Glass transition temperature prediction method based on zebra optimization algorithm

Abstract

The present invention relates to the field of electrical digital data processing. Disclosed is a glass transition temperature prediction method based on a zebra optimization algorithm. The method comprises: constructing a transition temperature database of glass materials having different components; constructing a feature descriptor; providing a training set sample, and initializing parameters of a zebra search algorithm; setting a zebra optimization algorithm to optimize parameters, using the zebra optimization algorithm to optimize the parameters of a neural network algorithm so as to obtain the optimal parameters of the neural network algorithm, and establishing a glass transition temperature prediction model; and inputting test sample data, and predicting a glass transition temperature. The present invention uses the zebra optimization algorithm for optimizing the parameters of the neural network algorithm, has a simple structure, and improves the convergence speed and precision, the optimal parameters of the neural network algorithm obtained by optimization can obviously improve the performance of the neural network algorithm, and the method has practical significance for improving the accuracy of predicting the glass transition temperature.

Description

Glass transition temperature prediction method based on zebra optimization algorithm Technical Field

[0001] The present invention belongs to the field of electrical digital data processing, and in particular relates to a glass transition temperature prediction method based on a zebra optimization algorithm. Background Art

[0002] Glass is a non-equilibrium, amorphous material that can spontaneously relax to a supercooled liquid state. Unlike crystals, glass does not need to meet strict stoichiometric rules and can be considered as a continuous solution of chemical elements. Therefore, a large number of elements may become components of glass materials. 80 chemical elements with a 1 mol% change can produce 10 52 However, the number of reported inorganic glasses is only about 106, which means there is still a huge space to explore new glass-forming compositions with special properties.

[0003] Understanding the glass transition temperature (Tg) is fundamental to the development of new glass materials. The Tg can be defined as the temperature at which a glass material transitions from a hard and brittle state to a viscous and soft state. Its importance relates, for example, to the elimination of residual stresses, glass stability against crystallization, and mechanical stability. Glass compositions with extremely low Tgs are continually sought to reduce manufacturing costs, while glasses with extremely high Tgs are being developed as refractory materials. Currently, the commonly used methods for determining glass transition temperatures include the Vogel-Fulcher-Tammann method and the Avramov-Milchev method. Both predict the Tg of a glass based on its viscosity relative to temperature. However, both methods have limitations. The Vogel-Fulcher-Tammann method fails to predict the Tg when the glass is in a low-temperature, dense state, while the Avramov-Milchev method fails to predict the Tg when the glass is in an ultra-cold liquid state. Currently, no single prediction method can effectively address the problem of inaccurate glass transition temperature predictions.

[0004] Summary of the Invention

[0005] The purpose of the present invention is to provide a glass transition temperature prediction method based on the zebra optimization algorithm to solve the problem that the glass transition temperature cannot be accurately and quickly predicted in the prior art.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] A glass transition temperature prediction method based on a zebra optimization algorithm, the method comprising:

[0008] Step 1: Build a glass transition temperature database, mapping glass components and their corresponding transition temperatures one by one;

[0009] Step 2: Descriptors with element molar content and preparation process parameters as input parameters;

[0010] Step 3: Using the descriptor as the input of the model and the glass transition temperature database as the output of the model, a training set and a test set are constructed to establish a neural network model;

[0011] Step 4: Introduce the zebra optimization algorithm to optimize the parameters of the selected neural network model;

[0012] Step 5: Establish a neural network model with optimal performance based on the optimized parameters;

[0013] Step 6: For the glass component to be predicted, use the optimal neural network model to predict the glass transition temperature of the glass component.

[0014] Furthermore, step 2 includes the following steps:

[0015] Step 2-1: The molar content of each component of the glass is used as a set of descriptors;

[0016] Step 2-2: Construct a descriptor based on glass preparation process parameters.

[0017] Furthermore, the glass preparation process parameters include heating rate, melting temperature and holding time.

[0018] Beneficial effects of the present invention:

[0019] The present invention uses the zebra optimization algorithm to optimize the neural network algorithm parameters. It has a simple structure, improves the convergence speed and accuracy, and the optimal neural network algorithm parameters obtained by optimization can significantly improve the performance of the neural network algorithm, which has practical significance for improving the accuracy of predicting the glass transition temperature. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The present invention will be further described below with reference to the accompanying drawings.

[0021] FIG1 is a flow chart of a glass transition temperature prediction method based on a zebra optimization algorithm according to the present invention. DETAILED DESCRIPTION

[0022] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0023] Example 1

[0024] A glass transition temperature prediction method based on the zebra optimization algorithm, as shown in FIG1 , includes the following steps:

[0025] Step 1: Collect transition temperature data of oxide glasses with different components and build a glass transition temperature database. Specifically, collect 800 sets of transition temperature data of SiO2-Na2O-CaO-Al2O3 glasses and build a glass transition temperature database.

[0026] Step 2: Descriptors with element molar content and preparation process parameters as input parameters;

[0027] Step 2-1: Take the molar contents of elements of 800 groups of SiO2-Na2O-CaO-Al2O3 glasses as a set of descriptors;

[0028] Step 2-2: Construct a descriptor based on the glass preparation process parameters (heating rate, melting temperature, and holding time).

[0029] Step 3: Using the descriptors constructed in step 2 as the input of the model and the transition temperature database constructed in step 1 as the output of the model, a training set of 640 data sets and a test set of 160 data sets were constructed to establish a neural network model;

[0030] Step 4: Introduce the zebra optimization algorithm to optimize the parameters of the selected neural network model;

[0031] Step 4-1: The hyperparameters of the neural network algorithm (learning rate, batch size, and number of iterations) are the parameters that need to be optimized. The position coordinates of the zebras are the parameters that need to be optimized. The population size of the zebra optimization algorithm is defined as 50, and the maximum number of iterations is 1000.

[0032] Step 4-2: During the foraging phase (the first phase), the best member of the population is considered the pioneer zebra and guides other population members towards its position in the search space. Therefore, updating the position of the zebra during the foraging phase can be mathematically modeled using the following formula:

[0033] in, The new position of the i-th zebra during the foraging phase; The j-dimensional value of the i-th zebra; F i new,P1 New objective function value; PZ refers to the pioneer zebra, that is, the best zebra; PZ j Refers to the j-dimensional value of the Pioneer Zebra; R is a random number between [0,1]; I = round(1+rand), rand is a random number between [0,1].

[0034] Step 4-3: Calculate the model accuracy of the parameters represented by each zebra position based on the position of each zebra in step 4-2, and select the best pioneer zebra position;

[0035] Step 4-4: Defensive behavior against predators (second stage): Simulate the zebra's defensive strategy against predator attacks to update the position of zebra population members in the search space. There are two cases: (1) A lion attacks a zebra, so the zebra chooses an escape strategy (S1); (2) Another predator attacks the zebra, so the zebra chooses an attack strategy (S2). The formula is as follows:

[0036] in, The new state of the i-th zebra in the second stage; The j-dimensional value of the i-th zebra in the second stage; F i new，P2 The objective function value of the second stage; t is the number of iterations; T is the maximum number of iterations; R is a constant (0.01); P s Is a random number between [0,1]; AZ is the state of the attacked zebra; AZ j is the j-th dimension value of AZ.

[0037] Step 4-5: Repeat steps 4-2 to 4-4 until the population size N and the maximum number of iterations T are met, and then the optimization process ends;

[0038] Step 4-6: Output the position and accuracy of the zebra. The position of the zebra is the parameter value of the neural network.

[0039] Step 5: Establish a neural network model with optimal performance based on the optimized parameters;

[0040] Step 6: For the glass component to be predicted, use the optimal neural network model to predict the glass transition temperature of the glass component.

[0041] Example 2

[0042] A glass transition temperature prediction method based on the zebra optimization algorithm is different from Example 1 in that:

[0043] Specifically, 1,600 sets of SiO2-Na2O-CaO-Al2O3 glass transition temperature data were collected to build a glass transition temperature database;

[0044] The molar contents of elements in 1600 groups of SiO2-Na2O-CaO-Al2O3 glasses are used as a set of descriptors;

[0045] Construct 1280 sets of data as training set and 320 sets of data as test set to build neural network model;

[0046] The other steps are the same as those in Example 1.

[0047] The performance of the embodiment model is shown in the following table:

[0048] Table 1: Performance of the models of Examples 1 and 2

[0049] The above contents are merely examples and explanations of the present invention. Those skilled in the art may make various modifications or additions to the described specific embodiments or replace them in similar ways. As long as they do not deviate from the invention or exceed the scope defined by the claims, they should all fall within the scope of protection of the present invention.

Claims

1. A method for predicting the glass transition temperature based on the zebra optimization algorithm, characterized in that, The method includes: Step 1: Construct a glass transition temperature database, in which the glass components and their corresponding transition temperatures are mapped one by one; Step 2: Use the elemental molar content and preparation process parameters as descriptors of input parameters; Step 3: Use the descriptors as the input of the model and the glass transition temperature database as the output of the model to construct a training set and a test set, and establish a neural network model; Step 4: Introduce the zebra optimization algorithm to optimize the parameters of the selected neural network model; Step 5: Based on the optimized parameters, establish a neural network model with the best performance; Step 6: For the glass component to be predicted, use the optimal neural network model to predict the glass transition temperature of the glass component.

2. The glass transition temperature prediction method based on the zebra optimization algorithm according to claim 1, characterized in that Step 2 includes the following steps: Step 2-1: Use the molar content of each component constituting the glass as a set of descriptors; Step 2-2: Construct descriptors with the glass preparation process parameters.

3. The glass transition temperature prediction method based on the zebra optimization algorithm according to claim 2, wherein The glass preparation process parameters include the heating rate, melting temperature, and holding time.

4. The glass transition temperature prediction method based on the zebra optimization algorithm according to claim 1, characterized in that Step 4 includes the following steps: Step 4-1: Select the parameters to be optimized by the neural network algorithm. The position coordinates of the zebras are the parameters to be optimized. Define the population size N and the maximum number of iterations T of the zebra optimization algorithm; Step 4-2: Update the position of the zebra during the foraging phase and perform mathematical modeling using the following formula: Among them, The new position of the i-th zebra during the foraging stage; The j - dimensional value of the i - th zebra; F i new,P1 The new objective function value; PZ refers to the pioneer zebra; PZ j Refers to the j - dimensional value of the pioneer zebra; R is a random number between [0, 1]; I = round(1+rand), where rand is a random number between [0, 1]; Step 4-3: Calculate the model accuracy of the parameters represented by the position of each zebra according to the position of each zebra in Step 4-2, and select the best pioneer zebra position; Step 4-4: Simulate the defensive strategy of zebras against predator attacks to update the positions of zebra population members in the search space. The formula is as follows: where x i new,P2 is the new state of the i-th zebra; is the j-dimensional value of the i-th zebra; F i new，P2 is the objective function value; t is the number of iterations; T is the maximum number of iterations; R is a constant (0.01); P s is a random number between [0, 1]; AZ is the attacked zebra The state of; AZ j Is the j-th dimensional value of AZ; Step 4-5: Repeat Steps 4-2 to 4-4. When the population size N and the maximum number of iterations T are satisfied, end the optimization process; Step 4-6: Output the position and accuracy of the zebras. The position of the zebras is the parameter value of the neural network.

Images