Artificial intelligence-based method and system for generating composition-temperature phase diagram of ferroelectric material

Abstract

The present invention relates to an artificial intelligence-based method and system for generating a composition-temperature phase diagram of a ferroelectric material. The method comprises: acquiring a variable-doped ferroelectric material chemical formula, and setting same in the form of a character string; setting variation ranges and step sizes for both the doping ratio and the temperature; processing chemical elements and doping ratios in the variable-doped ferroelectric material chemical formula in the form of the character string to obtain a chemical formula numerical vector list; on the basis of the variation range and step size set for the temperature, forming a temperature list, and performing normalization processing to obtain a normalized temperature list; performing merging on the basis of the chemical formula numerical vector list and the normalized temperature list, and inputting data into a pre-trained crystalline phase transition artificial intelligence model to output a crystal structure type; and on the basis of the crystal structure type, drawing composition-temperature phase diagrams of the variable-doped ferroelectric material chemical formula under different doping ratios and different temperature conditions. Compared with the prior art, the present invention has the advantages such as automatically generating component-temperature phase diagrams.

Description

An artificial intelligence generation method and system of a ferroelectric material composition-temperature phase diagram TECHNICAL FIELD

[0001] The present application relates to the technical field of new materials, in particular to an artificial intelligence generation method and system of a ferroelectric material composition-temperature phase diagram. BACKGROUND

[0002] With the rapid development of new generation information technology, new energy, integrated circuits and other frontier fields, the demand for electronic functional materials has increased dramatically. Ferroelectric materials, as a basic electronic functional material, are widely used in piezoelectric transduction, infrared sensing, energy conversion and other fields. Traditional typical ferroelectric materials such as barium titanate and lead titanate materials have been difficult to meet the rich requirements of material performance in frontier fields. Therefore, finding a feasible solution to accurately and quickly develop new high-performance ferroelectric materials is a key technology that needs to be solved urgently. Researchers have found that in the composition-temperature phase diagram of ferroelectric materials, the ferroelectric materials at the intersection of multiple crystal structures, i.e. at the phase boundary, have multiple excellent electrical properties. For example, in the case of calcium and zirconium co-doped barium titanate material, at room temperature, the doping ratio at the cubic, tetragonal and trigonal phase boundary generally improves the piezoelectric coefficient, dielectric constant and other electrical properties of the pure barium titanate material by more than one order of magnitude. Designing doped ferroelectric materials based on the phase boundary of multiple crystal structures has become a guiding principle for the rapid development of ferroelectric materials in this field. Therefore, accurately and quickly constructing a composition-temperature phase diagram of ferroelectric materials can directly determine the interface of ferroelectric materials among multiple phases and directly synthesize ferroelectric materials with high electrical properties. However, in the traditional method of constructing a composition-temperature phase diagram, experiments are designed, a large number of materials with different doping ratios are synthesized, and the crystal structure and phase transition temperature of each material in the series are characterized one by one to draw a composition-temperature phase diagram of doped ferroelectric materials, and accordingly to find the multi-phase boundary. For a large number of ferroelectric materials with different doping ratios and types, the traditional method cannot achieve rapid phase diagram construction and phase boundary searching.

[0003] The patent application CN115410654A discloses a machine learning prediction method for ferroelectric-paraelectric phase transition temperature of molecular ferroelectrics, collects chemical formula of molecular ferroelectrics and experimental value of ferroelectric-paraelectric phase transition temperature thereof from literature; calculates using literature, first principle calculation and topological structure to obtain feature parameters for describing sample characteristics; quickly reduces dimension of features, then combines subset search strategy and embeds learning ware to obtain subset of optimal features; takes ferroelectric-paraelectric phase transition temperature as target variable and optimal features as independent variable, uses support vector machine regression algorithm to establish quantitative prediction model for ferroelectric-paraelectric phase transition temperature of molecular ferroelectrics; collects new molecular ferroelectric samples, obtains feature parameters of new samples, and predicts ferroelectric-paraelectric phase transition temperature of new molecular ferroelectric samples according to the quantitative prediction model. The input data of the method needs information such as first principle calculation and topological structure, and also needs to obtain feature parameters of sample characteristics, in addition, the method does not involve prediction of crystal structure type, generation of component-temperature phase diagram of ferroelectric materials, and automatic generation of component-temperature phase diagram of ferroelectric materials. SUMMARY

[0004] The purpose of the present application is to provide an artificial intelligence generation method and system for a component-temperature phase diagram of ferroelectric materials that realizes automatic generation of a component-temperature phase diagram.

[0005] The purpose of the present application can be achieved by the following technical solutions:

[0006] An artificial intelligence generation method for a component-temperature phase diagram of ferroelectric materials, comprising the following steps:

[0007] Obtain a chemical formula of a doped variable ferroelectric material and set it as a string form, wherein the chemical formula of the doped variable ferroelectric material includes chemical elements and corresponding doping proportions, and the doping variable is contained in the doping proportion;

[0008] Based on the chemical formula of the doped variable ferroelectric material in the string form, set the change range and step length of the doping proportion and temperature;

[0009] According to the set change range and step length of the doping proportion, process the chemical elements and doping proportions in the chemical formula of the doped variable ferroelectric material in the string form to obtain a chemical formula numerical vector list;

[0010] According to the set change range and step length of the temperature, form a temperature list and perform normalization processing to obtain a normalized temperature list;

[0011] Based on the chemical formula numerical vector list and the normalized temperature list, merge and input into a pre-trained crystal phase transition artificial intelligence model as input data, and output a crystal structure type;

[0012] Based on the crystal structure type, the component-temperature phase diagram of the doped variable ferroelectric material chemical formula under different doping ratios and different temperature conditions is drawn.

[0013] Further, the string form of the doped variable ferroelectric material chemical formula uses symbols to distinguish chemical elements and doping ratios, specifically:

[0014] Only chemical elements are placed in symbols, doping ratios are placed on the side of corresponding chemical elements, or both chemical elements and doping ratios are placed in different symbols, wherein the doping variable is contained in the doping ratio.

[0015] Further, the obtaining step of the chemical formula numerical vector list comprises:

[0016] Based on the string form of the doped variable ferroelectric material chemical formula, according to the set change range and step length of the doping ratio, chemical formulas with different doping ratios are batch generated to form a chemical formula list with different doping ratios, wherein the batch generation quantity is:

[0017] In the formula, N1 is the number of batch-generated chemical formulas, R1 and R2 are the upper and lower limits of the doping ratio respectively, and ΔR is the step length;

[0018] Based on the chemical formula list with different doping ratios, the doping ratio is normalized by using a logarithmic normalization method to obtain a logarithmically normalized doping ratio:

[0019] The chemical elements and the logarithmically normalized doping ratio in the chemical formula list are embedded and represented by using the periodic table of elements to obtain an initial chemical formula numerical vector list, wherein the expression for embedding and representing is:

[0020] In the formula, v[j] is the jth initial chemical formula numerical vector, m′ i is the logarithmically normalized doping ratio of the ith chemical element, E j is the jth chemical formula containing n elements, ε i is the element list in the periodic table of elements;

[0021] Based on the initial chemical formula numerical vector list, principal component analysis algorithm is used for dimension reduction to obtain a final chemical formula numerical vector list, wherein the dimension reduction is represented as: p z j = v k U

[0022] In the formula, z p is the dimension-reduced chemical formula numerical vector, v j is the jth initial chemical formula numerical vector, U kThe low-dimensional base matrix.

[0023] Further, the step of obtaining the logarithmically normalized doping ratio comprises:

[0024] Based on the chemical formula list of the different doping ratios, the doping ratio is normalized, and the expression of the normalization is:

[0025] In the formula, m * i is the normalized doping ratio coefficient of the i th chemical element, m i is the doping ratio of the i th chemical element.

[0026] The normalized doping ratio is logarithmically transformed to obtain the logarithmically normalized doping ratio, and the expression of the logarithmically normalized doping ratio is: m′ i = log(1+m * i )

[0027] In the formula, m′ i is the logarithmically normalized doping ratio of the i th chemical element.

[0028] Further, the step of obtaining the normalized temperature list comprises:

[0029] According to the set change range, step length of the temperature, batch generate each temperature point, form a temperature list, wherein each temperature point represents the total number of calculations of each chemical formula value vector in the chemical formula value vector list, and the calculation expression is:

[0030] In the formula, N2 is the total number of calculations of each chemical formula value vector at different temperature points, T1 and T2 are the upper limit and lower limit of the temperature respectively, and ΔT is the step length;

[0031] Based on the temperature list, the maximum and minimum normalization algorithm is used for normalization processing to obtain a normalized temperature list, and the expression of the maximum and minimum normalization algorithm is:

[0032] In the formula, T′ i is the i th normalized temperature point, T i is the i th original temperature, T min , T max are the minimum and maximum values of the temperature respectively.

[0033] Further, the crystal phase transformation artificial intelligence model comprises an input layer, a hidden layer and an output layer, the input layer is used for inputting input data obtained by merging the chemical formula numerical vector list and the normalized temperature list, the hidden layer is used for extracting deep-level implicit features of the input data, and the output layer is used for classification prediction based on the extracted deep-level implicit features, and outputting a predicted crystal structure type.

[0034] Further, the hidden layer is provided with multiple layers, each layer of the hidden layer is provided with a batch normalization operation and a dropout operation, and the execution steps of the hidden layer comprise:

[0035] In the first layer of the hidden layer, the input data is processed to obtain an initial output vector h 1 of the first layer of the hidden layer 1 = ReLU(W (1) Q + b (1) )

[0036] wherein W (1) is a weight matrix of the first layer of the hidden layer, Q is an input vector, b (1) is a bias vector of the first layer of the hidden layer;

[0037] The initial output vector h 1 of the first layer is subjected to a batch normalization operation to obtain a batch-normalized initial output vector

[0038] wherein γ, β and ε are training parameters, μ batch and are the mean and variance of the current batch;

[0039] The batch-normalized initial output vector is subjected to a dropout operation to obtain a final output vector of the first layer of the hidden layer, and the final output vector of the first layer of the hidden layer is taken as an input of a next layer of the hidden layer, and the above steps are repeated until a processing process of all the layers of the hidden layer is completed.

[0040] Further, the final output vector of the first layer of the hidden layer is represented as:

[0041] wherein h is the final output vector of the first layer of the hidden layer, and m is a mask.

[0042] Further, the step of outputting the crystal structure type comprises:

[0043] Based on the chemical formula numerical vector list and the normalized temperature list, a total number of executions of the pre-trained crystal phase transformation artificial intelligence model is calculated, and a calculation expression of the total number of executions is: N=N1×N2

[0044] In the formula, N is the total number of executions, N1 is the number of chemical formula value vectors in the chemical formula value vector list, that is, the number of chemical formulas with different doping proportions, and N2 is the number of temperature points in the normalized temperature list, that is, the total number of calculations of each chemical formula value vector at different temperature points;

[0045] The chemical formula value vectors in the chemical formula value vector list and the normalized temperatures in the normalized temperature list are combined as input data for one execution, and the pre-trained crystal phase transition artificial intelligence model is used for processing, to output the crystal structure type this time, and the step is repeated to perform N-1 times, to obtain the crystal structure types of all chemical formula value vectors at various temperature points.

[0046] The application also provides an artificial intelligence generation system for a ferroelectric material composition-temperature phase diagram, comprising:

[0047] A pre-trained artificial intelligence module is used to obtain a ferroelectric material crystal phase transition dataset, train a multilayer artificial neural network model, perform hyperparameter optimization and cross-validation in the training process, and obtain a pre-trained crystal phase transition artificial intelligence model;

[0048] A phase diagram automatic generation module is used to read a ferroelectric material chemical formula with a doping variable, a doping proportion and a temperature variation range, a step size, and perform standardization processing, cyclically call the pre-trained crystal phase transition artificial intelligence model for processing, output a crystal structure type, and draw a composition-temperature phase diagram of the ferroelectric material chemical formula with the doping variable under different doping proportions and different temperature conditions according to the crystal structure type, wherein the ferroelectric material chemical formula with the doping variable comprises chemical elements and corresponding doping proportions, and the doping variable is included in the doping proportion.

[0049] Compared with the prior art, the application has the following beneficial effects:

[0050] (1) The application sets the doping proportion and temperature of the ferroelectric material chemical formula with a doping variable to generate chemical formulas with different doping proportions, and uses a crystal phase transition artificial intelligence model to predict the crystal structure type by combining with different temperature points, so that a composition-temperature phase diagram under different doping proportions and different temperature conditions can be drawn, the application only needs to input the chemical formula of the ferroelectric material, without any additional information such as material properties, structure, atoms, etc., not only realizes the automatic generation of the ferroelectric material composition-temperature phase diagram, but also reduces the generation difficulty.

[0051] (2) By using the pre-trained crystal phase transition artificial intelligence model, the model does not need to be optimized by adjusting parameters and the like when predicting the crystal structure of the ferroelectric material, and the execution efficiency is improved.

[0052] (3) Compared with the traditional experimental method of constructing the component-temperature phase diagram, the artificial intelligence automatic generation method of the present application greatly shortens the time consumed for generating a complete component-temperature phase diagram and reduces the research and development cost. BRIEF DESCRIPTION OF DRAWINGS

[0053] Fig. 1 is a flowchart of the artificial intelligence generation method of the present application;

[0054] Fig. 2 is a schematic diagram of the artificial intelligence generation system of the present application;

[0055] Fig. 3 is a schematic diagram of the data processing flow of the chemical formula of the ferroelectric material of the present application;

[0056] Fig. 4 is the accuracy of the pre-trained crystal phase transition artificial intelligence model in the training set and the validation set of the present application;

[0057] Fig. 5 is the component-temperature phase diagram of the calcium-zirconium-doped barium titanate ferroelectric material (1-x)Ba(Zr 0.2 Ti 0.8 )O3-xBa 0.7 Ca 0.3 TiO3 generated in the present application;

[0058] Fig. 6 is the component-temperature phase diagram of the tin-doped barium titanate ferroelectric material BaSn x Ti 1-x O3 generated in the present application;

[0059] Fig. 7 is the component-temperature phase diagram of the lead zirconate titanate ferroelectric material PbZr 1-x Ti x O3 generated in the present application. DETAILED DESCRIPTION

[0060] The present application will be described in detail below in conjunction with the drawings and specific embodiments. The present embodiment is implemented on the premise of the technical solution of the present application, and gives a detailed implementation and specific operation process, but the protection scope of the present application is not limited to the following examples.

[0061] Example 1

[0062] The present embodiment provides an artificial intelligence generation method for a ferroelectric material component-temperature phase diagram, as shown in Fig. 1, which comprises the following steps:

[0063] (1) The natural language processing method is used to obtain the crystal phase transition data of the ferroelectric material from the existing literature, and the collected data is used to construct a ferroelectric material crystal phase transition data set.

[0064] The ferroelectric material crystal phase transition dataset is obtained by using natural language processing method or manual acquisition method from the accessible scientific research literature library. The dataset contains chemical formula, temperature, crystal structure, literature identification code, journal name, publication year, and source information. The crystal phase transition dataset is a linked list type data structure, which can also be an array, tree, graph, or other data structure.

[0065] (2) Build a multi-layer artificial neural network intelligent model to train the crystal structure transition data of ferroelectric materials. After cross-validation, the crystal phase transition artificial intelligence model is obtained.

[0066] The artificial neural network module includes a pre-trained multi-layer artificial neural network model, which is trained by the ferroelectric material crystal phase transition dataset.

[0067] The multi-layer artificial neural network model consists of one or more input layers, fully connected layers, and output layers. The data received by the input layer is processed data from the original data. The proportion of chemical elements in the original chemical formula is normalized. For a chemical formula, E i (i = 1, 2, 3…) represents a chemical formula containing n elements, and m i (i = 1, 2, 3…) represents the doping proportion of each element, which is calculated based on the following formula: * i (i = 1, 2, 3…),

[0068] According to the following formula, the proportion coefficient is logarithmically transformed to obtain the logarithmic normalized doping proportion m′ i (i = 1, 2, 3…), m′ i = log(1+m * i ) (2)

[0069] According to the following formula, the elements and proportions in the chemical formula are embedded in the element list ε i (i = 1, 2, 3…) in the periodic table of elements,

[0070] to obtain the numerical representation of the chemical formula. The original chemical formula is converted to a new vector v.

[0071] The vector v is calculated by principal component analysis algorithm according to its dimension, and the low-dimensional basis matrix U k is calculated according to the following formula to reduce it to the specified p dimension, and the reduced input vector z p is obtained; z p = v j Uk (4)

[0072] Temperature T in the original input data i The normalized temperature T' is calculated according to the Max-Min normalization algorithm as shown in the following formula:

[0073] The input vector z p The normalized temperature T' is merged into a new vector to form the input vector Q of the first layer neural network, and the dimension is d = p + 1;

[0074] Each layer of artificial neural network extracts deep hidden features of input data, and the features h extracted by the s-th layer neuron are calculated according to the following formula through the weight matrix W and the bias vector b; h s = ReLU(W (s) Q + b (s) ) (6)

[0075] Preferably, each layer of artificial neural network includes one or more neurons, activation functions, regularizers, batch normalizers, and droppers, which can improve the stability and generalization ability of the model; the hyperparameters in each layer of artificial neural network are obtained by grid search to obtain the hyperparameters with the highest accuracy; the artificial neural network module built is a multi-layer perceptron model for classification tasks, which can also be a model for other types of tasks; the last layer of artificial neural network uses an activation function to output the predicted class probability.

[0076] The output layer is the L-th layer, and the output vector o of the output layer is calculated according to the following formula, and the activation function is used to convert the output vector to the numerical data corresponding to the crystal structure type; o = W (L) h L-1 +b (L) (7)

[0077] The activation function used is the Softmax function, and the total number of crystal structure types is C. The probability of predicting the crystal structure numerical type is calculated by the following formula: The maximum probability is taken The crystal structure corresponding to the current input vector is taken as the final output of the artificial neural network model.

[0078] The accuracy of the artificial neural network is evaluated by loss function, accuracy and other evaluation indexes; cross-validation is performed on the training set and the validation set, which are obtained by dividing the crystal phase transition data set obtained in step (1) according to a certain proportion; the artificial neural network continuously optimizes the weight matrix W and the bias coefficient b with the increase of the training round. The accuracy of the crystal phase transition artificial intelligence model obtained after optimization is greater than 90%.

[0079] (3) Set the chemical formula of the ferroelectric material with doping variable to generate the phase diagram.

[0080] The data format of the set chemical formula is a string of one or more elements combined in the form of "element 1 [proportion 1] element 2 [proportion 2]", etc. This method enables the computer program to read the elements and proportions in the chemical formula; such structured strings can also be a combination of other symbols, such as using parentheses, commas, periods, semicolons, and other symbols to distinguish chemical elements and proportions, or using the form of placing elements in symbols, or using the form of placing elements and proportions in different symbols.

[0081] The doping variable refers to doping a chemical element in the matrix material, and the doping proportion x is used as the doping variable, which is included in the string of doping proportions. The string containing the doping variable can be converted into a form recognizable by the computer and can be subjected to mathematical operations.

[0082] (4) Set the upper and lower limits of the doping proportion as R1 and R2, and the step size as ΔR, and set the upper and lower limits of the temperature change as T1 and T2, and the step size as ΔT.

[0083] The change range of the doping proportion is the upper and lower limits of the component axis in the phase diagram to be generated; the change range of the specified temperature condition is between 0-2000K.

[0084] (5) According to the set range and step size of the doping proportion, calculate the proportion coefficient in the specified chemical formula of the ferroelectric material, and calculate the total number N1 of the chemical formula using the following formula,

[0085] After the calculation is completed, N1 chemical formulas are generated in batches, the numerical representation of the chemical formula is obtained using the embedding representation method, and the data dimension is optimized.

[0086] The batch-generated chemical formula is embedded by the periodic table of chemical elements, and the string of chemical elements and proportions is converted into a numerical list to realize the numerical representation of the chemical formula; and a logarithmic normalization algorithm is used to enhance the influence of trace doping elements.

[0087] Preferably, the doping proportion m of each element in the chemical formula is first calculated according to formula (1) i(i = 1, 2, 3…) is normalized to obtain the normalized doping ratio coefficient m * i (i = 1, 2, 3…), and the proportional coefficient is logarithmically transformed according to formula (2) to obtain the logarithmically normalized doping ratio m' i (i = 1, 2, 3…);

[0088] The principal component analysis, the neural network-based encoder and other algorithms can be used to reduce the dimension of the original data. The optimization of the data dimension can reduce the dimension of the original data and improve the feature representation of the data.

[0089] (6) According to the set temperature condition change range and step length, batch generate each temperature point to form a temperature list, and calculate the total number N2 of times that each chemical formula needs to be calculated at different temperature points by using the following formula:

[0090] The temperature is subjected to Min-Max normalization processing, and the normalized temperature list T'(i = 1, 2, 3…) is obtained according to formula (5).

[0091] (7) The reduced chemical formula numerical representation and the normalized temperature list are combined into a new numerical list as input data.

[0092] The input data is input into the pre-trained crystal phase transition artificial intelligence model one by one, and the model can display the output result one by one or display all the results after execution. The output result of the pre-trained crystal phase transition artificial intelligence model is a numerical representation of the crystal structure type, which is converted into a string of crystal structure by a crystal structure type converter.

[0093] (8) The total number N of times of calling the crystal phase transition artificial intelligence model is calculated according to the following formula, N = N1 x N2 (11)

[0094] Within the range of N times, the crystal phase transition artificial intelligence model is called one by one, the input data is read one by one, and the crystal structure under different doping ratios and temperatures is output.

[0095] (9) After N times of execution, based on the output result, a component-temperature phase diagram of the current material under different doping ratios and different temperature conditions is drawn.

[0096] This embodiment verifies the experiment using the method described above. The phase transition AI model used in the experiment is a multilayer perceptron model for classification tasks. Seven fully connected layers are set up, serving as one input layer, five hidden layers, and one output layer. The data received at the input layer is the processed form of the original data. Using the method described in this embodiment, after reading the chemical formula, the chemical formula is normalized, embedded, and dimensionality reduced to obtain a 20-dimensional chemical formula embedding vector and a normalized 1-dimensional temperature vector. In this embodiment, the phase transition dataset contains more than 10,000 data entries. A portion of the chemical formulas in the phase transition dataset, after processing, is shown in Figure 3.

[0097] By merging the 20-dimensional chemical formula embedding vector Q and the normalized 1-dimensional temperature vector, the input vector in the input layer becomes 21-dimensional, in the form Q = [Q1 Q2 …Q 20 temp] T (12)

[0098] In the first hidden layer, the ReLU activation function is specified; therefore, the output vector h of the first layer is... 1 for h 1 =ReLU(W (1) Q+b (1) (13)

[0099] Among them W (1) Let b be the weight matrix of the first layer. (1) This is the bias vector for the first layer. In the neurons of this layer, batch normalization and dropout operations are added to improve the model's generalization ability. Therefore, the output vector h... 1 Scaling and translation are performed using the following formulas:

[0100] Where μ batch and These represent the mean and variance of the current batch, while γ, β, and ε are trainable parameters. The batch-normalized output vector is obtained through calculation. In the discard operation, the mask m is calculated.

[0101] The output vector obtained after batch normalization and discarding operations is obtained. This vector is also the input data for the next layer of the neural network. In the subsequent second, third, fourth, and fifth hidden layers, the input vector is used to calculate the output vector using formulas (13) to (15). In the fifth hidden layer, the output vector is obtained as follows: The last layer of neurons is the output layer, calculated according to formulas (7) to (8). The output crystal structure is obtained.

[0102] Increase the training round, divide the training set and the validation set for cross-validation, and constantly optimize the weight matrix, bias vector and other hyperparameters in each layer of neurons, and finally obtain a crystal phase transition artificial intelligence model with an accuracy greater than 90% on the training set and the validation set. Figure 4 is the accuracy of the crystal phase transition artificial intelligence model built in this embodiment on the training set and the validation set with the training round.

[0103] Set the chemical formula of the ferroelectric material with doping variables for which the phase diagram is to be generated. In the computer program, input the chemical formula of three ferroelectric materials as shown in Table 1. The three materials are:

[0104] (1) The material numbered No. 1 is a calcium-zirconium-doped barium titanate ferroelectric material:

[0105] The chemical formula of the No. 1 material with variables is set as “(1-x)Ba(Zr 0.2 Ti 0.8 )O3-xBa 0.7 Ca 0.3 TiO3”. After standardization, it is obtained in the following form: “Ba[1-0.3x]Zr[0.2-0.2x]Ti[0.8+0.2x]O[3]Ca[0.3x]”, where x is a variable that adjusts the doping ratio.

[0106] The upper limit and the lower limit of the doping ratio range are set to 0.2 and 0.5, respectively.

[0107] The ratio change step is set to 0.002.

[0108] The upper limit and the lower limit of the temperature change range are set to 120K and 440K, respectively.

[0109] The temperature change step is set to 2K.

[0110] Table 1 Parameters for automatically generating three different material component-temperature phase diagrams

[0111] Execute the computer program to read the chemical formula with doping variables, doping range, and step information, and batch generate a list of chemical formulas with different doping ratios. In this material, there are 150 chemical formulas with different doping ratios. According to the generated chemical formula, the doping ratio is normalized using the logarithmic normalization method, and the elements and doping ratios are embedded using the periodic table of elements. The embedded representation of the chemical formula is reduced using the principal component analysis (PCA) method to obtain a 20-dimensional numerical representation of the chemical formula.

[0112] Generate a temperature list and normalize the temperature list using the Min-Max normalization method.

[0113] The 20-dimensional numerical values of the chemical formula and the temperature list after Min-Max normalization processing are combined as input data.

[0114] According to formulas (9) to (11), the total number of times of calling the pre-trained crystal phase transition artificial intelligence model is 24000. For each row of data in the input data, the crystal phase transition artificial intelligence model is called in sequence, and the crystal structure under different doping ratios and temperatures is output. After 24000 times of execution, the processing of all input data is completed.

[0115] According to the output results, the composition-temperature phase diagram of the calcium-zirconium-doped barium titanate ferroelectric material in the doping ratio range of 0.2-0.5 and the temperature range of 120-440K is drawn, and the results are shown in FIG. 5. Compared with the phase diagram constructed by Acosta, Matias (2014) and others through experiments, it can be seen that the phase boundary change trend in the phase diagram automatically generated in the embodiment of the application is consistent.

[0116] (2) The material numbered 2 is a tin-doped barium titanate material:

[0117] The chemical formula of the material numbered 2 is set as “BaSn x Ti 1-x O3”, and after standardization processing, the chemical formula is converted to the following form “Ba[1]Sn[x]Ti[1-x]O[3]”, where x is a variable for adjusting the doping ratio.

[0118] The upper limit and the lower limit of the doping ratio range are set as 0 and 0.3, respectively.

[0119] The ratio change step is set as 0.002.

[0120] The upper limit and the lower limit of the temperature change range are set as 150K and 440K, respectively.

[0121] The temperature change step is set as 2K.

[0122] Using the same processing method as the material numbered 1, according to formulas (9) to (11), the total number of times of calling the pre-trained crystal phase transition artificial intelligence model is calculated to be 21750. After sequential execution, the composition-temperature phase diagram of the tin-doped barium titanate ferroelectric material in the doping ratio range of 0-0.3 and the temperature range of 150-440K is drawn, and the results are shown in FIG. 6. The phase boundary change trend in the phase diagram is consistent with the experimental results of Liu, Wenfeng (2017) and others.

[0123] (3) The material numbered 3 is a lead zirconate titanate material:

[0124] The chemical formula of the material numbered 3 is set as “PbZr 1-xTi x After standardization, the chemical formula is converted to the following form: "Pb[1]Zr[1-x]Ti[x]O[3]", where x is a variable that adjusts the doping ratio.

[0125] The upper and lower limits of the doping ratio range are set to 0.35 and 0.46, respectively.

[0126] The step size of the ratio change is set to 0.001.

[0127] The upper and lower limits of the temperature change range are set to 500K and 690K, respectively.

[0128] The temperature change step size is set to 2K.

[0129] Using the same processing method as the material numbered 1, the total number of times the pre-trained crystal phase transition artificial intelligence model needs to be called is calculated to be 10450 times according to formulas (9) to (11). After each execution is completed, the composition-temperature phase diagram of the material in the doping ratio change range of 0.35-0.46 and 500-690K is plotted, as shown in Figure 7. The resulting phase boundary change trend is consistent with the experimental results published by A. Bouzid (2003) et al.

[0130] In this embodiment, the composition-temperature phase diagram of three different materials is automatically generated using the above-mentioned intelligent generation method of ferroelectric material composition-temperature phase diagram. Table 1 also lists the time required to construct the composition-temperature phase diagrams of the above-mentioned three different materials in this embodiment, which is less than 1000s. Compared with the time consumed by traditional experimental methods (usually 1-2 years), the above-mentioned method greatly shortens the research and development time and improves the research and development efficiency. Using high-performance graphics cards and processors can further shorten the execution time of computer programs.

[0131] Embodiment 2

[0132] The present embodiment provides an artificial intelligence generation system for ferroelectric material composition-temperature phase diagram, as shown in Figure 2, which comprises:

[0133] Pre-trained artificial intelligence module: used to obtain the crystal phase transition data set of the electric material, train the multi-layer artificial neural network model, and perform hyperparameter optimization and cross-validation during the training process to obtain the pre-trained crystal phase transition artificial intelligence model;

[0134] The phase diagram automatic generation module is used for reading a doped variable ferroelectric material chemical formula, a doping ratio and a change range of temperature, and a step, and performing standardization processing, cyclically calling a pre-trained crystal phase transition artificial intelligence model for processing, outputting a crystal structure type, and drawing a component-temperature phase diagram of the doped variable ferroelectric material chemical formula under different doping ratios and different temperature conditions according to the crystal structure type, wherein the doped variable ferroelectric material chemical formula includes chemical elements and corresponding doping ratios, and a doping variable is contained in the doping ratio.

[0135] The rest is as in Example 1.

[0136] If the above functions are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the part of the prior art that essentially contributes or the part of the technical solutions can be embodied in the form of a software product, which is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in various embodiments of the present application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various program code storage media.

[0137] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can 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. The solutions in the embodiments of the present application can be implemented in various computer languages, such as object-oriented programming languages Java and interpreted scripting language JavaScript.

[0138] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks.

[0139] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart block or blocks.

[0140] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks.

[0141] While the preferred embodiments of the application have been described, it should be understood that various modifications and changes can be made by those skilled in the art which follow in the spirit of the application and the scope of the appended claims. Therefor, the description is intended to cover all modifications and changes as fall within the scope of the application, together with all of the equivalents thereof.

[0142] Obviously, numerous modifications and variations of the present application are possible in light of the above teachings. It is therefore to be understood that within the scope of the appended claims and their equivalents, the application can be practiced otherwise than as specifically described.

Claims

1. An artificial intelligence generated method of ferroelectric material composition-temperature phase diagrams, characterized in that, The method comprises the following steps: obtaining a chemical formula of a doped variable ferroelectric material and setting it in a string form, wherein the chemical formula of the doped variable ferroelectric material comprises chemical elements and corresponding doping proportions, and the doping variable is contained in the doping proportion; setting a change range and a step length of the doping proportion and temperature based on the chemical formula of the doped variable ferroelectric material in the string form; processing the chemical elements and the doping proportion in the chemical formula of the doped variable ferroelectric material in the string form according to the set change range and step length of the doping proportion to obtain a chemical formula numerical vector list; forming a temperature list according to the set change range and step length of the temperature and performing normalization processing to obtain a normalized temperature list; merging the chemical formula numerical vector list and the normalized temperature list as input data and inputting them into a pre-trained crystal phase transition artificial intelligence model to output a crystal structure type; drawing a component-temperature phase diagram of the chemical formula of the doped variable ferroelectric material under different doping proportions and different temperatures based on the crystal structure type.

2. The artificial intelligence generating method of a ferroelectric material composition-temperature phase diagram according to claim 1, characterized by, The chemical formula of the doped variable ferroelectric material in the string form distinguishes the chemical elements and the doping proportion by symbols, specifically: only the chemical elements are placed in the symbols, and the doping proportion is placed on the side of the corresponding chemical elements, or both the chemical elements and the doping proportion are placed in different symbols, wherein the doping variable is contained in the doping proportion.

3. The artificial intelligence generating method of a ferroelectric material composition-temperature phase diagram according to claim 1, characterized by, The obtaining step of the chemical formula numerical vector list comprises: Based on the string form of the doped variable ferroelectric material chemical formula, according to the set change range and step of the doping ratio, a chemical formula with different doping ratios is batch generated to form a chemical formula list with different doping ratios, and the number of batch generation is: wherein N1 is the number of batch-generated chemical formulas, R1 and R2 are the upper limit and the lower limit of the doping proportion respectively, and ΔR is the step length; based on the chemical formula list of different doping proportions, the doping proportion is normalized by using a logarithmic normalization method to obtain a logarithmically normalized doping proportion: The chemical formula list is embedded with the chemical elements and the logarithmic normalized doping proportions of the chemical formula in the periodic table of elements to obtain an initial chemical formula numerical vector list, wherein the expression embedded is: where v[j] is the jth initialized chemical formula numerical vector, m' i is the logarithm normalized doping proportion of the ith chemical element, E j is the jth chemical formula containing n elements, ε i is the list of elements in the periodic table; based on the initial chemical formula numerical vector list, a principal component analysis algorithm is used for dimension reduction to obtain a final chemical formula numerical vector list, wherein the dimension reduction is represented as: z p = v j U k wherein z p is the reduced dimensionality chemical formula numerical vector, v j is the jth initialized chemical formula numerical vector, U k is the low-dimensional basis matrix.

4. The artificial intelligence generating method of a ferroelectric material composition-temperature phase diagram according to claim 3, characterized in that, The step of obtaining the logarithmically normalized doping proportion comprises: Based on the chemical formula list of the different doping ratios, the doping ratio is normalized, and the expression of the normalization is: wherein m * i is the normalized doping proportion coefficient of the i-th chemical element, m i is the doping proportion of the i-th chemical element; performing logarithmic transformation on the normalized doping proportion to obtain the logarithmically normalized doping proportion, and the expression of the logarithmically normalized doping proportion is: m' i = log(l + m * i ) where m' is the logarithmically normalized doping ratio of the i-th chemical element. i is the logarithmically normalized doping ratio of the i-th chemical element.

5. The artificial intelligence generating method of a ferroelectric material composition-temperature phase diagram according to claim 1, characterized in that, The step of obtaining the normalized temperature list comprises: According to the change range of the set temperature, the step, each temperature point is generated in batches, and a temperature list is formed, wherein each temperature point represents the total number of calculations of each chemical formula numerical vector in the chemical formula numerical vector list, and the calculation expression is: wherein N2 is the total number of calculations of each chemical formula numerical vector at different temperature points, T1 and T2 are the upper limit and the lower limit of the temperature respectively, and ΔT is the step length; Based on the temperature list, a maximum minimum normalization algorithm is used for normalization processing to obtain a normalized temperature list, wherein an expression of the maximum minimum normalization algorithm is: where T' is the normalized temperature i is the i-th normalized temperature point, T i is the i-th original temperature, T min , T max are the minimum and maximum temperatures, respectively.

6. The artificial intelligence generating method of a ferroelectric material composition-temperature phase diagram according to claim 1, wherein The crystal phase transition artificial intelligence model comprises an input layer, a hidden layer and an output layer, the input layer is used for inputting input data obtained by merging the chemical formula numerical vector list and the normalized temperature list, the hidden layer is used for extracting deep-level implicit features of the input data, and the output layer is used for classification prediction based on the extracted deep-level implicit features to output a predicted crystal structure type.

7. The artificial intelligence generating method of a ferroelectric material composition-temperature phase diagram according to claim 6, characterized in that, The hidden layer is provided with multiple layers, each hidden layer increases batch normalization operation and dropout operation, and the execution steps of the hidden layer comprise: In the first layer hidden layer, the input data is processed to obtain an initial output vector h of the first layer hidden layer 1 : h 1 = ReLU(W (1) Q + b (1) ) where W (1) is the weight matrix of the first hidden layer, Q is the input vector, b (1) is the bias vector of the first hidden layer; to the first layer h 1 performing a batch normalization operation to obtain a batch-normalized initial output vector where γ, β, and ε are training parameters, μ batch and is the mean and variance of the current batch; based on the batch-normalized initial output vector performing the dropout operation to obtain the final output vector of the first layer of the hidden layer and taking the final output vector as the input of the next layer of the hidden layer, and repeating the above steps until the processing process of all the hidden layers is completed.

8. The artificial intelligence generating method of a ferroelectric material composition-temperature phase diagram according to claim 7, characterized in that, The final output vector of the first hidden layer is represented as: In the formulae, is the final output vector of the first hidden layer, m is the mask.

9. The artificial intelligence generating method of a ferroelectric material composition-temperature phase diagram according to claim 1, wherein, The step of outputting the crystal structure type comprises: Based on the chemical formula numerical vector list and the normalized temperature list, the total number of executions of the pre-trained crystal phase transition artificial intelligence model is calculated, and the calculation expression of the total number of executions is: N=N1×N2 In the formula, N is the total number of executions, N1 is the number of chemical formula numerical vectors in the chemical formula numerical vector list, that is, the number of chemical formulas with different doping proportions, and N2 is the number of temperature points in the normalized temperature list, that is, the total number of calculations of each chemical formula numerical vector at different temperature points; The chemical formula numerical vectors in the chemical formula numerical vector list and the normalized temperatures in the normalized temperature list are combined as input data for one execution, and the pre-trained crystal phase transition artificial intelligence model is used for processing, to output the crystal structure type of this execution, and the step is repeated for N-1 times to obtain the crystal structure types of all chemical formula numerical vectors at various temperature points.

10. An artificial intelligence generated system of ferroelectric material composition-temperature phase diagrams, characterized in that, It comprises: A pre-trained artificial intelligence module is used to obtain an electrical material crystal phase transition data set, train a multi-layer artificial neural network model, perform hyperparameter optimization and cross-validation during the training process, and obtain a pre-trained crystal phase transition artificial intelligence model; An automatic phase diagram generation module is used to read a doped variable ferroelectric material chemical formula, a doping proportion, and a temperature variation range and step, perform standardization processing, cyclically call the pre-trained crystal phase transition artificial intelligence model for processing, output a crystal structure type, and draw a component-temperature phase diagram of the doped variable ferroelectric material chemical formula under different doping proportions and different temperature conditions according to the crystal structure type, wherein the doped variable ferroelectric material chemical formula comprises chemical elements and corresponding doping proportions, and a doping variable is contained in the doping proportion.

Images