging simulation model and the space where the forging simulation model is located into multi-layer space units, and make each space unit have two attributes: geometric shape and stress.

[0034] In this embodiment, the space corresponding to the forging simulation model discretizes the space where the forging simulation model is located into multiple two-dimensional matrices, that is, by generating multiple layers and splicing the matrices of the multiple layers, a large two-dimensional matrix is ​​generated. The two-dimensional matrix has two channels, namely, Figure 2 Schematic representation of the geometry and stress properties shown, Figure 2 In FIG. 1 , the geometric shape attribute J is a physical attribute that characterizes the forging simulation model, and the stress attribute Y is an attribute that characterizes the stress distribution of the forging simulation model.

[0035] In this embodiment, Figure 3a and Figure 3b is a cross-sectional diagram of the forging simulation model, which includes: prediction space K, upper die M1, lower die M2 and forging workpiece G. The upper die M1 and the lower die M2 are rigid bodies and will not deform. Since forging is mostly a symmetrical model, in order to save calculation, one eighth of the entire forging simulation model is taken for analysis and prediction, which is Figure 3b The black dotted box is shown. The forging simulation model is divided into 100 layers, each layer is divided into 768*854 pixels, and each pixel has two attributes. Therefore, it can be understood that the matrix form of the entire space is as follows {2, {768, 854, 100}}, where 2 represents two channels, {768, 854, 100}-dimensional space pixels. If it is a simpler model, the pixel value can be adjusted appropriately, such as reducing it to {200, 200, 50}, which can greatly reduce the amount of calculation.

[0036] In this embodiment, the space unit is the smallest unit obtained after dividing the forging simulation model and the space where the forging simulation model is located. Specifically, the forging simulation model can be divided according to development requirements, for example, the forging simulation model is divided into multiple layers of space units horizontally, and each layer can have one or more space units. For another example, the forging simulation model is divided into multiple layers of space units vertically, and each layer can have one or more space units.

[0037] In this embodiment, the space corresponding to the forging simulation model is divided into spatial units located in multiple layers, and the gridding of the space corresponding to the forging simulation model is realized, thereby replacing the original workpiece grid division. In this way, the number of spatial units in the space is fixed, and the prediction of forging processing results for workpieces with various structures is realized.

[0038] Step 103 , based on the simulation data, matrices of different channels are used to respectively characterize the geometric shape and stress properties of each spatial unit to obtain a sample data set.

[0039] In this embodiment, the simulation data includes the first model data of the forged workpiece before forging and the second model data and stress data after forging. The first model data is used to characterize the shape of the forged workpiece before forging and the positional relationship between the forged workpiece and other objects (such as dies located above and below the forged workpiece), and the second model data is used to characterize the shape of the forged workpiece after forging and the positional relationship between the forged workpiece and other objects. Generally, the simulation data is three-dimensional data, while the sample data set is two-dimensional data. A certain algorithm can be used to convert the simulation data into values ​​on matrices of different channels that characterize the geometric shape and stress properties of each spatial unit.

[0040] In this embodiment, the sample data set is the data for training the forging prediction model, the forged workpiece is the workpiece that forms the sample data set, the workpiece to be forged is the workpiece in the state after unsimulated forging, and the sample data set includes: an input matrix of specific values ​​of the geometric shape attributes of the forged workpiece before forging (corresponding to the forging matrix of the workpiece to be forged), an output matrix of specific values ​​of the geometric shape attributes and specific values ​​of the stress attributes of the forged workpiece after forging, and the output matrix is ​​also the true value matrix that the forging prediction model needs to predict. The forging prediction model is a model that predicts the state of the forged workpiece after forging. Through the forging matrix, the forging prediction model can determine the characterization of the geometric shape of the forged workpiece before forging; through the truth matrix, the prediction matrix obtained by the forging prediction model in the training process can be gradually close to the truth matrix.

[0041] Step 104 , based on the sample data set, the initial forging prediction model is trained to obtain a trained forging prediction model.

[0042] In this embodiment, the forging prediction model can adopt the UNetEx model architecture, and the forging prediction model consists of two main parts: the encoder and the decoder. The encoder extracts the features of the input data (i.e., the forging matrix) by downsampling layer by layer, and makes a jump connection with the corresponding layer of the decoder at each layer to retain the detail information. The decoder gradually restores the spatial resolution of the data, and uses the features extracted by the encoder for accurate prediction, giving a prediction result, which is a two-dimensional matrix, i.e., a prediction matrix including the geometric shape channel and the stress channel of the object.

[0043] In this embodiment, the forging matrix is ​​a matrix input into the forging prediction model reflecting the data of the forged workpiece before forging, and the prediction matrix is ​​a matrix output by the forging prediction model reflecting the shape and stress distribution of the forged workpiece after forging.

[0044] In this embodiment, during each iterative training of the forging prediction model, sample data is selected from the sample data set, and the selected sample data is input into the forging prediction model, and the loss value of the forging prediction model is calculated based on the prediction result output by the forging prediction model.

[0045] In this embodiment, the loss function of the forging prediction model can adopt the mean square error function. The mean square error (MSE) is a measure of the difference between the estimator and the estimated quantity. The specific function of the mean square error is shown in formula (1). The smaller the value of the mean square error function MSE, the better the prediction effect of the forging prediction model.

[0046]

[0047] In formula (1), n ​​represents the amount of data, Y iRepresents the true value, which is also the estimated value. Represents the predicted value, which is also an estimate.

[0048] Optionally, the loss function of the forging prediction model can also adopt a weighted mean square error function. The loss functions implemented by the average error function and the weighted mean square error function can both handle the problem of uneven stress distribution, thereby improving the reliability of the forging prediction model.

[0049] In this embodiment, the Adam optimizer can be used to adjust the model parameters. The Adam optimizer has the characteristics of adaptive learning rate, which helps to accelerate convergence. The specific parameter adjustment process is: first, the initial parameters of the forging prediction model are generated immediately, and then the prediction results of the forging prediction model are compared with the actual data, and the loss function value is calculated. Generally, the loss function is the mean square error function. By calculating the gradient and momentum, etc., the new hyperparameters are solved, the number of iterative predictions (can be set based on development needs, such as 1000 times), and the final result is used as the optimal model parameters.

[0050] In this embodiment, a learning rate decay strategy may also be adopted to gradually reduce the learning rate of the forging prediction model to improve the stability and accuracy of the forging prediction model in the later stage of training.

[0051] For the forging process, the present invention proposes to use a deep learning algorithm to predict the shape and stress distribution after forging, and proposes two channels for characterization, one is the object property channel and the other is the stress property channel. The forging prediction model can be used to achieve better shape and stress prediction.

[0052] The forging prediction model training method provided by the present disclosure first obtains the simulation data of the forging simulation model corresponding to the forging workpiece; secondly, divides the forging simulation model and the space where the forging simulation model is located into multi-layer space units, and makes each space unit have two attributes of geometric shape and stress; then, based on the simulation data, uses matrices of different channels to respectively characterize the geometric shape and stress attributes of each space unit to obtain a sample data set; finally, based on the sample data set, trains the initial forging prediction model to obtain a trained forging prediction model. Thus, after dividing the space units, the geometric shape and stress attributes are set for the space units, and the matrices of different channels are used to respectively characterize the geometric shape attributes and stress attributes of the space units, so that the forging prediction model can predict the stress distribution of fixed-shape forgings under these two attributes, providing a reliable means for using the model to predict the stress of forgings, and improving the reliability of forging prediction.

[0053] In some optional implementations of the present disclosure, the above-mentioned simulation data includes: first model data of the forged workpiece before forging, second model data and stress data of the forged workpiece after forging. Based on the simulation data, matrices of different channels are used to respectively characterize the geometric shape and stress properties of each spatial unit, and the sample data set is obtained, including: performing a position matrix transformation on the first model data of each spatial unit to obtain a first three-dimensional matrix, performing a two-dimensional transformation on the first three-dimensional matrix to obtain a single-channel input matrix; performing a position matrix transformation on the second model data of each spatial unit to obtain a second three-dimensional matrix, performing a two-dimensional transformation on the second three-dimensional matrix and the corresponding stress data to obtain a dual-channel output matrix; combining the input matrix and the output matrix to obtain the sample data set.

[0054] In this optional implementation, the simulation data is generally three-dimensional data, and the sample data in the sample data set is two-dimensional data. Through position matrix transformation and stress data addition, the simulation data can be converted into a sample data set of the forging workpiece that is easy to understand by the forging prediction model and can characterize the forging workpiece.

[0055] In this optional implementation, the first model data of the forged workpiece before forging reflects the relationship between the forged workpiece and the die in the forging simulation model before forging, and specifically can reflect the shape data of the forged workpiece before forging and the forged workpiece (at this time, the stress of the forged workpiece is 0). The second model data of the forged workpiece after forging reflects the relationship between the forged workpiece and the die in the forging simulation model after forging, and the stress data of the forged workpiece after forging is the stress value simulated after forging.

[0056] In this optional implementation, the position matrix transformation refers to the transformation of the first model data of the forging workpiece before forging into a three-dimensional matrix according to a certain algorithm. The three-dimensional matrix is ​​organized according to certain rules and can also be restored to the original three-dimensional first model data before forging.

[0057] In this optional implementation, the two-dimensional transformation is a transformation that converts three-dimensional data into two-dimensional data. The first three-dimensional matrix, the second three-dimensional matrix, and the stress data are transformed in the same way. The two-dimensional transformation can be achieved by using a conventional three-dimensional to two-dimensional method. For example, in Python, the reshape() function in the NumPy library can be used to convert three-dimensional data into two-dimensional data. Specifically, assuming there is a three-dimensional array with a shape of (m, n, p), it can be converted into a two-dimensional array with a shape of (m*n, p) by using the reshape() function.

[0058] In this optional implementation, the matrix obtained after the spatial unit performs position matrix transformation has a corresponding relationship with the stress data of the forged workpiece after forging. For this purpose, the second three-dimensional matrix and the corresponding stress data can be simultaneously subjected to two-dimensional changes to obtain a dual-channel output matrix corresponding to the second three-dimensional matrix and stress data respectively.

[0059] In this optional implementation, the sample data set includes multiple sample data, each sample data includes an input matrix and an output matrix, the input matrix is ​​a single-channel matrix reflecting the shape data before forging, and the output matrix is ​​a dual-channel matrix reflecting the shape data and stress data after forging. The input matrix of the sample data is input into the forging prediction model, so that the forging prediction model predicts the input matrix, and the prediction matrix gradually approaches the output matrix corresponding to the input matrix, so that the forging prediction model can realize the prediction of the forging process of the forging workpiece.

[0060] The method for obtaining a sample data set provided by this optional implementation manner performs a position matrix transformation on the first model data of each spatial unit to obtain a first three-dimensional matrix, performs a two-dimensional transformation on the first three-dimensional matrix to obtain a single-channel input matrix; performs a position matrix transformation on the second model data of each spatial unit to obtain a second three-dimensional matrix, performs a two-dimensional transformation on the second three-dimensional matrix and corresponding stress data to obtain a dual-channel output matrix, thereby improving the reliability and accuracy of obtaining the sample data set.

[0061] In some optional implementations of the present disclosure, the above-mentioned position matrix transformation of the first model data of each spatial unit to obtain the first three-dimensional matrix includes: determining the domain of the forging simulation model based on the first model data of the forging workpiece before forging; traversing all spatial units in the domain and judging the position of each spatial unit relative to the forging simulation model; and obtaining the first three-dimensional matrix based on the position.

[0062] In this optional implementation, the forging simulation model is a model including a forging workpiece and upper and lower dies, and the data structure and geometric information of the forging simulation model are accessed by loading the file of the forging simulation model (such as a .vtk file).

[0063] A cuboid (or cube) domain is generated according to the boundary of the forging simulation model, and this domain is defined by calculating the bounding box of the forging simulation model. Specifically, the boundary is a cuboid domain generated according to the maximum x, y, and z coordinates of the model, which ensures that the model is definitely within the domain. The bounding box is a cuboid space generated with the maximum xyz coordinates. This space is of fixed size in all models. For example, if there are 100 simulation models, it should be ensured that this domain is applicable to these 100 models, and these 100 models all use this domain.

[0064] In this example, we need to decide how to discretize the domain of the cuboid into a fixed number of spatial units, which can be generated using a uniform grid. The generate_grid_points function uses numpy.linspace to generate uniformly distributed spatial units in each dimension of the cuboid. np.meshgrid is used to generate a 3D grid, which is then flattened into a list of spatial units using .reshape(-1,3).

[0065] Then, determine one by one whether these spatial units are in the predicted space K, the upper die M1, the lower die M2, and the forging workpiece G. The model is a solid, not hollow, so in the model is on the model. Specifically, first convert the forging simulation model file to be judged into a .brep file, and then use the BRepClass3d_SolidClassifier function to make a judgment. BRepClass3d_SolidClassifier is a class in the OpenCASCADE Technology Library (OCCT), which is mainly used to determine the positional relationship between spatial units and three-dimensional geometric bodies, such as determining whether a point is inside, outside, or on the boundary of a solid, so as to obtain the position of each spatial unit relative to the forging simulation model.

[0066] The method for obtaining a single-channel input matrix provided by this optional implementation method determines the domain of the forging simulation model based on the first model data of the forging workpiece before forging; in the domain, traverses all spatial units and determines the position of each spatial unit relative to the forging simulation model; based on the position, obtains a first three-dimensional matrix, thereby converting the actual simulation data into a single-channel input matrix that can characterize the simulation data and can be recognized by the forging prediction model, thereby improving the convenience and reliability of data acquisition.

[0067] Optionally, the above-mentioned position matrix transformation of each spatial unit based on the first model data of the forged workpiece before forging to obtain a single-channel input matrix includes: traversing all spatial units, determining the position relationship of the spatial unit relative to the forging simulation model, determining the shape and position data of each spatial unit based on the position relationship and the first model data of the forged workpiece before forging, and converting the shape data of each spatial unit into single-channel matrix data according to a predetermined three-dimensional to two-dimensional algorithm to obtain a single-channel input matrix.

[0068] In this embodiment, the transformation form of the second three-dimensional matrix obtained by performing position matrix transformation on the second model data of each spatial unit is the same as the transformation form of the first three-dimensional matrix obtained by performing position matrix transformation on the first model data of each spatial unit.

[0069] Optionally, the spatial unit can be a point, and the above-mentioned position matrix transformation of each spatial unit includes: The judgment process is mainly: 1. Import a geometric object. 2. Traverse each corresponding point in the space. 3. Use the BRepBuilderAPI_MakeVertex function in the topological algorithm to create a vertex object, and add the above-mentioned point object to the model. 4. Use the BRepClass3d_SolidClassifier function in the geometric algorithm, and pass the model and point object as parameters to it. 5. Call the lsinside method of the SoidClassifier function to check whether the point is inside the model. If the return value is TopAbs_IN, it means that the point is inside the model; if the return value is TopAbs_OUT, it means that the point is outside the model.

[0070] Through the above operations, the points on the forging model can be marked as 1, the points on the die of the forging simulation model can be marked as 10, and the points in the space can be marked as 100. After marking, it is restored to the original two-dimensional matrix. In this way, in the two-dimensional matrix of all slices, the data values ​​are composed of 1, 10, and 100. This marking method helps to clearly display the boundaries and boundaries of objects when the data is reduced. Specifically, this marking method helps to highlight the differences between the model, the die, and the surrounding space, so that the boundaries of the object can be more obvious and clear when the image is reduced or processed. This practice can ensure better visual separation and identification of different areas.

[0071] In some optional implementations of the present disclosure, the above-mentioned training of the initial forging prediction model based on the sample data set to obtain the trained forging prediction model includes: selecting sample data from the sample data set; inputting the sample data into the initial forging prediction model to obtain the prediction result of the forging prediction model; calculating the loss of the forging prediction model based on the prediction result and the sample data; and obtaining the trained forging prediction model in response to determining that the forging prediction model meets the training completion condition based on the loss.

[0072] In this optional implementation, the data before forging in the simulation data will be generated into a single-channel input matrix x, and the data after forging in the simulation data will be generated as follows Figure 2 The output matrix y of the two attributes (i.e., two channels) shown in FIG. 1 is a matrix of the output matrix y of the two attributes (i.e., two channels) shown in FIG. 1 , and the two matrices constitute a set of sample data {x, y}. Multiple sets of such sample data constitute a sample data set. Standardizing or normalizing the sample data set can improve the training effect. The sample data set is divided into a training set, a validation set, and a test set. Usually, the training set is used to train the forging prediction model, the validation set is used to adjust the hyperparameters of the forging prediction model and prevent overfitting, and the test set is used to finally evaluate the performance of the forging prediction model.

[0073] In this embodiment, the sample data includes an input matrix and an output matrix. When sample data is selected from the sample data set to train the forging prediction model, the forging prediction model can predict the forging state of the input matrix to obtain a prediction result, wherein the prediction result is the prediction result of the forging prediction model corresponding to the output matrix. When the difference between the prediction result and the output matrix is ​​smaller, it means that the loss of the forging prediction model is smaller and the prediction result of the forging prediction model is more accurate.

[0074] In this optional implementation, the loss of the forging prediction model can be calculated by the loss function set for the forging simulation model. The loss function is used to characterize the parameters of the prediction results and the output matrix. The prediction results and the output matrix are input into the loss function to obtain the loss of the forging simulation model. It should be noted that the output matrix is ​​the true value in the sample data, which provides a training target for the forging prediction model.

[0075] In this optional implementation, the training completion condition includes: the loss value is less than the loss threshold value. When the loss of the forging prediction model meets the training completion condition, it is determined that the forging prediction model training is completed.

[0076] The method for training a forging prediction model provided by this optional implementation method selects sample data from a sample data set; inputs the sample data into an initial forging prediction model to obtain a prediction result of the forging prediction model; calculates the loss of the forging prediction model based on the prediction result and the sample data; and in response to determining that the forging prediction model meets a training completion condition based on the loss, obtains a trained forging prediction model, thereby providing a reliable training basis for the training of the forging prediction model through the training completion condition and the loss, thereby improving the reliability of the forging prediction model training.

[0077] With respect to the forging prediction model obtained by the forging prediction model training method, the present disclosure also provides a method for using the forging prediction model, which is also a forging simulation method. Figure 4 A process 400 of an embodiment of a forging simulation method is shown, and the forging simulation method includes the following steps:

[0078] Step 401, obtaining a forging simulation model of a part to be forged.

[0079] In this embodiment, the workpiece to be forged is a workpiece to be forged. A forging simulation model can be obtained by setting a forging die on the workpiece to be forged and simulating the state of the workpiece to be forged in the forging die. It should be noted that the forging simulation model only establishes the spatial position relationship between the workpiece to be forged and the die, and does not simulate the data of the workpiece after forging.

[0080] Step 402, obtaining a single-channel forging matrix based on the forging simulation model.

[0081] like Figure 5 As shown, the forging matrix D is a single-channel matrix representing the geometry of the part to be forged.

[0082] In this embodiment, for the forging parts whose shapes are inconsistent before and after forging, the forging simulation model can be represented by a single-channel forging matrix, and the space corresponding to the forging simulation model can be gridded, thereby replacing the original workpiece grid division. In this way, the number of spatial units in the space is fixed, and the forging processing results of workpieces with various structures can be predicted.

[0083] Step 403, inputting the forging matrix into the forging prediction model to obtain the prediction matrix output by the forging prediction model.

[0084] In this embodiment, the forging prediction model is a model trained by using any one of the forging prediction model training methods of claims 1-4.

[0085] like Figure 5 In the forging matrix D, the forging prediction model X is input to obtain the prediction matrix Y. The prediction matrix Y is a matrix with two channels, and the data in each channel is also represented by a matrix.

[0086] Step 404, based on the prediction matrix, obtain the shape and stress of the forged part after forging.

[0087] In this embodiment, the prediction matrix is ​​a two-dimensional matrix including two channels: a geometry channel and a stress channel.

[0088] In actual use, after obtaining the prediction matrix, the geometric model can be used to restore the prediction matrix into three-dimensional geometric data corresponding to the simulation data according to a predetermined algorithm.

[0089] The forging simulation method provided in this embodiment first obtains the forging simulation model of the part to be forged; secondly, based on the forging simulation model, obtains the forging matrix of a single channel; thirdly, inputs the forging matrix into the forging prediction model to obtain the prediction matrix output by the forging prediction model; finally, based on the prediction matrix, obtains the shape and stress of the part to be forged after forging. Thus, there is no need to calculate the shape and stress of the part to be forged by the forging simulation model to be forged, and only the forging matrix of the forging simulation model to be forged is predicted by the trained forging prediction model, which improves the efficiency of obtaining the forging results and improves the reliability and accuracy of obtaining the shape and stress of the part to be forged after forging.

[0090] In some optional implementations of the present disclosure, the forging matrix of a single channel is obtained based on the forging simulation model, including: determining each spatial unit based on the forging simulation model; performing position matrix transformation on each spatial unit to obtain the forging matrix of a single channel.

[0091] In this optional implementation, the method of determining the spatial units and performing position matrix transformation on each spatial unit can refer to the method of dividing the spatial units and performing position matrix transformation on the spatial units in the forging prediction model training method, which will not be repeated here.

[0092] The method for obtaining a single-channel forging matrix provided by this optional implementation method determines each spatial unit based on a forging simulation model; performs position matrix transformation on each spatial unit to obtain a single-channel forging matrix, and transforms the forging simulation model into a matrix that can be understood by a forging prediction model, thereby improving the reliability of the forging data prediction of the forging parts to be forged.

[0093] In some optional implementations of the present disclosure, the above-mentioned prediction matrix includes: a geometric shape matrix and a stress distribution matrix. Based on the prediction matrix, the shape and stress of the part to be forged after forging are obtained, including: restoring the prediction matrix to a three-dimensional space according to the spatial cutting information corresponding to the prediction matrix; in the three-dimensional space, using the data in the geometric shape matrix to reconstruct the three-dimensional structure of the part to be forged; using the stress distribution matrix to reconstruct the stress distribution of the part to be forged so that the stress is distributed on the three-dimensional structure.

[0094] In this optional implementation, after prediction, two two-dimensional matrices are obtained: one represents the discretization of the geometric shape, and the other represents the predicted result of the stress distribution. In order to restore these two-dimensional matrices to the original three-dimensional object, the following steps need to be followed:

[0095] First, the predicted two-dimensional matrix is ​​spliced ​​according to the original discretization method. Specifically, according to the spatial cutting information corresponding to each two-dimensional matrix (such as the position and attributes of the small cuboid), the matrix information is restored to the three-dimensional space layer by layer. Each element of each two-dimensional matrix corresponds to a small cuboid in the three-dimensional space, and its attributes can be reconstructed by stacking layer by layer.

[0096] Next, the data in the geometry matrix is ​​used to reconstruct the structure of the 3D object, which includes restoring the geometric outline and volume of the object. The stress distribution matrix is ​​used to distribute stress information on the reconstructed 3D object, ensuring that the predicted stress distribution is correctly mapped to each small cuboid. This process may require interpolation techniques to smooth and adjust the predicted values ​​to ensure the continuity and accuracy of stress on the actual 3D object.

[0097] Finally, the restored 3D object and stress distribution are checked through visualization tools to ensure that they are consistent with the actual geometric structure and stress distribution. These steps ensure that the restoration process from 2D matrix to 3D object can accurately reflect the predicted results and provide effective 3D data support for practical applications.

[0098] This optional implementation provides a method for obtaining the shape and stress of the part to be forged after forging, which restores the prediction matrix to three-dimensional space. In the three-dimensional space, the data of the geometric shape matrix of the prediction matrix is ​​reconstructed into the three-dimensional structure of the part to be forged, and the stress distribution matrix is ​​reconstructed into the stress distribution of the part to be forged, which provides a reliable implementation method for obtaining the forging stress and geometric shape of the part to be forged, and improves the reliability of forging simulation.

[0099] Further references Figure 6 As an implementation of the forging prediction model training method shown in the above figures, the present disclosure provides an embodiment of a forging prediction model training device. Figure 1 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0100] like Figure 6 As shown, the forging prediction model training device 600 provided in this embodiment includes: a data acquisition unit 601, a division unit 602, a characterization unit 603, and a training unit 604. Among them, the above-mentioned data acquisition unit 601 can be configured to obtain simulation data of the forging simulation model corresponding to the forging workpiece. The above-mentioned division unit 602 can be configured to divide the forging simulation model and the space where the forging simulation model is located into multiple layers of spatial units, and make each spatial unit have two properties of geometric shape and stress. The above-mentioned characterization unit 603 can, based on the simulation data, use matrices of different channels to respectively characterize the geometric shape and stress properties of each spatial unit to obtain a sample data set. The above-mentioned training unit 604 can be configured to train the initial forging prediction model based on the sample data set to obtain a trained forging prediction model.

[0101] In this embodiment, in the forging prediction model training device 200, the specific processing of the data acquisition unit 601, the division unit 602, the characterization unit 603, and the training unit 604 and the technical effects thereof can be referred to respectively. Figure 1 The relevant descriptions of step 101, step 102, step 103, and step 104 in the corresponding embodiment are not repeated here.

[0102] In some optional implementations of the present disclosure, the above-mentioned simulation data includes: first model data of the forged workpiece before forging, second model data and stress data of the forged workpiece after forging, and the above-mentioned characterization unit 603 is configured to: perform position matrix transformation on the first model data of each spatial unit to obtain a first three-dimensional matrix, perform two-dimensional transformation on the first three-dimensional matrix to obtain a single-channel input matrix; perform position matrix transformation on the second model data of each spatial unit to obtain a second three-dimensional matrix, perform two-dimensional transformation on the second three-dimensional matrix and the corresponding stress data to obtain a dual-channel output matrix; combine the input matrix and the output matrix to obtain a sample data set.

[0103] In some optional implementations of the present disclosure, the characterization unit 603 is further configured to: determine the domain of the forging simulation model based on the first model data of the forging workpiece before forging; traverse all spatial units in the domain and determine the position of each spatial unit relative to the forging simulation model; and obtain a first three-dimensional matrix based on the position.

[0104] In some optional implementations of the present disclosure, the training unit 604 is further configured to: select sample data from a sample data set; input the sample data into an initial forging prediction model to obtain a prediction result of the forging prediction model; calculate the loss of the forging prediction model based on the prediction result and the sample data; in response to determining that the forging prediction model meets the training completion condition based on the loss, obtain a trained forging prediction model.

[0105] In the forging prediction model training device provided in the embodiment of the present disclosure, first, the data acquisition unit 601 acquires the simulation data of the forging simulation model corresponding to the forging workpiece; secondly, the division unit 602 divides the forging simulation model and the space where the forging simulation model is located into multi-layer space units, and makes each space unit have two attributes of geometric shape and stress; then, the characterization unit 603 can use matrices of different channels to respectively characterize the geometric shape and stress attributes of each space unit based on the simulation data to obtain a sample data set; finally, the training unit 604 trains the initial forging prediction model based on the sample data set to obtain a trained forging prediction model. Thus, after dividing the space unit, the geometric shape and stress attributes are set for the space unit, and the matrices of different channels are used to respectively characterize the geometric shape attributes and stress attributes of the space unit, so that the forging prediction model can predict the stress distribution of the fixed shape forging under these two attributes, which provides a reliable implementation means for using the model to predict the stress of the forging, and improves the reliability of the forging prediction.

[0106] Further references Figure 7 As an implementation of the forging prediction model training method shown in the above figures, the present disclosure provides an embodiment of a forging prediction model training device. Figure 1 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0107] like Figure 7As shown, the forging simulation device 700 provided in this embodiment includes: a model acquisition unit 701, a matrix acquisition unit 702, a prediction unit 703, and a result acquisition unit 704. Among them, the above-mentioned model acquisition unit 701 can be configured to obtain a forging simulation model of the part to be forged. The above-mentioned matrix acquisition unit 702 can be configured to obtain a single-channel forging matrix based on the forging simulation model. The above-mentioned prediction unit 703 can be configured to input the forging matrix into the forging prediction model to obtain a prediction matrix output by the forging prediction model; the forging prediction model is a model trained by the forging prediction model training device of the above-mentioned embodiment. The above-mentioned result acquisition unit 704 can be configured to obtain the shape and stress of the part to be forged after forging based on the prediction matrix.

[0108] In this embodiment, in the forging simulation device 700, the specific processing of the model acquisition unit 701, the matrix acquisition unit 702, the prediction unit 703, and the result acquisition unit 704 and the technical effects thereof can be referred to respectively. Figure 5 The relevant descriptions of step 501, step 502, step 503, and step 504 in the corresponding embodiment are not repeated here.

[0109] In some optional implementations of the present disclosure, the matrix obtaining unit 702 is configured to: determine each spatial unit based on a forging simulation model; and perform position matrix transformation on each spatial unit to obtain a single-channel forging matrix.

[0110] In some optional implementations of the present disclosure, the above-mentioned prediction matrix includes: a geometric shape matrix and a stress distribution matrix, and the above-mentioned result obtaining unit 704 is configured to: restore the prediction matrix to a three-dimensional space according to the spatial cutting information corresponding to the prediction matrix; in the three-dimensional space, use the data in the geometric shape matrix to reconstruct the three-dimensional structure of the part to be forged; use the stress distribution matrix to reconstruct the stress distribution of the part to be forged so that the stress is distributed on the three-dimensional structure.

[0111] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

[0112] Figure 8A schematic block diagram of an example electronic device 800 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their modes are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0113] like Figure 8 As shown, the device 800 includes a computing unit 801, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 into a random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the device 800 can also be stored. The computing unit 801, the ROM 802, and the RAM 803 are connected to each other via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0114] A number of components in the device 800 are connected to the I / O interface 805, including: an input unit 806, such as a keyboard, a mouse, etc.; an output unit 807, such as various types of displays, speakers, etc.; a storage unit 808, such as a disk, an optical disk, etc.; and a communication unit 809, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 809 allows the device 800 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0115] The computing unit 801 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 801 performs the various methods and processes described above, such as a forging prediction model training method or a forging simulation method. For example, in some embodiments, the forging prediction model training method or the forging simulation method may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 808. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 800 via the ROM 802 and / or the communication unit 809. When the computer program is loaded into the RAM 803 and executed by the computing unit 801, one or more steps of the forging prediction model training method or the forging simulation method described above may be performed. Alternatively, in other embodiments, the computing unit 801 may be configured to execute the forging prediction model training method or the forging simulation method in any other appropriate manner (eg, by means of firmware).

[0116] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0117] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable forging prediction model training device or forging simulation device, so that the program code, when executed by the processor or controller, implements the mode / operation specified in the flow chart and / or block diagram. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0118] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0119] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0120] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0121] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0122] The foregoing description of specific exemplary embodiments of the present disclosure is for the purpose of illustration and demonstration. These descriptions are not intended to limit the present disclosure to the precise form disclosed, and it is clear that many changes and variations can be made based on the above teachings. The purpose of selecting and describing the exemplary embodiments is to explain the specific principles of the present disclosure and its practical application, so that those skilled in the art can realize and utilize various different exemplary embodiments of the present disclosure and various different selections and changes. The scope of the present disclosure is intended to be defined by the claims and their equivalents.

Claims

1. A forging prediction model training method, characterized in that: The method comprises: Acquiring simulation data of a forging simulation model corresponding to the forging workpiece; Dividing the forging simulation model and the space where the forging simulation model is located into multi-layer space units, and making each space unit have two attributes of geometric shape and stress; Based on the simulation data, matrices of different channels are used to respectively characterize the geometric shape and stress properties of each spatial unit to obtain a sample data set; Based on the sample data set, the initial forging prediction model is trained to obtain a trained forging prediction model.

2. The method according to claim 1, wherein: The simulation data includes: first model data of the forged workpiece before forging, second model data and stress data of the forged workpiece after forging, and based on the simulation data, matrices of different channels are used to respectively characterize the geometric shape and stress properties of each spatial unit, and the sample data set obtained includes: Performing a position matrix transformation on the first model data of each spatial unit to obtain a first three-dimensional matrix, and performing a two-dimensional transformation on the first three-dimensional matrix to obtain a single-channel input matrix; Performing a position matrix transformation on the second model data of each spatial unit to obtain a second three-dimensional matrix, and performing a two-dimensional transformation on the second three-dimensional matrix and corresponding stress data to obtain a dual-channel output matrix; The input matrix and the output matrix are combined to obtain a sample data set.

3. The method according to claim 2, wherein: The performing position matrix transformation on the first model data of each spatial unit to obtain a first three-dimensional matrix includes: determining a domain of the forging simulation model based on first model data of the forging workpiece before forging; In the domain, traverse all spatial units and determine the position of each spatial unit relative to the forging simulation model; Based on the positions, a first three-dimensional matrix is ​​obtained.

4. The method according to claim 1, wherein: The initial forging prediction model is trained based on the sample data set to obtain a trained forging prediction model including: Select sample data from the sample data set; Inputting the sample data into an initial forging prediction model to obtain a prediction result of the forging prediction model; Calculating the loss of the forging prediction model based on the prediction result and the sample data; In response to determining that the forging prediction model meets a training completion condition based on the loss, a trained forging prediction model is obtained.

5. A forging simulation method, the method comprising: Obtaining a forging simulation model of a part to be forged; Based on the forging simulation model, a single-channel forging matrix is ​​obtained; Inputting the forging matrix into the forging prediction model to obtain the prediction matrix output by the forging prediction model; the forging prediction model is a model trained by the forging prediction model training method according to any one of claims 1 to 4; Based on the prediction matrix, the shape and stress of the forged part after forging are obtained.

6. The method according to claim 5, wherein: The forging matrix of a single channel obtained based on the forging simulation model includes: Based on the forging simulation model, determining each spatial unit; Perform position matrix transformation on each spatial unit to obtain a single-channel forging matrix.

7. The method according to claim 5, wherein: The prediction matrix includes: a geometric shape matrix and a stress distribution matrix. The shape and stress of the forged part after forging are obtained based on the prediction matrix, including: According to the spatial cutting information corresponding to the prediction matrix, the prediction matrix is ​​restored to the three-dimensional space; In the three-dimensional space, the three-dimensional structure of the forged part is reconstructed using the data in the geometric shape matrix; The stress distribution matrix is ​​used to reconstruct the stress distribution of the forged part so that the stress is distributed on the three-dimensional structure.

8. A forging prediction model training device, characterized in that: The device comprises: A data acquisition unit is configured to acquire simulation data of a forging simulation model corresponding to the forging workpiece; A dividing unit is configured to divide the forging simulation model and the space where the forging simulation model is located into multi-layer space units, and make each space unit have two attributes of geometric shape and stress; A characterization unit, based on the simulation data, uses matrices of different channels to respectively characterize the geometric shape and stress properties of each spatial unit to obtain a sample data set; The training unit is configured to train the initial forging prediction model based on the sample data set to obtain a trained forging prediction model.

9. A forging simulation device, characterized in that: The device comprises: A model acquisition unit is configured to acquire a forging simulation model of a part to be forged; The matrix obtains the unit, and the configured layer obtains the forging matrix of the single channel based on the forging simulation model; A prediction unit is configured to input the forging matrix into a forging prediction model to obtain a prediction matrix output by the forging prediction model; the forging prediction model is a model trained by the forging prediction model training device according to claim 8; The result obtaining unit is configured to obtain the shape and stress of the forged part after forging based on the prediction matrix.

10. An electronic device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 7.

Images