Component design method for encrypted multi-component alloy based on machine learning and product

Abstract

The invention discloses a machine learning-based component design method for encrypted multi-component alloy and a product, and relates to the technical field of material genes and privacy computing, and the method comprises the following steps: obtaining encrypted multi-component alloy components and target performance requirements sent by a data holder; generating a plurality of to-be-predicted encrypted multi-component alloys with different component mass proportions in a permutation and combination manner according to the components of the encrypted multi-component alloys; calculating the characteristics of each to-be-predicted encrypted multi-element alloy according to the components of the encrypted multi-element alloy and the component mass ratio; inputting the characteristics of each to-be-predicted encrypted multi-element alloy into the trained encrypted alloy performance prediction model to obtain corresponding performance data; and sending the component mass proportion of the encrypted multi-component alloy component corresponding to the performance data meeting the target performance requirement to a data holder. Cooperative innovation of alloy material research and development can be promoted, and meanwhile privacy security of data is ensured.

Description

Technical Field

[0001] The present application relates to the field of material genetics and privacy computing technology, and in particular to a composition design method and product for encrypted multi-element alloys based on machine learning. Background Art

[0002] In the development of modern industry and science and technology, alloy materials have become indispensable key materials due to their excellent comprehensive properties. In particular, precious metal bonding materials, with their excellent electrical conductivity, thermal stability and corrosion resistance, play a vital role in the fields of microelectronics and semiconductor manufacturing. The performance of such materials is directly related to the reliability, performance and life of electronic devices. Precious metal bonding materials usually involve precious metals such as gold, silver and platinum. These metals are widely used in the interconnection process of electronic packaging and semiconductor chips due to their high chemical stability and good electrical conductivity.

[0003] As electronic devices develop towards miniaturization and high performance, the performance requirements for alloy materials are becoming higher and higher. Traditional alloy materials can no longer meet these new demands. However, there are some limitations in the traditional trial-and-error method of developing alloys. First, the preparation process of alloy materials is complex and relies on a large number of experimental trial and error. This method is not only inefficient but also costly. Secondly, the performance optimization of alloy materials lacks systematic theoretical guidance and often relies on the experience and intuition of researchers, which limits the space for improving material performance. In order to solve these problems, researchers began to explore new alloy material design methods. The introduction of the Materials Genome Project provides new ideas for the research and development of alloy materials. The program aims to quickly screen and optimize the performance of materials through high-throughput experiments and computational methods, and accelerate the discovery and development of new materials. However, in the process of material research and development, data security and privacy protection are becoming increasingly important. With the strengthening of global data privacy regulations, how to balance the open sharing of data and privacy protection in material research and development has become an urgent problem to be solved. Summary of the invention

[0004] The purpose of this application is to provide a composition design method and product for encrypted multi-element alloys based on machine learning, which can both protect data privacy and accelerate the material research and development process.

[0005] To achieve the above objectives, this application provides the following solutions:

[0006] In a first aspect, the present application provides a composition design method for encrypted multi-element alloys based on machine learning, and the composition design method for encrypted multi-element alloys based on machine learning includes:

[0007] Obtaining the encrypted multi-element alloy composition and target performance requirements sent by the data holder; the encrypted multi-element alloy composition is encrypted by encrypting the element atomic parameter characteristic table;

[0008] According to the encrypted multi-element alloy composition, a plurality of to-be-predicted encrypted multi-element alloys with different component mass ratios are generated by permutation and combination;

[0009] Calculate the characteristics of each predicted encrypted multi-element alloy according to the encrypted multi-element alloy components and component mass ratios;

[0010] Input the characteristics of each of the encrypted multi-element alloys to be predicted into a trained encrypted alloy performance prediction model to obtain corresponding performance data; the encrypted alloy performance prediction model is obtained by training the characteristics of the encrypted training multi-element alloy and the training performance of the encrypted training multi-element alloy;

[0011] The encrypted component mass ratio of the multi-element alloy component corresponding to the performance data meeting the target performance requirement is sent to the data holder.

[0012] Optionally, the training method of the encrypted alloy performance prediction model includes:

[0013] Obtain the encrypted training multi-element alloy composition, the mass ratio of each component of the encrypted training multi-element alloy and the training performance sent by the data holder;

[0014] Calculating the characteristics of the encrypted training multi-element alloy according to the encrypted training multi-element alloy composition and the mass ratio of each component of the encrypted training multi-element alloy;

[0015] The encrypted training multi-element alloy features are used as input, and the training performance is used as output to train a machine learning model to obtain an encrypted alloy performance prediction model.

[0016] Optionally, the machine learning model is a regression model.

[0017] Optionally, the encryption method for encrypting the multi-element alloy composition includes:

[0018] The element atomic parameter characteristic table is encrypted by using a monotonic function to obtain an encrypted element atomic parameter characteristic table;

[0019] The encrypted element atomic parameter characteristic table is used to encrypt the multi-component alloy composition to obtain the encrypted multi-component alloy composition.

[0020] Optionally, according to the encrypted multi-element alloy components and component mass ratios, the characteristics of each to-be-predicted encrypted multi-element alloy are calculated, specifically including:

[0021] According to the encrypted multi-element alloy components and component mass ratios, the characteristics of each to-be-predicted encrypted multi-element alloy are calculated by a weighted method.

[0022] Optionally, the method for calculating the characteristics of the to-be-predicted encrypted multi-element alloy is: using a weighted method to calculate a weighted average value and characteristic dispersion of each characteristic of the to-be-predicted encrypted multi-element alloy.

[0023] Optionally, the weighted average value and characteristic dispersion of each characteristic of the predicted encrypted multi-element alloy are calculated as follows:

[0024]

[0025] in, represents the weighted average of the jth characteristic of the alloy, δf j Indicates the discreteness of the feature, c i represents the mass ratio of the ith element in the n-element alloy, f ij Represents the value of the jth feature of the ith element.

[0026] In a second aspect, the present application provides a computer device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any of the above-described composition design methods for encrypted multi-element alloys based on machine learning.

[0027] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-mentioned machine learning-based composition design methods for encrypted multi-element alloys.

[0028] In a fourth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned machine learning-based composition design methods for encrypted multi-element alloys.

[0029] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0030] The present application provides a composition design method and product for encrypted multi-element alloys based on machine learning, the method comprising: obtaining the encrypted multi-element alloy composition and target performance requirements sent by the data holder; the encrypted multi-element alloy composition is encrypted by encrypting the element atomic parameter feature table; according to the encrypted multi-element alloy composition, a number of encrypted multi-element alloys to be predicted with different component mass ratios are generated by permutation and combination; according to the encrypted multi-element alloy composition and the component mass ratio, the characteristics of each encrypted multi-element alloy to be predicted are calculated; the characteristics of each encrypted multi-element alloy to be predicted are respectively input into the trained encrypted alloy performance prediction model to obtain the corresponding performance data; the encrypted alloy performance prediction model is obtained by training the characteristics of the encrypted training multi-element alloy and the training performance of the encrypted training multi-element alloy; the component mass ratio of the encrypted multi-element alloy composition corresponding to the performance data that meets the target performance requirements is sent to the data holder. The present application provides a solution for multi-element alloy materials that protects data privacy and accelerates the material research and development process by combining privacy computing technology with a composition design method for alloy materials based on machine learning. The present application can promote collaborative innovation in alloy material research and development while ensuring data privacy security. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0032] Figure 1 A schematic flow chart of a composition design method for encrypted multi-element alloys based on machine learning provided in one embodiment of the present application;

[0033] Figure 2 A schematic diagram of an encryption process provided in an embodiment of the present application;

[0034] Figure 3 A schematic diagram of the added proportions and encrypted features corresponding to 8 virtual sample data generated according to an embodiment of the present application;

[0035] Figure 4 A technical flow diagram of a composition design method for encrypted multi-element alloys based on machine learning provided in one embodiment of the present application;

[0036] Figure 5 A schematic diagram of encrypted alloy data of a portion of gold-based bonding materials provided in an embodiment of the present application;

[0037] Figure 6A schematic diagram of a principal component analysis projection of target performance provided by an embodiment;

[0038] Figure 7 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0039] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0040] As electronic devices develop towards miniaturization and high performance, the performance requirements for alloy materials are becoming higher and higher. Traditional alloy materials can no longer meet these new demands. However, there are some limitations in the traditional trial-and-error method of developing alloys. First, the preparation process of alloy materials is complex and relies on a large number of experimental trial and error. This method is not only inefficient but also costly. Secondly, the performance optimization of alloy materials lacks systematic theoretical guidance and often relies on the experience and intuition of researchers, which limits the space for improving material performance. In order to solve these problems, researchers began to explore new alloy material design methods. The introduction of the Materials Genome Project provides new ideas for the research and development of alloy materials. The program aims to quickly screen and optimize the performance of materials through high-throughput experiments and computational methods, and accelerate the discovery and development of new materials. Machine learning algorithms play an important role in this process. They can process a large amount of experimental data, discover the intrinsic connection between the data, and provide a theoretical basis for the design of alloy materials. Material design methods based on machine learning have been shown to play an important role in the prediction of alloy material performance and composition design. Machine learning models can process a large amount of experimental data, discover the intrinsic connection between the data, and provide a theoretical basis for the design of alloy materials.

[0041] In the process of material research and development, data security and privacy protection are becoming increasingly important. With the strengthening of data privacy regulations around the world, how to balance the open sharing of data and privacy protection in material research and development has become an urgent problem to be solved. Privacy computing provides a solution that allows data analysis and calculation under the premise of protecting data from being leaked. The key to this technology is to ensure the security and privacy of data during processing, so that data owners can share data with confidence without worrying about the leakage of sensitive information. Applying privacy computing to the research and development of alloy materials can provide material scientists with a secure data sharing platform. On this platform, researchers can use encrypted data from different sources to design the composition and predict the performance of alloy materials without worrying about the leakage of original data. This not only promotes open sharing and collaborative innovation of data, but also brings new opportunities for the research and development of alloy materials. Especially in the field of precious metal multi-element alloy materials, the application of privacy computing can protect alloy composition and process parameters involving commercial secrets. Through privacy computing, enterprises can share data with other research institutions or upstream and downstream partners in the industrial chain without leaking core technologies, and jointly promote the advancement of material technology. However, the application of privacy computing technology in alloy material research and development is still in its infancy. How to effectively integrate these technologies into the existing material research and development process, how to ensure efficiency and accuracy in the calculation process, and how to formulate corresponding technical standards and specifications are all current challenges.

[0042] The purpose of this application is to overcome the blindness of the experimental trial and error method and protect data privacy, and to provide a composition design technology for encrypted multi-element alloys based on machine learning. This technology is a data-driven material genetic method that designs alloy materials while ensuring data privacy.

[0043] This application aims to solve the above problems. For multi-element alloy materials, by combining privacy computing technology with the composition design method of alloy materials based on machine learning, a solution that can both protect data privacy and accelerate the material research and development process is provided. Through this application, it is hoped that collaborative innovation in alloy material research and development can be promoted while ensuring data privacy security.

[0044] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0045] In an exemplary embodiment, Figure 1As shown, a composition design method for encrypted multi-element alloys based on machine learning is provided. The method is executed by a computer device, specifically, it can be executed by a computer device such as a terminal or a server alone, or it can be executed by a terminal and a server together. In the embodiment of the present application, the method is applied to a server as an example for explanation. The execution subject of the present embodiment is the material designer. The technical personnel in this field can also use the data holder as the execution subject in actual application. The present embodiment does not limit this, and includes the following steps S1 to S5. Among them:

[0046] S1. Obtain the encrypted multi-element alloy composition and target performance requirements sent by the data holder; the encrypted multi-element alloy composition is encrypted by encrypting the element atomic parameter characteristic table.

[0047] The data holder prepares the composition of the multi-element alloy and the corresponding target performance requirement data for alloy material design. The data holder encrypts the atomic characteristic table of the element through a self-selected monotonic function, which is private to the data holder and used for subsequent decryption. The data holder provides the encrypted data set to the material designer.

[0048] The encryption method for encrypting multi-element alloy components includes: encrypting an element atomic parameter characteristic table using a monotonic function to obtain an encrypted element atomic parameter characteristic table; and encrypting the multi-element alloy components using the encrypted element atomic parameter characteristic table to obtain an encrypted multi-element alloy component.

[0049] In this embodiment, the composition of the multi-element alloy is encrypted by encrypting the element atomic parameter characteristic table. The specific principle is as follows: suppose each chemical element of the alloy has m atomic parameters, and the atomic parameter characteristic matrix of the chemical element before encryption is X, X is a k×m matrix, the total number of elements is k, and the row X of the matrix is i Contains information on m atomic parameters of a chemical element i, such as atomic radius, electronegativity, ionization energy, density, etc. i =[x i1 ,…,x ij ,…,x im ], where x ij Represents the jth feature of chemical element i. The data encryption party sets a monotonic function f(X) to encrypt the atomic parameter features, such as f(X) = k*3X+c, where k and c are constants. X' = f(X) is the encrypted atomic parameter feature matrix, where each row X' i is the encrypted atomic characteristic parameter of an element, i is a chemical element. f is kept confidential by the data encryption party and can be a linear or nonlinear monotonic function. After the original alloy data is encrypted using this function, it is provided to data mining and material design parties. The encryption process diagram is shown in Figure 2 shown.

[0050] S2. Based on the encrypted multi-element alloy composition, a plurality of to-be-predicted encrypted multi-element alloys with different component mass proportions are generated by permutation and combination.

[0051] In this embodiment, this step can also be referred to as a high-throughput virtual sample generation process:

[0052] If the encryption party provides the modeling party with an encrypted element atomic parameter characteristic table, the forward high-throughput virtual sample can be generated by permuting and combining different positions and alloy proportions in the alloy encryption data set provided by the data holder. For example, a ternary gold alloy is represented by Au (1-x-y) B x C y , and in the training set, it is assumed that there may be 2 different elements (Cu or Fe) at the B position, and the added content is 0.01 or 0.02ppm. Assuming that there may be 2 elements (Al or Mg) at the C position, and the addition ratio can only be 0.03ppm, then the permutation combination space is 2*2*2*1=8. The addition ratios and encrypted features corresponding to the generated 8 virtual sample data are as follows Figure 3 shown.

[0053] S3. Calculate the characteristics of each predicted encrypted multi-element alloy according to the encrypted multi-element alloy components and component mass ratios. The material designer calculates the characteristics by a weighted characteristic calculation method.

[0054] According to the encrypted multi-element alloy components and component mass ratios, the characteristics of each to-be-predicted encrypted multi-element alloy are calculated by a weighted method.

[0055] The method for calculating the characteristics of the multi-element alloy to be predicted is: using a weighted method to calculate the weighted average value and characteristic dispersion of each characteristic of the multi-element alloy to be predicted.

[0056] The calculation formula for the weighted average value and characteristic dispersion of each feature of the predicted encrypted multi-element alloy is:

[0057]

[0058] in, represents the weighted average of the jth characteristic of the alloy, δf j Indicates the discreteness of the feature, c i represents the mass ratio of the ith element in the n-element alloy, f ij Represents the value of the jth feature of the ith element.

[0059] S4. Input the characteristics of each of the encrypted multi-element alloys to be predicted into the trained encrypted alloy performance prediction model to obtain corresponding performance data; the encrypted alloy performance prediction model is trained by the characteristics of the encrypted training multi-element alloy and the training performance of the encrypted training multi-element alloy.

[0060] S5. Send the encrypted component mass ratio of the multi-element alloy components corresponding to the performance data that meets the target performance requirements to the data holder.

[0061] The training method of the encrypted alloy performance prediction model includes:

[0062] Obtain the encrypted training multi-element alloy composition, the encrypted training multi-element alloy component mass ratio and training performance sent by the data holder.

[0063] The characteristics of the encrypted training multi-element alloy are calculated according to the encrypted training multi-element alloy composition and the mass ratio of each component of the encrypted training multi-element alloy.

[0064] The encrypted training multi-element alloy features are used as input and the training performance is used as output to train a machine learning model to obtain an encrypted alloy performance prediction model. The machine learning model is a regression model.

[0065] Compared with the prior art, this embodiment has the following obvious outstanding substantial features and significant advantages:

[0066] 1. The encryption technology in this embodiment does not affect the physical and chemical laws in the original multi-element alloy data set;

[0067] 2. The encrypted feature calculation technology used in the method of this embodiment can meet the consistency of the rotation symmetry of multi-element alloys, that is, the feature vectors calculated for ABC, BCA and CAB alloys are consistent;

[0068] 3. The method of this embodiment combines qualitative pattern recognition and quantitative regression analysis to reduce prediction risks;

[0069] 4. The generation of high-throughput virtual samples in this embodiment is achieved by arranging and combining the features of the elements in the original data set, and no additional encryption process is required.

[0070] In an exemplary embodiment, Figure 4 As shown, a composition design method for encrypted multi-element alloys based on machine learning is provided. The technology is a data-driven material gene method. The alloy material is designed while ensuring data privacy, including the following steps:

[0071] 1) Data holders prepare the composition of multi-element alloys and corresponding target performance data for alloy material design;

[0072] 2) The data holder encrypts the atomic characteristic table of the element through a self-selected monotonic function, which is private to the data holder and used for subsequent decryption. The data holder provides the encrypted data set to the material designer;

[0073] 3) The material designer calculates the features using a weighted feature calculation method;

[0074] 4) Machine learning algorithm modeling and pattern recognition;

[0075] 5) The material designer generates and predicts high-throughput virtual samples and provides an encrypted form of the designed samples;

[0076] 6) The data holder decrypts and conducts experimental synthesis and characterization based on the recommended samples provided by the material designer;

[0077] The encryption method in step 2) is to encrypt the atomic characteristics of the element through a monotonic function selected by the data holder, and the encrypted monotonic function is private to the data holder.

[0078] The weighted feature calculation method in step 3) is to perform weighted feature calculation on the encrypted features.

[0079] Only the data holder holds the encryption function. The sharing here mainly refers to sharing between a single data holder and the material designer, rather than sharing between multiple data holders.

[0080] Step 5) Predict and give the encrypted form of the sample involved. The material designer models based on the encrypted data, so the data before encryption or the encryption function is not needed. For example, the original data x is 1, 3, 5, and is encrypted into 2, 6, 10 using a 2x function. The designer only uses the data of 2, 6, 10 for modeling and returns the design result of the same form as 2, 6, 10, such as 20. The data holder can know from the result that the designed sample is x=10.

[0081] The generation of high-throughput virtual samples in step 4) is achieved by permuting and combining the encrypted feature vectors.

[0082] As a specific implementation method,

[0083] 1) Dataset encryption and decryption

[0084] The composition of the multi-element alloy is encrypted by encrypting the atomic parameter characteristic table. The specific principle is as follows: suppose each chemical element of the alloy has m atomic parameters, and the atomic parameter characteristic matrix of the chemical element before encryption is X, which is a k×m matrix with a total number of elements of k, where the row X of the matrix iContains information on m atomic parameters of a chemical element i, such as atomic radius, electronegativity, ionization energy, density, etc. i =[x i1 ,…,x ij ,…,x im ], where x ij Represents the jth feature of chemical element i. The data encryption party sets a monotonic function f(X) to encrypt the atomic parameter features, such as f(X) = k*3X+c, where k and c are constants. X' = f(X) is the encrypted atomic parameter feature matrix, where each row X' i is the encrypted atomic characteristic parameter of an element, i is a chemical element. f is kept confidential by the data encryption party, and can be a linear or nonlinear monotonic function. After the original alloy data is encrypted using this function, it is provided to data mining and material design parties. Figure 2 A schematic diagram of the encryption process.

[0085] 2) Weighted feature calculation

[0086] The characteristics of multiple alloy elements are calculated by weighting the characteristics. Specific principle: Assume that the ratio of the i-th element in the n-element alloy is c i , the value of the jth feature of the i-th element is f ij The weighted average of the jth characteristic of the alloy with this characteristic is calculated by weighted method. and the discreteness of the feature δf j .

[0087]

[0088] The advantage of this feature calculation method is that it maps the features of multi-component alloys to a non-sparse feature space and ensures the rotation symmetry of the alloys, that is, the weighted feature calculation results of the chemically identical alloys ABC, BCA and ACB are the same.

[0089] 3) High-throughput virtual sample generation

[0090] If the encryption party provides the modeling party with an encrypted atomic parameter table, the forward high-throughput virtual sample can be generated by permuting and combining different positions and alloy proportions in the alloy encryption data set provided by the data holder. For example, a ternary gold alloy is represented by Au (1-x-y) B x C y , and in the training set, it is assumed that there may be 2 different elements (Cu or Fe) at the B position, and the added content is 0.01 or 0.02ppm. Assuming that there may be 2 elements (Al or Mg) at the C position, and the addition ratio can only be 0.03ppm, then the permutation combination space is 2*2*2*1=8. The addition ratios and encrypted features corresponding to the generated 8 virtual sample data are as follows Figure 3 shown.

[0091] 4) Machine learning algorithm modeling and pattern recognition

[0092] The supervised learning algorithms such as support vector machine regression are used to build a regression model with the target performance to be optimized as the dependent variable and the calculated weighted features as the independent variable. The pattern recognition part uses principal component analysis technology to help understand the relationship between the independent variable and the alloy performance by visualizing the dimensionality reduction of the independent variable space, and find the range of alloys that meet the target performance in this independent variable space.

[0093] 5) Recommended experimental ingredients

[0094] The machine learning model is used to predict the virtual samples generated in 3), and the encrypted alloy composition that meets the design requirements is selected as an experimental recommendation and given to the data holder. The data holder can obtain the alloy composition formula through reverse decryption and conduct experiments.

[0095] This embodiment uses an encryption technology that does not affect the physical and chemical laws in the original multi-element alloy data set to encrypt the data while ensuring the privacy of the data holder, and designs the composition of the multi-element alloy based on the encrypted data set, which helps to accelerate the cooperative research and development of alloy materials.

[0096] In an exemplary embodiment, a composition design method for encrypted multi-element alloys based on machine learning is provided. This embodiment is basically the same as the above embodiment, with the special feature that: the data set in step 1) is encrypted and decrypted. The data is data of gold-based bonding materials of different compositions obtained from 38 experiments; wherein Target1 and Target2 are the target performance of the material to be optimized, and the higher the target performance, the better. B% and C% are the addition amounts of a certain microalloying element. Bi and Ci (i=1,2,3) are the encrypted data of atomic radius, electronegativity and ionization energy of B-position elements and C-position elements, respectively. The design requirement is that the performance of Target1 and Target2 is as high as possible. The encrypted alloy data of some gold-based bonding materials are as follows: Figure 5 shown.

[0097] The step 5) recommends experimental components. After calculating the virtual samples according to the weighted features, they are input into the constructed support vector regression model to obtain the predicted values ​​of Target1 and Target2 for each virtual sample, and finally the virtual samples with the highest predicted values ​​of Target1 and Target2 are selected as the recommended experimental samples.

[0098] In an exemplary embodiment, Figure 4As shown, a composition design method for encrypted multi-element alloys based on machine learning is provided. This embodiment is basically the same as the previous embodiment, with the special features that:

[0099] In the step 4), pattern recognition is added, and principal component analysis is used to define the top 25% of the target performance data as excellent samples, and the remaining 75% of the samples with poor performance as inferior samples. The data is standardized to avoid the influence of data scale and unit on the generated features. For feature x i The normalization formula is as follows:

[0100]

[0101] where u i is the x of the data set i The average value, σ i is the x of the data set i The standard deviation of the feature;

[0102] Select a rectangular area with a large intersection of multiple target performance excellent classes. Specifically, search for a rectangular box in which the proportion of excellent class samples of multiple target performance is the highest and the number of samples in the rectangular box is greater than a certain threshold. As the excellent class area, all high-throughput generated virtual samples are also projected onto the two-dimensional plane in this way, and the virtual samples in the excellent class area are selected and combined with the predicted values ​​of the machine learning model as the recommended experimental components, such as Figure 6 As shown, Figure 6 a) is the projection of target1, and b) is the projection of target2. (× is a good sample, · is a bad sample, and the threshold of the number of samples in the rectangular box is 3).

[0103] This embodiment uses a composition design technology for encrypted multi-element alloys based on machine learning to guide the composition design of multi-element alloys. This embodiment uses principal component analysis to add pattern recognition technology to the composition design of alloys on the basis of machine learning regression modeling, and designs the alloy composition from both qualitative and quantitative perspectives, thereby improving the reliability of the design.

[0104] In summary, the above-mentioned embodiment is a composition design technology for encrypted multi-element alloys based on machine learning, which includes the following steps: the material data holder encrypts the multi-element alloy data and provides it to the material designer; the material designer calculates the characteristics of the alloy material and generates a high-throughput virtual sample based on the encrypted data; constructs a corresponding machine learning model and pattern recognition model with multiple target performances; and selects and recommends experimental samples based on the predicted value of the model and the superior region of the model recognition model.

[0105] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 7 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a composition design method for encrypted multi-element alloys based on machine learning is implemented.

[0106] Those skilled in the art will understand that Figure 7 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0107] In an exemplary embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0108] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0109] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0110] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0111] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0112] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0113] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0114] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A composition design method for encrypted multi-element alloys based on machine learning, characterized in that: The composition design method for encrypted multi-element alloys based on machine learning includes: Obtaining the encrypted multi-element alloy composition and target performance requirements sent by the data holder; the encrypted multi-element alloy composition is encrypted by encrypting the element atomic parameter characteristic table; According to the encrypted multi-element alloy composition, a plurality of to-be-predicted encrypted multi-element alloys with different component mass ratios are generated by permutation and combination; Calculate the characteristics of each predicted encrypted multi-element alloy according to the encrypted multi-element alloy components and component mass ratios; Input the characteristics of each of the encrypted multi-element alloys to be predicted into a trained encrypted alloy performance prediction model to obtain corresponding performance data; the encrypted alloy performance prediction model is obtained by training the characteristics of the encrypted training multi-element alloy and the training performance of the encrypted training multi-element alloy; The encrypted component mass ratio of the multi-element alloy component corresponding to the performance data meeting the target performance requirement is sent to the data holder.

2. The composition design method for encrypted multi-element alloys based on machine learning according to claim 1, characterized in that: The training method of the encrypted alloy performance prediction model includes: Obtain the encrypted training multi-element alloy composition, the mass ratio of each component of the encrypted training multi-element alloy and the training performance sent by the data holder; Calculating the characteristics of the encrypted training multi-element alloy according to the encrypted training multi-element alloy composition and the mass ratio of each component of the encrypted training multi-element alloy; The encrypted training multi-element alloy features are used as input, and the training performance is used as output to train a machine learning model to obtain an encrypted alloy performance prediction model.

3. The composition design method for encrypted multi-element alloys based on machine learning according to claim 2, characterized in that: The machine learning model is a regression model.

4. The composition design method for encrypted multi-element alloys based on machine learning according to claim 1, characterized in that: The encryption method for encrypting the multi-element alloy composition comprises: The element atomic parameter characteristic table is encrypted by using a monotonic function to obtain an encrypted element atomic parameter characteristic table; The encrypted element atomic parameter characteristic table is used to encrypt the multi-component alloy composition to obtain the encrypted multi-component alloy composition.

5. The composition design method for encrypted multi-element alloys based on machine learning according to claim 1, characterized in that: According to the encrypted multi-element alloy components and component mass ratios, the characteristics of each to-be-predicted encrypted multi-element alloy are calculated, specifically including: According to the encrypted multi-element alloy components and component mass ratios, the characteristics of each to-be-predicted encrypted multi-element alloy are calculated by a weighted method.

6. The composition design method for encrypted multi-element alloys based on machine learning according to claim 1, characterized in that: The method for calculating the characteristics of the multi-element alloy to be predicted is: using a weighted method to calculate the weighted average value and characteristic dispersion of each characteristic of the multi-element alloy to be predicted.

7. The composition design method for encrypted multi-element alloys based on machine learning according to claim 1, characterized in that: The calculation formula for the weighted average value and characteristic dispersion of each feature of the predicted encrypted multi-element alloy is: in, represents the weighted average of the jth characteristic of the alloy, δf j Indicates the discreteness of the feature, c i represents the mass ratio of the ith element in the n-element alloy, f ij Represents the value of the jth feature of the ith element.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the composition design method for encrypted multi-element alloys based on machine learning as described in any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the composition design method for encrypted multi-element alloys based on machine learning described in any one of claims 1 to 7 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the composition design method for encrypted multi-element alloys based on machine learning described in any one of claims 1 to 7 is implemented.

Images