Polycrystalline material mechanical property prediction method and device, storage medium and equipment

Abstract

The invention discloses a polycrystalline material mechanical property prediction method and device, a storage medium and equipment, the method can improve the material property prediction precision by fusing the advantages of a convolutional neural network and a graph neural network, the convolutional neural network is responsible for capturing the shape information of crystal grains, and the graph neural network is responsible for capturing the shape information of the crystal grains. And the graph neural network considers the shape, size, orientation and grain boundary information of the grains. Through combination of the convolutional neural network and the graph neural network, information of crystal grains and crystal boundaries in the polycrystalline material can be utilized at the same time, and continuous prediction of nonlinear mechanical properties of the polycrystalline material can be realized. In addition, the properties of the base material can be input as grain attributes, so that stress prediction of polycrystalline materials under different materials is supported, and the applicability and accuracy of the model are enhanced.

Description

Technical Field

[0001] The present specification relates to the field of material science and technology, and in particular to a method, device, storage medium and equipment for predicting the mechanical properties of polycrystalline materials. Background Art

[0002] Polycrystalline materials are solid materials composed of many small crystal particles (called grains). Each grain is a single crystal, but the orientations of the grains are different, so the whole material presents a polycrystalline structure. Polycrystalline materials are very common in nature and industry, such as metals, ceramics, semiconductors, and many other types of materials. For polycrystalline materials, a variety of microstructural features (such as grain size, shape, and crystal orientation) will significantly affect their mechanical response. Therefore, predicting the macroscopic mechanical properties of polycrystalline materials based on their microstructure, that is, the nonlinear stress-strain response curve, is extremely important for guiding the production of advanced mechanical materials.

[0003] Normally, the stress of polycrystalline materials under different loads can be predicted through the data-driven proxy model of neural networks. However, due to the limitations of the neural network structure itself, the microstructure of polycrystalline materials cannot be fully expressed, that is, it is difficult to fully extract and represent the complex features of polycrystalline microstructures such as grain shape, grain orientation, and grain boundaries through neural networks, resulting in poor accuracy in predicting the stress of polycrystalline materials under different loads through neural networks.

[0004] Therefore, how to improve the accuracy of predicting the stress of polycrystalline materials under different loads through neural network models is an urgent problem to be solved. Summary of the invention

[0005] This specification provides a method, device, storage medium and equipment for predicting the mechanical properties of polycrystalline materials to partially solve the above-mentioned problems existing in the prior art.

[0006] This manual adopts the following technical solutions: This specification provides a method for predicting the mechanical properties of polycrystalline materials, including: Acquire characteristic graph data and target load of a target polycrystalline material, wherein different nodes in the characteristic graph data correspond to different grains contained in the target polycrystalline material, and for each node, if the grain corresponding to the node is adjacent to other grains through a grain boundary, the node is connected to the nodes corresponding to the other grains, and the node feature of the node includes at least one of a grain shape feature, a grain volume feature, a substrate property feature, and a grain orientation feature, wherein the grain shape feature of the node is obtained by performing convolution extraction based on a discretized structure array corresponding to the target polycrystalline material by a convolution model; Inputting the characteristic graph data and the target load into a pre-trained prediction model, so as to predict the stress of the target polycrystalline material under the target load according to the node characteristics of each node in the characteristic graph data and the connection relationship between each node and other nodes through the prediction model; The task is executed according to the stress of the target polycrystalline material under the target load.

[0007] Optionally, obtaining characteristic graph data of the target polycrystalline material specifically includes: Obtaining a three-dimensional array of the microstructure of the target polycrystalline material after discretization, wherein different elements contained in the three-dimensional array correspond to different sampling points uniformly distributed in the target polycrystalline material, and the value of each element is used to characterize the grain to which the sampling point corresponding to each element belongs; Encode each element contained in the three-dimensional array to obtain a coding feature vector array corresponding to the three-dimensional array, and input the coding feature vector array into a pre-trained convolution model, so as to convolve each coding feature vector in the coding feature vector array that matches the grain for each grain contained in the target polycrystalline material through the convolution model to obtain a grain shape feature of the grain; According to the grain shape characteristics of each grain, characteristic graph data of the target polycrystalline material is obtained.

[0008] Optionally, acquiring characteristic graph data of the target polycrystalline material according to the grain shape characteristics of each grain specifically includes: For each crystal grain included in the target polycrystalline material, determining, from the different elements included in the three-dimensional array, an element whose corresponding sampling point belongs to the crystal grain as each matching element of the crystal grain; Determining a grain volume feature of the grain according to the number of each matching element of the grain and the total number of each element included in the three-dimensional array; According to the grain shape characteristics and grain volume characteristics of each grain, characteristic graph data of the target polycrystalline material is obtained.

[0009] Optionally, the characteristic graph data and the target load are input into a pre-trained prediction model, so as to predict the stress of the target polycrystalline material under the target load according to the node characteristics of each node in the characteristic graph data and the connection relationship between each node and other nodes through the prediction model, specifically including: The characteristic graph data and the target load are input into a pre-trained prediction model, so as to determine the neighbor nodes of each node according to the connection relationship between the node and other nodes through the prediction model, and determine the aggregate node characteristics of the node according to the node characteristics of the node and the node characteristics of the neighbor nodes, and predict the stress of the target polycrystalline material under the target load according to the aggregate node characteristics of each node.

[0010] Optionally, predicting the stress of the target polycrystalline material under the target load according to the aggregate node characteristics of each node specifically includes: The aggregate node features of each node are globally pooled to fuse the aggregate node features of each node to obtain the global features of the target polycrystalline material, and the stress of the target polycrystalline material under the target load is predicted based on the global features.

[0011] Optionally, training the prediction model specifically includes: Obtaining sample characteristic diagram data and sample load of a sample polycrystalline material; Inputting the sample characteristic graph data and the sample load into a prediction model to be trained, so as to predict the stress of the sample polycrystalline material under the sample load as a predicted stress according to the node characteristics of each node in the sample characteristic graph data and the connection relationship between each node and other nodes through the prediction model to be trained; According to the deviation between the predicted stress and the actual stress of the sample polycrystalline material under the sample load, a target loss value is determined, and the prediction model to be trained is trained with minimizing the target loss value as the optimization goal to obtain a prediction model, wherein the deviation between the predicted stress and the actual stress of the sample polycrystalline material under the sample load is proportional to the target loss value.

[0012] Optionally, the method further comprises: Acquire at least one pair of stress-strain data input by a user as a target stress-strain data pair, wherein the stress-strain data pair includes a specified load and an expected stress; According to the target stress-strain data pair, a candidate polycrystalline material whose corresponding stress-strain response curve matches the target stress-strain data pair is determined from the preset candidate polycrystalline materials as a reference polycrystalline material, and the identification data of the reference polycrystalline material is recommended to the user, wherein the stress-strain response curve is determined in advance by the prediction model based on the characteristic graph data of the candidate polycrystalline material.

[0013] This specification provides a device for predicting mechanical properties of polycrystalline materials, comprising: An acquisition module is used to acquire characteristic graph data and a target load of a target polycrystalline material, wherein different nodes in the characteristic graph data correspond to different grains contained in the target polycrystalline material, and for each node, if the grain corresponding to the node is adjacent to other grains through a grain boundary, the node is connected to the nodes corresponding to the other grains, and the node feature of the node includes at least one of a grain shape feature, a grain volume feature, a substrate property feature, and a grain orientation feature, wherein the grain shape feature of the node is obtained by performing convolution extraction based on a discretized structure array corresponding to the target polycrystalline material by a convolution model; A prediction module, used for inputting the characteristic graph data and the target load into a pre-trained prediction model, so as to predict the stress of the target polycrystalline material under the target load according to the node characteristics of each node in the characteristic graph data and the connection relationship between each node and other nodes through the prediction model; An execution module is used to execute a task according to the stress of the target polycrystalline material under the target load.

[0014] This specification provides a computer-readable storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned method for predicting the mechanical properties of polycrystalline materials is implemented.

[0015] This specification provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned method for predicting the mechanical properties of polycrystalline materials when executing the program.

[0016] At least one of the above technical solutions adopted in this specification can achieve the following beneficial effects: In the method for predicting the mechanical properties of polycrystalline materials provided in the present specification, the characteristic graph data and the target load of the target polycrystalline material are first obtained. In the characteristic graph data, different nodes correspond to different grains contained in the target polycrystalline material. For each node, if the grain corresponding to the node is adjacent to other grains through a grain boundary, the node is connected to the nodes corresponding to other grains. The node characteristics of the node include: at least one of grain shape characteristics, grain volume characteristics, substrate property characteristics, and grain orientation characteristics. The grain shape characteristics of the node are obtained by convolution extraction based on a discretized structure array corresponding to the target polycrystalline material by a convolution model. The characteristic graph data and the target load are then input into a pre-trained prediction model, so that the prediction model predicts the stress of the target polycrystalline material under the target load according to the node characteristics of each node in the characteristic graph data and the connection relationship between each node and other nodes, and performs the task according to the stress of the target polycrystalline material under the target load.

[0017] From the above method, it can be seen that the accuracy of material property prediction can be improved by integrating the advantages of convolutional neural networks and graph neural networks. Among them, the convolutional neural network is responsible for capturing the shape information of the grains, while the graph neural network predicts the mechanical properties based on the shape information of the grains, the size of the grains, the orientation of the grains and the grain boundary information. By combining the above-mentioned convolutional neural network and graph neural network, it is possible to simultaneously utilize the information of the grains and grain boundaries in the polycrystalline material to achieve continuous prediction of the nonlinear mechanical properties of the polycrystalline material, and the substrate properties can be used as the attribute input of the grains, thereby supporting the stress prediction of polycrystalline materials under different materials and enhancing the accuracy of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings described herein are used to provide a further understanding of this specification and constitute a part of this specification. The illustrative embodiments and descriptions of this specification are used to explain this specification and do not constitute an improper limitation on this specification. In the drawings: Figure 1 A schematic diagram of a polycrystalline material provided in this specification; Figure 2 A schematic diagram of a method for predicting mechanical properties of polycrystalline materials provided in this specification; Figure 3 This is the only schematic diagram of the prediction process of the mechanical properties of polycrystalline materials that can be provided in this specification; Figure 4 A schematic diagram of a polycrystalline material mechanical property prediction device provided in this specification; Figure 5 A method corresponding to the Figure 1 Schematic diagram of electronic equipment. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solutions and advantages of this specification more clear, the technical solutions of this specification will be clearly and completely described below in combination with the specific embodiments of this specification and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of this specification, not all of them. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this specification.

[0020] Generally speaking, polycrystalline materials are solid materials composed of many small crystal particles (called grains). Each grain is a single crystal, but the orientations of the grains are different, so the whole material presents a polycrystalline structure, such as Figure 1 shown.

[0021] Figure 1 This is a schematic diagram of the polycrystalline material provided in this specification.

[0022] Combination Figure 1 It can be seen that polycrystalline materials can be composed of several grains. The atomic arrangement inside each grain is orderly and has the same crystal structure and orientation. The size, shape and orientation of each grain have an important influence on the performance of polycrystalline materials.

[0023] Based on this, polycrystalline materials with different microstructures often have different mechanical properties. For example, polycrystalline materials with small and equiaxed grains usually have higher strength and toughness. This is because smaller grain size can reduce dislocation movement between grain boundaries, thereby improving the material's ability to resist deformation. Another example is that polycrystalline materials with grains of a specific shape (such as long strips or sheets) have different mechanical properties in different directions. For example, long strips of grains may show higher strength and lower ductility along their length. Therefore, how to predict the macroscopic mechanical properties of polycrystalline materials based on their microstructure, that is, the nonlinear stress-strain response curve, is extremely important for guiding the production of advanced mechanical materials.

[0024] The technical solutions provided by the embodiments of this specification are described in detail below in conjunction with the accompanying drawings.

[0025] Figure 2 A schematic flow chart of a method for predicting mechanical properties of polycrystalline materials provided in this specification includes the following steps: S201: Acquire characteristic graph data and target load of a target polycrystalline material, wherein different nodes in the characteristic graph data correspond to different grains contained in the target polycrystalline material, and for each node, if the grain corresponding to the node is adjacent to other grains through a grain boundary, the node is connected to the nodes corresponding to the other grains, and the node feature of the node includes at least one of grain shape feature, grain volume feature, substrate property feature, and grain orientation feature, wherein the grain shape feature of the node is obtained by convolution extraction based on a discretized structure array corresponding to the target polycrystalline material by a convolution model.

[0026] In this specification, the business platform can determine the polycrystalline material that needs to be predicted for mechanical properties as the target polycrystalline material, and determine the load acting on the target polycrystalline material (i.e., the degree of strain applied to the polycrystalline material to cause the polycrystalline material to deform) as the target load according to the operation instructions input by the user. Then, a three-dimensional array of the microstructure of the target polycrystalline material after discretization can be obtained, wherein the different elements contained in the three-dimensional array of the microstructure of the target polycrystalline material after discretization correspond to different sampling points uniformly distributed in the target polycrystalline material, and the value of each element is used to characterize the grain to which the sampling point corresponding to each element belongs, as shown in detail. Figure 3 shown.

[0027] Figure 3 This is the only schematic diagram of the process of predicting the mechanical properties of polycrystalline materials that can be provided in this specification.

[0028] Combination Figure 3 It can be seen that the business platform can discretize the microstructure of the target polycrystalline material. After the discretization, the microstructure of the target polycrystalline material can obtain a three-dimensional array D corresponding to r1*r2*r3 discrete sampling points uniformly distributed in the target polycrystalline material, where r1, r2, and r3 are the resolutions of the discretization. Here, r1, r2, and r3 can be set according to actual needs. The element indexed as (x, y, z) in the three-dimensional array D corresponds to the sampling point located at the (x, y, z) coordinate in the microstructure of the target polycrystalline material, and the value of the element indexed as (x, y, z) in the three-dimensional array D is the identifier of the grain to which the sampling point located at the (x, y, z) coordinate in the microstructure of the target polycrystalline material belongs. For example, if the identifier of the grain to which the sampling point located at the (x, y, z) coordinate in the microstructure of the target polycrystalline material belongs is i, then the value of the element indexed as (x, y, z) in the three-dimensional array D is i.

[0029] In addition, the business platform can also obtain the substrate property characteristics and grain orientation characteristics {O1, O2...On} of the target polycrystalline material. The substrate property characteristics here are used to reflect the physical or mechanical properties of the polycrystalline material itself. For example, the substrate property characteristics here can be the rigid matrix of the base material used by the target polycrystalline material, which can be determined based on the material itself. The grain orientation characteristics here are used to identify the positional relationship of the direction of the internal atomic arrangement of each grain contained in the polycrystalline material relative to the external reference system.

[0030] Furthermore, the business platform can encode each element contained in the three-dimensional array to obtain a coded feature vector array corresponding to the three-dimensional array, and input the coded feature vector array into a pre-trained convolution model, so as to convolve each coded feature vector in the coded feature vector array that matches the grain for each grain contained in the target polycrystalline material through the convolution model (if the sampling point corresponding to a coded feature vector belongs to the grain, then it can be considered that the coded feature vector matches the grain) to obtain the grain shape characteristics of the grain.

[0031] Specifically, the business platform can use one-hot encoding to convert the three-dimensional array D from the r1*r2*r3 format to the n*r1*r2*r3 format, where n is the number of grains contained in the target polycrystalline material, and convert the grain identifiers in the three-dimensional array D into an n-dimensional vector [0,0…,0,1,0,0…0], and the position of element 1 is the i-th dimension in the n-dimensional vector. Thus, the converted encoding feature vector array T of the shape [n,r1,r2,r3] can be obtained, and then the encoding feature vector array T can be input into the pre-trained convolution model, so that the convolution model can be used to convolve each encoding feature vector in the encoding feature vector array that matches each grain contained in the target polycrystalline material to obtain the grain shape feature of the grain, and finally obtain n grain shape features, and the dimension of the grain shape feature of each grain is w.

[0032] In addition, the business platform can also determine, for each grain contained in the target polycrystalline material, the elements whose corresponding sampling points belong to the grain from the different elements contained in the three-dimensional array, as the matching elements of the grain, and then determine the grain volume characteristics of the grain according to the number of matching elements of the grain and the total number of elements contained in the three-dimensional array, and obtain the characteristic graph data of the target polycrystalline material according to the grain shape characteristics and grain volume characteristics of each grain.

[0033] Specifically, the business platform can use the ratio between the number of matching elements of each grain (i.e., the frequency of occurrence of the grain identifier i in the three-dimensional array D) and the total number of elements contained in the three-dimensional array as the grain volume feature for each grain contained in the target polycrystalline material. The grain volume feature here can reflect the volume proportion of the grain in the microstructure of the entire polycrystalline material.

[0034] Furthermore, the business platform can construct characteristic graph data of the target polycrystalline material based on the above-mentioned three-dimensional array. In the characteristic graph data of the target polycrystalline material, different nodes correspond to different grains contained in the target polycrystalline material. For each node, if the grain corresponding to the node is adjacent to other grains through a grain boundary, then the node is connected to the node corresponding to the other grain. After determining each node in the characteristic graph data of the target polycrystalline material and the connection relationship between each node, at least one node feature of the grain shape feature, grain volume feature, substrate property feature, and grain orientation feature of each node determined by the above method can be assigned to each node to obtain the characteristic graph data of the target polycrystalline material.

[0035] It should be noted that, in this specification, the microstructure of the above-mentioned target polycrystalline material can be acquired by a preset image acquisition device, and the image acquisition device here can be: a scanning electron microscope, a transmission electron microscope, a camera, etc.

[0036] Of course, the microstructure of the target polycrystalline material mentioned above may also be generated by a specified algorithm according to configuration parameters input by the user.

[0037] In this specification, the execution entity used to implement the method for predicting the mechanical properties of polycrystalline materials may refer to a designated device such as a server set up on a business platform, or may refer to a terminal device such as a desktop computer or a laptop computer. For the sake of ease of description, the method for predicting the mechanical properties of polycrystalline materials provided in this specification is described below using the server as an example of the execution entity.

[0038] S202: Input the characteristic graph data and the target load into a pre-trained prediction model, so as to predict the stress of the target polycrystalline material under the target load according to the node characteristics of each node in the characteristic graph data and the connection relationship between each node and other nodes through the prediction model.

[0039] In this specification, after obtaining the characteristic graph data of the target polycrystalline material, the server can input the obtained characteristic graph data and the target load into a pre-trained prediction model to predict the stress of the target polycrystalline material under the target load based on the node characteristics of each node in the characteristic graph data of the target polycrystalline material and the connection relationship between each node and other nodes through the prediction model.

[0040] Specifically, the above-mentioned prediction model can be composed of a graph sampling and aggregation layer (Graph Sample and Aggregate, GraphSAGE) layer and several fully connected layers, wherein the server can input the characteristic graph data and the target load of the target polycrystalline material into the graph sampling and aggregation layer of the prediction model, so as to determine the neighbor nodes of the node for each node according to the connection relationship between the node and other nodes through the graph sampling and aggregation layer of the prediction model, and determine the aggregate node features of the node according to the node features of the node and the node features of the neighbor nodes, so that the aggregate node features of each node contained in the characteristic graph data of the target polycrystalline material can be globally pooled to fuse the aggregate node features of each node to obtain the global features of the target polycrystalline material, and then the global features and the target load of the target polycrystalline material can be input into the fully connected layer, so as to predict the stress of the target polycrystalline material under the target load according to the global features and the target load of the target polycrystalline material through the fully connected layer.

[0041] It should be noted that the above-mentioned prediction model can be deployed to the server only after training, wherein the training method of the prediction model can be to obtain sample characteristic map data and sample load of the sample polycrystalline material, and input the obtained sample characteristic map data and sample load into the prediction model to be trained, so as to predict the stress of the sample polycrystalline material under the sample load according to the node characteristics of each node in the sample characteristic map data, and the connection relationship between each node and other nodes through the prediction model to be trained, as the predicted stress, and then according to the deviation between the predicted stress and the actual stress of the sample polycrystalline material under the sample load, the target loss value can be determined, and the minimization of the target loss value is used as the optimization goal, and the prediction model to be trained is trained to obtain the prediction model.

[0042] The deviation between the predicted stress and the actual stress of the sample polycrystalline material under the sample load is proportional to the target loss value.

[0043] The actual stress of the sample polycrystalline material under the sample load can be obtained by simulation software through finite element analysis calculation.

[0044] It is worth noting that the above-mentioned convolution model and the above-mentioned prediction model can be trained separately or together, wherein the method for training the above-mentioned convolution model and the above-mentioned prediction model together can be to obtain the sample characteristic map data and the sample load of the sample polycrystalline material, determine the three-dimensional array of the sample polycrystalline material after discretization processing, and input the determined sample coding feature vector array into the convolution model to be trained according to the sample coding feature vector array corresponding to the sample polycrystalline material, so as to convolve each coding feature vector in the coding feature vector array that matches the grain for each grain contained in the sample polycrystalline material through the convolution model to be trained, so as to obtain the grain shape characteristics of the grain, and then the sample polycrystalline material can be used to predict the shape of the grain. The grain shape characteristics and other node characteristics of each grain contained in the material are used to obtain the characteristic graph data of the sample polycrystalline material, and the characteristic graph data of the sample polycrystalline material and the sample load are input into the prediction model to be trained, so as to predict the stress of the sample polycrystalline material under the sample load according to the node characteristics of each node in the sample characteristic graph data and the connection relationship between each node and other nodes through the prediction model to be trained, as the predicted stress, and then the target loss value can be determined according to the deviation between the predicted stress and the actual stress of the sample polycrystalline material under the sample load, and the minimization of the target loss value is used as the optimization goal, and the convolution model to be trained and the prediction model to be trained are trained to obtain the convolution model and the prediction model.

[0045] S203: Execute a task according to the stress of the target polycrystalline material under the target load.

[0046] In this specification, after determining the stress of the target polycrystalline material under the target load, the server can display the stress of the target polycrystalline material under the target load to the user, so that the user can perform tasks according to the stress of the target polycrystalline material under the target load.

[0047] In addition, the server can also input the stress of the target polycrystalline material under the target load into a preset detection model, so as to predict the probability value of the target polycrystalline material undergoing an abnormality under the target load based on the stress of the target polycrystalline material under the target load through the detection model, and generate a safety assessment result based on the probability value of the target polycrystalline material undergoing an abnormality under the target load, and display the safety assessment result to the user. The abnormalities here include: fracture, crack extension, etc.

[0048] In addition, the server can also input the stress of the target polycrystalline material under the target load into a preset recommendation model, so as to determine the optimized processing parameters for the target polycrystalline material according to the stress of the target polycrystalline material under the target load through the recommendation model, and display the optimized processing parameters to the user. The optimized processing parameters here are used to configure the 3D printer so that the 3D printer performs 3D printing according to the optimized processing parameters, thereby improving the yield of 3D printing.

[0049] In actual application scenarios, the server can also obtain at least one pair of stress-strain data pairs input by the user as a target stress-strain data pair, and input the target stress-strain data pair into a preset search recommendation system, so as to determine the candidate polycrystalline material whose corresponding stress-strain response curve matches the target stress-strain data pair from the preset candidate polycrystalline materials through the search recommendation system as a reference polycrystalline material, and recommend the identification data of the reference polycrystalline material to the user.

[0050] The above stress-strain data pair includes a specified load and an expected stress.

[0051] In the above content, for each candidate polycrystalline material, the stress-strain response curve of the candidate polycrystalline material is determined in advance by a prediction model based on the characteristic graph data of the candidate polycrystalline material. The stress-strain response curve here is used to characterize the relationship between different loads applied to the candidate polycrystalline material and the stress of the candidate polycrystalline material under different loads.

[0052] From the above method, it can be seen that the server can improve the accuracy of material property prediction by integrating the advantages of convolutional neural networks and graph neural networks. Among them, convolutional neural networks are responsible for capturing the shape information of grains, while graph neural networks consider the size, orientation and grain boundary information of grains. Through the combination of the above-mentioned convolutional neural networks and graph neural networks, it is possible to simultaneously utilize the information of grains and grain boundaries in polycrystalline materials to achieve continuous prediction of nonlinear mechanical properties of polycrystalline materials, rather than just stress-strain curves based on discrete points. In addition, the server can also input substrate properties as grain attributes, thereby supporting stress prediction of polycrystalline materials under different materials, enhancing the applicability and accuracy of the model.

[0053] The above are one or more embodiments of the polycrystalline material mechanical property prediction method of this specification. Based on the same idea, this specification also provides a corresponding polycrystalline material mechanical property prediction device, such as Figure 4 shown.

[0054] Figure 4 A schematic diagram of a polycrystalline material mechanical property prediction device provided in this specification includes: An acquisition module 401 is used to acquire characteristic graph data and a target load of a target polycrystalline material, wherein different nodes in the characteristic graph data correspond to different grains contained in the target polycrystalline material, and for each node, if the grain corresponding to the node is adjacent to other grains through a grain boundary, the node is connected to the nodes corresponding to the other grains, and the node feature of the node includes at least one of a grain shape feature, a grain volume feature, a substrate property feature, and a grain orientation feature, wherein the grain shape feature of the node is obtained by performing convolution extraction based on a discretized structure array corresponding to the target polycrystalline material by a convolution model; A prediction module 402, for inputting the characteristic graph data and the target load into a pre-trained prediction model, so as to predict the stress of the target polycrystalline material under the target load according to the node characteristics of each node in the characteristic graph data and the connection relationship between each node and other nodes through the prediction model; The execution module 403 is used to execute the task according to the stress of the target polycrystalline material under the target load.

[0055] Optionally, the acquisition module 401 is specifically used to obtain a three-dimensional array of the microstructure of the target polycrystalline material after discretization processing, wherein different elements contained in the three-dimensional array correspond to different sampling points uniformly distributed in the target polycrystalline material, and the value of each element is used to characterize the grain to which the sampling point corresponding to each element belongs; encode each element contained in the three-dimensional array to obtain a coded feature vector array corresponding to the three-dimensional array, and input the coded feature vector array into a pre-trained convolution model, so as to convolve each coded feature vector in the coded feature vector array that matches the grain for each grain contained in the target polycrystalline material through the convolution model to obtain the grain shape characteristics of the grain; and obtain the characteristic graph data of the target polycrystalline material according to the grain shape characteristics of each grain.

[0056] Optionally, the acquisition module 401 is specifically used to, for each grain contained in the target polycrystalline material, determine the element whose corresponding sampling point belongs to the grain from the different elements contained in the three-dimensional array as the matching elements of the grain; determine the grain volume characteristics of the grain according to the number of matching elements of the grain and the total number of elements contained in the three-dimensional array; and acquire the characteristic graph data of the target polycrystalline material according to the grain shape characteristics and grain volume characteristics of each grain.

[0057] Optionally, the prediction module 402 is specifically used to input the characteristic graph data and the target load into a pre-trained prediction model, so as to determine the neighbor nodes of each node according to the connection relationship between the node and other nodes through the prediction model, and determine the aggregate node characteristics of the node according to the node characteristics of the node and the node characteristics of the neighbor nodes, and predict the stress of the target polycrystalline material under the target load according to the aggregate node characteristics of each node.

[0058] Optionally, the prediction module 402 is specifically used to perform global pooling processing on the aggregated node features of each node to fuse the aggregated node features of each node to obtain the global features of the target polycrystalline material, and predict the stress of the target polycrystalline material under the target load based on the global features.

[0059] Optionally, the device further comprises: a training module 404; The training module 404 is specifically used to obtain sample characteristic graph data and sample load of a sample polycrystalline material, and input the sample characteristic graph data and sample load into a prediction model to be trained, so as to predict the stress of the sample polycrystalline material under the sample load according to the node characteristics of each node in the sample characteristic graph data and the connection relationship between each node and other nodes through the prediction model to be trained, as the predicted stress, determine the target loss value according to the deviation between the predicted stress and the actual stress of the sample polycrystalline material under the sample load, and train the prediction model to be trained with minimizing the target loss value as the optimization target to obtain a prediction model, wherein the deviation between the predicted stress and the actual stress of the sample polycrystalline material under the sample load is proportional to the target loss value.

[0060] Optionally, the execution module 403 is also used to obtain at least one pair of stress-strain data pairs input by the user as a target stress-strain data pair, wherein the stress-strain data pair includes a specified load and an expected stress; determine, from each preset candidate polycrystalline material, a candidate polycrystalline material whose corresponding stress-strain response curve matches the target stress-strain data pair as a reference polycrystalline material, and recommend identification data of the reference polycrystalline material to the user, wherein the stress-strain response curve is determined in advance by the prediction model based on the characteristic graph data of the candidate polycrystalline material.

[0061] This specification also provides a computer-readable storage medium, which stores a computer program, which can be used to execute the above Figure 1 A method for predicting the mechanical properties of polycrystalline materials is provided.

[0062] This manual also provides Figure 5 The one shown corresponds to Figure 1 A schematic diagram of the electronic device. Figure 5 As mentioned above, at the hardware level, the electronic device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory, and may also include other hardware required for the business. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it to achieve the above Figure 1 The method for predicting the mechanical properties of polycrystalline materials. Of course, in addition to the software implementation, this specification does not exclude other implementations, such as logic devices or a combination of software and hardware, etc., that is, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0063] For the improvement of a technology, it can be clearly distinguished whether it is a hardware improvement (for example, improvement of the circuit structure of diodes, transistors, switches, etc.) or a software improvement (improvement of the method flow). However, with the development of technology, many improvements of the method flow today can be regarded as direct improvements of the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that the improvement of a method flow cannot be implemented with a hardware entity module. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is such an integrated circuit whose logical function is determined by the user's programming of the device. Designers can "integrate" a digital system on a PLD by programming themselves, without having to ask chip manufacturers to design and make dedicated integrated circuit chips. Moreover, nowadays, instead of manually making integrated circuit chips, this kind of programming is mostly implemented by "logic compiler" software, which is similar to the software compiler used when developing and writing programs, and the original code before compilation must also be written in a specific programming language, which is called hardware description language (HDL). There is not only one kind of HDL, but many kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also know that it is only necessary to program the method flow slightly in the above-mentioned hardware description languages ​​and program it into the integrated circuit, and then it is easy to obtain the hardware circuit that implements the logic method flow.

[0064] The controller may be implemented in any suitable manner, for example, the controller may take the form of a microprocessor or processor and a computer-readable medium storing a computer-readable program code (e.g., software or firmware) executable by the (micro)processor, a logic gate, a switch, an application-specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller, examples of which include but are not limited to the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320, and the memory controller may also be implemented as part of the control logic of the memory. It is also known to those skilled in the art that, in addition to implementing the controller in a purely computer-readable program code manner, the controller may be implemented in the form of a logic gate, a switch, an application-specific integrated circuit, a programmable logic controller, and an embedded microcontroller by logically programming the method steps. Therefore, such a controller may be considered as a hardware component, and the devices for implementing various functions included therein may also be considered as structures within the hardware component. Or even, the devices for implementing various functions may be considered as both software modules for implementing the method and structures within the hardware component.

[0065] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0066] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0067] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Therefore, this specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0068] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0069] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0070] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0071] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0072] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0073] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0074] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0075] It should be understood by those skilled in the art that the embodiments of this specification may be provided as methods, systems or computer program products. Therefore, this specification may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0076] This specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0077] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0078] The above description is only an embodiment of the present specification and is not intended to limit the present specification. For those skilled in the art, the present specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present specification shall be included in the scope of the claims of the present specification.

Claims

1. A method for predicting mechanical properties of polycrystalline materials, characterized in that: include: Acquire characteristic graph data and target load of a target polycrystalline material, wherein different nodes in the characteristic graph data correspond to different grains contained in the target polycrystalline material, and for each node, if the grain corresponding to the node is adjacent to other grains through a grain boundary, the node is connected to the nodes corresponding to the other grains, and the node feature of the node includes at least one of a grain shape feature, a grain volume feature, a substrate property feature, and a grain orientation feature, wherein the grain shape feature of the node is obtained by performing convolution extraction based on a discretized structure array corresponding to the target polycrystalline material by a convolution model; Inputting the characteristic graph data and the target load into a pre-trained prediction model, so as to predict the stress of the target polycrystalline material under the target load according to the node characteristics of each node in the characteristic graph data and the connection relationship between each node and other nodes through the prediction model; The task is executed according to the stress of the target polycrystalline material under the target load.

2. The method according to claim 1, characterized in that Obtain the characteristic map data of the target polycrystalline material, including: Obtaining a three-dimensional array of the microstructure of the target polycrystalline material after discretization, wherein different elements contained in the three-dimensional array correspond to different sampling points uniformly distributed in the target polycrystalline material, and the value of each element is used to characterize the grain to which the sampling point corresponding to each element belongs; Encode each element contained in the three-dimensional array to obtain a coding feature vector array corresponding to the three-dimensional array, and input the coding feature vector array into a pre-trained convolution model, so as to convolve each coding feature vector in the coding feature vector array that matches the grain for each grain contained in the target polycrystalline material through the convolution model to obtain a grain shape feature of the grain; According to the grain shape characteristics of each grain, characteristic graph data of the target polycrystalline material is obtained.

3. The method according to claim 2, characterized in that According to the grain shape characteristics of each grain, characteristic graph data of the target polycrystalline material is obtained, specifically including: For each crystal grain included in the target polycrystalline material, determining, from the different elements included in the three-dimensional array, an element whose corresponding sampling point belongs to the crystal grain as each matching element of the crystal grain; Determining a grain volume feature of the grain according to the number of each matching element of the grain and the total number of each element included in the three-dimensional array; According to the grain shape characteristics and grain volume characteristics of each grain, characteristic graph data of the target polycrystalline material is obtained.

4. The method according to claim 1, characterized in that Inputting the characteristic graph data and the target load into a pre-trained prediction model, so as to predict the stress of the target polycrystalline material under the target load according to the node characteristics of each node in the characteristic graph data and the connection relationship between each node and other nodes through the prediction model, specifically includes: The characteristic graph data and the target load are input into a pre-trained prediction model, so as to determine the neighbor nodes of each node according to the connection relationship between the node and other nodes through the prediction model, and determine the aggregate node characteristics of the node according to the node characteristics of the node and the node characteristics of the neighbor nodes, and predict the stress of the target polycrystalline material under the target load according to the aggregate node characteristics of each node.

5. The method according to claim 4, characterized in that Predicting the stress of the target polycrystalline material under the target load according to the aggregate node characteristics of each node, specifically comprising: The aggregate node features of each node are globally pooled to fuse the aggregate node features of each node to obtain the global features of the target polycrystalline material, and the stress of the target polycrystalline material under the target load is predicted based on the global features.

6. The method according to claim 1, characterized in that Training the prediction model specifically includes: Obtaining sample characteristic diagram data and sample load of a sample polycrystalline material; Inputting the sample characteristic graph data and the sample load into a prediction model to be trained, so as to predict the stress of the sample polycrystalline material under the sample load as a predicted stress according to the node characteristics of each node in the sample characteristic graph data and the connection relationship between each node and other nodes through the prediction model to be trained; According to the deviation between the predicted stress and the actual stress of the sample polycrystalline material under the sample load, a target loss value is determined, and the prediction model to be trained is trained with minimizing the target loss value as the optimization goal to obtain a prediction model, wherein the deviation between the predicted stress and the actual stress of the sample polycrystalline material under the sample load is proportional to the target loss value.

7. The method according to claim 1, characterized in that The method further comprises: Acquire at least one pair of stress-strain data input by a user as a target stress-strain data pair, wherein the stress-strain data pair includes a specified load and an expected stress; According to the target stress-strain data pair, a candidate polycrystalline material whose corresponding stress-strain response curve matches the target stress-strain data pair is determined from the preset candidate polycrystalline materials as a reference polycrystalline material, and the identification data of the reference polycrystalline material is recommended to the user, wherein the stress-strain response curve is determined in advance by the prediction model based on the characteristic graph data of the candidate polycrystalline material.

8. A device for predicting mechanical properties of polycrystalline materials, characterized in that: include: An acquisition module is used to acquire characteristic graph data and a target load of a target polycrystalline material, wherein different nodes in the characteristic graph data correspond to different grains contained in the target polycrystalline material, and for each node, if the grain corresponding to the node is adjacent to other grains through a grain boundary, the node is connected to the nodes corresponding to the other grains, and the node feature of the node includes at least one of a grain shape feature, a grain volume feature, a substrate property feature, and a grain orientation feature, wherein the grain shape feature of the node is obtained by performing convolution extraction based on a discretized structure array corresponding to the target polycrystalline material by a convolution model; A prediction module, used for inputting the characteristic graph data and the target load into a pre-trained prediction model, so as to predict the stress of the target polycrystalline material under the target load according to the node characteristics of each node in the characteristic graph data and the connection relationship between each node and other nodes through the prediction model; An execution module is used to execute a task according to the stress of the target polycrystalline material under the target load.

9. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method described in any one of claims 1 to 7 is implemented.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method described in any one of claims 1 to 7 is implemented.

Images