A material design method, device, terminal and medium based on artificial intelligence

Abstract

The present invention discloses a material design method, device, terminal and medium based on artificial intelligence, the method comprising: training an artificial intelligence module so that the artificial intelligence module obtains a mapping rule between material structure and material properties; inputting expected performance parameters into the trained artificial intelligence module; the artificial intelligence model outputs a target material that meets the expected performance parameters. The present invention obtains the mapping rule of material structure-material properties by training the artificial intelligence module, so that the artificial intelligence module can explore the target material with the expected performance parameters according to the expected performance parameters and mapping rules set by the researchers, avoiding the thinking limitation caused by the researchers' exploration based on their own experience in the prior art, thereby greatly improving the chances of discovering new material applications.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a material design method, device, terminal and medium based on artificial intelligence. Background Art

[0002] Materials with suitable properties have always been an important foundation for the development of human industry. The exploration of new material applications requires repeated material synthesis experiments to find material structures with suitable properties. With the development of automated robot systems, automated robot systems have been applied to the field of chemical synthesis experiments, effectively reducing the burden on researchers. Existing automated experimental robot systems are usually set to perform relevant experimental steps according to the target materials input by researchers to obtain products, but the target materials need to be set by researchers, and researchers often trust their own experience and judgment more, and it is difficult to jump out of the limitations of inherent thinking to explore, reducing the opportunity to find new materials. The rapid development of artificial intelligence in recent years has greatly promoted the iteration of many fields, including the field of material research and development. In the past, the exploration of this field mainly relied on the personal experience of researchers, and it was difficult for researchers to trust the exploration beyond the scope of experience, resulting in the loss of potential new materials. Artificial intelligence can search for the objective laws of a system from a qualitative and quantitative perspective, and is not interfered by human personal consciousness. The mapping rules between material structure and material performance happen to be able to be qualitatively and quantitatively analyzed through experiments or calculations. Therefore, an artificial intelligence-based material design method that can break predefined barriers can achieve more efficient material research and development exploration. Summary of the invention

[0003] In order to solve the above technical problems, the first aspect of the present invention discloses a material design method based on artificial intelligence, the method comprising:

[0004] Training the artificial intelligence module so that the artificial intelligence module acquires a mapping rule between material structure and material property;

[0005] Input the expected performance parameters into the trained AI module;

[0006] The artificial intelligence model outputs a target material that meets the desired performance parameters.

[0007] In a further embodiment, the artificial intelligence module includes a supervision model and a generation model, the generation model includes an encoder and a decoder connected in sequence, and the training method for the artificial intelligence module to obtain the mapping rule between the material structure and the material property includes the following steps: the encoder represents the input material in the form of a vector X, each element in the vector X represents the number of atoms of a chemical element arranged in a certain order; the encoder maps the input vector X to a compressed latent vector V, and the decoder maps the latent vector V to reconstruct the material in the form of a generated vector Z; the generation model calculates the square of the Euclidean distance between the vector X and the vector Z to obtain a first loss function; the supervision model calculates the square of the difference between at least one element y in the latent vector V that is pre-designated as representing the predicted value of the material property and the actual value Y of the material property to obtain a second loss function; in the training stage, the generation model is trained to minimize the sum of the first loss function and the second loss function to obtain the mapping rule between the material structure and the material property; in the use stage, the latent vector V containing the expected performance parameters is input to the decoder, and the decoder outputs the target material in the form of the generated vector Z.

[0008] In a further embodiment, when the expected performance parameter is input to the artificial intelligence module, at least one element y in the potential vector V is designated as the expected performance parameter, and the target material containing the expected performance parameter is output through the decoder.

[0009] In a further embodiment, the latent vector V includes at least one element y representing a desired performance parameter and a plurality of random items, and the numerical range of the random items is generated and recorded by the artificial intelligence module during the training process.

[0010] The second aspect of the present invention discloses an artificial intelligence-based material design device, the device comprising an artificial intelligence module; the artificial intelligence module comprising a supervision model and a generation model, the generation model comprising an encoder and a decoder, the encoder is used to represent the input material in the form of a vector X, each element in the vector X represents the number of atoms of a chemical element arranged in a determined order; the encoder maps the input vector X to a compressed latent vector V, the decoder maps the latent vector V to the reconstructed material in the form of a generated vector Z; the generation model calculates the square of the Euclidean distance between the vector X and the vector Z to obtain a first loss function;

[0011] The supervised model is used to calculate the square of the difference between at least one element y in the latent vector V that is pre-specified as representing a predicted value of a material property and the actual value Y of the material property to obtain a second loss function.

[0012] The third aspect of the present invention discloses a material design recommendation terminal based on artificial intelligence, the terminal comprising:

[0013] A memory storing executable program code;

[0014] a processor coupled to the memory;

[0015] The processor calls the executable program code stored in the memory to execute the artificial intelligence-based material design method as described in the first aspect of the present invention.

[0016] The fourth aspect of the present invention discloses a computer storage medium, which stores computer instructions. When the computer instructions are called, they are used to execute some or all of the steps in the artificial intelligence-based material design method disclosed in the first aspect of the present invention.

[0017] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0018] The present invention obtains the mapping rules of material structure-material performance by training the artificial intelligence module, so that the artificial intelligence module can explore the target materials with expected performance parameters according to the expected performance parameters and mapping rules set by the researchers, avoiding the thinking limitations caused by the researchers' exploration based on their own experience in the prior art, thereby greatly improving the chances of discovering new material applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0020] Figure 1 It is a flow chart of a material design method based on artificial intelligence disclosed in an embodiment of the present invention;

[0021] Figure 2 It is a structural schematic diagram of a material design device based on artificial intelligence disclosed in an embodiment of the present invention;

[0022] Figure 3 It is a structural schematic diagram of a material design terminal based on artificial intelligence disclosed in an embodiment of the present invention;

[0023] Figure 4 It is a structural schematic diagram of a computer storage medium disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] The terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish different objects rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, device, product or end including a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units inherent to these processes, methods, products or ends.

[0026] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0027] The present invention obtains the mapping rules of material structure-material performance by training the artificial intelligence module, so that the artificial intelligence module can explore the target material with the expected performance parameters according to the expected performance parameters and mapping rules set by the researchers, avoiding the thinking limitations caused by the researchers' own experience exploration in the prior art, thereby greatly improving the chances of discovering new material applications. The following are detailed descriptions.

[0028] Embodiment 1

[0029] See also Figure 1 , Figure 1 Schematic diagram of a material design method based on artificial intelligence disclosed in an embodiment of the present invention. Figure 1 As shown, the artificial intelligence-based material design method may include the following operations:

[0030] 101 trains the artificial intelligence module so that the artificial intelligence module obtains the mapping rules between the material structure and the material properties; a specific embodiment, such as Figure 2As shown, the artificial intelligence module includes a supervision model and a generation model, and the generation model includes an encoder and a decoder. The encoder is used to represent the input material in the form of a vector X, and each element in the vector X represents the number of atoms of a chemical element arranged in a certain order, that is, the composition and structure of the material; the encoder maps the input vector X to a compressed potential vector V, and the decoder maps the potential vector V to a generated vector Z; the generation model calculates the square of the Euclidean distance between the vector X and the vector Z to obtain a first loss function loss_rebuilt, that is, a reconstruction loss function; the supervision model is used to calculate the square of the difference between at least one element y in the potential vector V that is pre-specified as representing the predicted value of the material property and the actual value Y of the material property to obtain a second loss function loss_property, that is, a property loss function; in the training stage, the training generation model minimizes the sum of the first loss function loss_rebuilt and the second loss function loss_property; in the use stage, the potential vector V containing the desired performance parameters is input to the decoder, and the decoder outputs the target material in the form of a generated vector Z. It can be understood that the artificial intelligence module qualitatively and quantitatively analyzes the mapping rules between material structure and material properties through bidirectional training of structure-property. Therefore, the artificial intelligence module can more accurately obtain the relationship between material structure and material properties, and can more easily break the limitations caused by researchers' pre-set target materials.

[0031] 102 Input the expected performance parameters into the trained artificial intelligence module; it is understandable that in practical applications, after the artificial intelligence module is trained, researchers only need to set the expected performance parameters y of the material, such as conductivity, which is achieved by setting at least one element y in the latent vector V that represents the predicted value of the material performance. In theory, any one or more elements in the latent vector V can be designated as y representing the predicted value of the material performance. For other elements in the latent vector V, because they also contain information about the composition and structure of the material, their numerical ranges are also recorded during the training process and used as random items when generated.

[0032] 103 The artificial intelligence model outputs a target material that meets the expected performance parameters. By inputting the expected performance parameter y, we can obtain a synthetic vector containing the expected performance parameter y and other random items, which is input into the decoder of the generative model to obtain a material structure that may have the expected performance parameters. Usually, 3 to 5 target materials are generated to increase the possibility of finding the required material.

[0033] The embodiments of the present invention have at least the following beneficial effects:

[0034] (1) By training the artificial intelligence module to obtain the mapping rules between material structure and material properties, the artificial intelligence module can explore the target materials with the expected performance parameters according to the expected performance parameters set by the researchers and the mapping rules. This avoids the thinking limitations caused by the researchers' exploration based on their own experience in the existing technology, thereby greatly increasing the chances of discovering new material applications.

[0035] (2) By minimizing the sum of the first loss function and the second loss function, the target material performance output by the artificial intelligence module is more in line with the needs of researchers, thereby improving the efficiency of material search.

[0036] Embodiment 2

[0037] See also Figure 2 , Figure 2 Schematic diagram of a material design device based on artificial intelligence disclosed in an embodiment of the present invention. Figure 2 As shown, the artificial intelligence-based material design device may include an artificial intelligence module; the artificial intelligence module includes a supervision model and a generation model, the generation model includes an encoder and a decoder, the encoder is used to represent the input material in the form of a vector X, each element in the vector X represents the number of atoms of a chemical element arranged in a certain order; the encoder maps the input vector X to a compressed latent vector V, and the decoder maps the latent vector V to reconstruct the material in the form of a generated vector Z; the generation model calculates the square of the Euclidean distance between the vector X and the vector Z to obtain a first loss function; the supervision model is used to calculate the square of the difference between at least one element y in the latent vector V that is pre-designated as representing a predicted value of a material property and the actual value Y of the material property to obtain a second loss function.

[0038] For the specific description of the above-mentioned artificial intelligence-based material design device, reference can be made to the specific description of the above-mentioned artificial intelligence-based material design method, which will not be repeated here.

[0039] Embodiment 3

[0040] See also Figure 3 , Figure 3 Schematic diagram of a material design terminal based on artificial intelligence disclosed in an embodiment of the present invention. Figure 3 As shown, the artificial intelligence-based material design terminal may include:

[0041] A memory 301 storing executable program codes;

[0042] a processor 302 coupled to the memory 301;

[0043] The processor 302 calls the executable program code stored in the memory 301 to execute the steps in the artificial intelligence-based material design method disclosed in the first embodiment of the present invention.

[0044] Embodiment 4

[0045] See also Figure 4 , Figure 4 Schematic diagram of the structure of a computer storage medium disclosed in an embodiment of the present invention. Figure 4 As shown, an embodiment of the present invention discloses a computer storage medium 401, which stores computer instructions. When the computer instructions are called, they are used to execute the steps in the artificial intelligence-based material design method disclosed in Embodiment 1 of the present invention.

[0046] The device embodiments described above are only illustrative, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, i.e., they may be located in one place, or they may be distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those of ordinary skill in the art may understand and implement it without creative work.

[0047] Through the specific description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the above technical solution can be essentially or partly contributed to the prior art in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable rewritable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, a magnetic disk storage, a magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0048] Finally, it should be noted that the artificial intelligence-based material design method, device, terminal and storage medium disclosed in the embodiments of the present invention are only preferred embodiments of the present invention, which are only used to illustrate the technical solution of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, it should be understood by those skilled in the art that the technical solutions described in the aforementioned embodiments can still be modified, or some of the technical features therein can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A material design method based on artificial intelligence, characterized in that: The method comprises: Training an artificial intelligence module so that the artificial intelligence module obtains a mapping rule between a material structure and a material property; the artificial intelligence module includes a supervision model and a generation model, the generation model includes an encoder and a decoder connected in sequence, and the training method for the artificial intelligence module to obtain a mapping rule between a material structure and a material property includes the following steps: The encoder represents the input material in the form of a vector X, each element in the vector X represents the atomic number of a chemical element arranged in a certain order; the encoder maps the input vector X to a compressed latent vector V, and the decoder maps the latent vector V to the reconstructed material in the form of a generated vector Z; the generative model calculates the square of the Euclidean distance between the vector X and the vector Z to obtain a first loss function; The supervised model calculates the square of the difference between at least one element y in the latent vector V that is pre-specified as representing the predicted value of the material property and the actual value Y of the material property to obtain a second loss function; In the training phase, the generative model is trained to minimize the sum of the first loss function and the second loss function to obtain a mapping rule between material structure and material properties; in the use phase, a potential vector V containing desired performance parameters is input to the decoder, and the decoder outputs the target material in the form of a generated vector Z; Input the expected performance parameters into the trained AI module; The artificial intelligence model outputs a target material that meets the desired performance parameters.

2. The material design method based on artificial intelligence according to claim 1, characterized in that: When the expected performance parameter is input into the artificial intelligence module, at least one element y in the potential vector V is designated as the expected performance parameter, and the target material containing the expected performance parameter is output through the decoder.

3. The material design method based on artificial intelligence according to claim 2, characterized in that: The potential vector V includes at least one element y representing an expected performance parameter and a number of random items, and the numerical range of the random items is generated and recorded by the artificial intelligence module during the training process.

4. A material design device based on artificial intelligence, characterized in that: The device includes an artificial intelligence module; The artificial intelligence module includes a supervision model and a generation model, the generation model includes an encoder and a decoder, the encoder is used to represent the input material in the form of a vector X, each element in the vector X represents the number of atoms of a chemical element arranged in a certain order; the encoder maps the input vector X to a compressed potential vector V, and the decoder maps the potential vector V to the reconstructed material in the form of a generated vector Z; the generation model calculates the square of the Euclidean distance between the vector X and the vector Z to obtain a first loss function; The supervised model is used to calculate the square of the difference between at least one element y in the latent vector V that is pre-specified as representing a predicted value of a material property and the actual value Y of the material property to obtain a second loss function.

5. A material design terminal based on artificial intelligence, characterized in that: The terminal comprises: A memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the artificial intelligence-based material design method as described in any one of claims 1-3.

6. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the artificial intelligence-based material design method as described in any one of claims 1 to 3 is implemented.

Images