Glass fiber component design method and generated glass fiber

Through the multi-objective optimization method of machine learning model and AGE algorithm, glass fibers with high modulus and high specific modulus are designed, solving the problems of long research and development cycle and high cost of glass fibers in the existing technology, and achieving high efficiency and low cost fiberglass composition design.

WO2024230031A1PCT designated stage expired Publication Date: 2024-11-14NANJING FIBERGLASS RES & DESIGN INST CO LTD +1

Patent Information

Application Number
PCT/CN2023/117978
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-05-10
Filing Date
2023-09-11
Publication Date
2024-11-14

AI Technical Summary

Technical Problem

The existing glass fiber component design methods are difficult to meet the needs of high modulus and high specific modulus at the same time, and the traditional methods are affected by component concentration and oxide synergistic effects, resulting in long R&D cycles and high cost.

Method used

Using a multi-objective optimization method based on machine learning model and AGE algorithm, the objective function of modulus and specific modulus is constructed. Through the multi-objective optimization algorithm, glass fiber components that meet the limitations are obtained, and the calculation complexity and time are reduced.

Benefits of technology

A glass fiber design with high modulus (≥95GPa) and high specific modulus (>3.68) was achieved, reducing the use of rare metal oxides, reducing raw material costs, and significantly shortening the calculation time.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN2023117978_14112024_PF_FP_ABST
    Figure CN2023117978_14112024_PF_FP_ABST
Patent Text Reader

Abstract

Disclosed in the present invention are a glass fiber component design method and a generated glass fiber. The glass fiber component design method comprises: acquiring data such as modulus, density and forming temperature of a glass fiber material, and building a machine learning model for predicting the performance of the glass fiber material; constructing an objective function according to requirements for modulus and specific modulus, and constructing constraints according to the actual conditions of the modulus, density and forming temperature of the glass fiber; and solving a multi-objective model on the basis of AGE algorithm, so as to obtain a plurality of glass fiber component solutions that meet the constraints. The tensile modulus of the glass fiber made of the glass fiber composition designed in the present invention reaches more than 95 Gpa, the specific modulus thereof is greater than 3.68, and the components of the glass fiber do not contain rare metal oxides Y2O3 and La2O3. Thus, the glass fiber has low raw material cost, and can be used for production of large wind power blades.
Need to check novelty before this filing date? Find Prior Art

Description

A glass fiber composition design method and generated glass fiber Technical Field

[0001] The present invention belongs to the technical field of glass fibers, and in particular relates to a method for designing a high modulus and high specific modulus glass fiber component and the resulting glass fiber. Background Art

[0002] Wind energy, as an important component of renewable energy, continues to grow rapidly under the background of dual carbon. The market has increasingly higher requirements for the technological development of wind turbine blades. Wind turbine blades are developing in the direction of large-scale, high modulus, low cost and lightweight. As a result, lightweight and high modulus glass fiber has become the first choice for ultra-large blade reinforcement materials.

[0003] Modulus reflects the rigidity of a material. The higher the modulus of glass fiber used in wind turbine blades, the greater its resistance to deformation. In practical applications, in addition to high modulus, lightweight glass fiber is also essential. Specific modulus is the ratio of a material's modulus to its density. A higher specific modulus indicates a higher modulus and lower density, which reduces the manufacturing difficulty of large-scale wind turbine blades. Therefore, glass fiber with both high modulus and high specific modulus has long been a research focus in the glass fiber field.

[0004] Chinese patent CN115432932A discloses a glass fiber composition and glass fiber with ultra-high specific modulus, wherein the tensile elastic modulus of the glass fiber reaches above 93.5GPa and the specific modulus is greater than 3.65. However, by comparing the current high specific modulus glass fiber products, there are very few products that can simultaneously meet the requirements of a modulus greater than 95GPa and a specific modulus greater than 3.68, and they contain relatively high amounts of expensive oxides such as Li2O (content ≥0.3%) and La2O3. The industry is in urgent need of an efficient method and system for developing a glass fiber formulation with high modulus and high specific modulus.

[0005] However, traditional glass fiber R&D is based on a trial-and-error approach. This is influenced by component concentrations and the synergistic effects of multiple oxides. The regulation of a single oxide is not a simple, monotonic relationship, and the mechanisms by which different oxides influence glass properties remain unclear. This leads to lengthy R&D cycles and high costs for high-performance glass fiber. The glass fiber industry is focusing on how to cost-effectively and efficiently design high-performance glass fiber that meets multiple performance requirements.

[0006] In response to the common fundamental issues in the development of high-performance glass fiber materials, the ideas and methods of materials genetic engineering provide new solutions to these technical challenges. With the development of big data and artificial intelligence in the materials field, materials genetics technology has made it possible to calculate and predict glass element properties. By collaborating high-throughput computing, experiments, and machine learning methods, and integrating materials science, mathematics, informatics, physics, and other disciplines, it can establish statistical mapping relationships and quantitative models between glass fiber chemical elements and properties based on a data-driven approach, without requiring a clear understanding of the complex physical mechanisms of glass materials. This allows for the efficient design of new glass fiber materials that meet specific functions, providing a new research path for the design and development of new glass fiber materials. Consequently, materials genetics technology has been called the "fourth paradigm" of scientific research.

[0007] There are two existing glass fiber composition design methods:

[0008] (1) Single-objective optimization method

[0009] The single-objective optimization model can only consider the modulus as the objective function, and cannot take into account the specific modulus of the glass fiber. It needs to repeatedly set multiple values, which has high computational cost and low efficiency, and cannot directly find the most reasonable solution.

[0010] (2) Optimization method based on NSGA-III algorithm

[0011] The NSGA-III algorithm can simultaneously consider the modulus and the specific modulus as objective functions. Although a feasible solution can be obtained, the algorithm itself needs to make assumptions about the shape of the Pareto front. If the problem itself is different from the assumption, it will lead to unsatisfactory results. At the same time, the time complexity of this algorithm is high, which makes the calculation time longer.

[0012] Summary of the Invention

[0013] The purpose of the present invention is to provide a glass fiber composition design method to address the problems existing in the above-mentioned prior art.

[0014] The technical solution for achieving the purpose of the present invention is as follows: On the one hand, a method for designing glass fiber composition is provided, the method comprising the following steps:

[0015] Collect data on various performance parameters of glass fiber materials and build a machine learning model to predict their performance;

[0016] Based on the machine learning model, the objective function is constructed according to the requirements of modulus and specific modulus, and the constraints are constructed according to the modulus, density and molding temperature of the glass fiber to build a multi-objective optimization model;

[0017] The multi-objective optimization model is solved using a multi-objective optimization algorithm based on the AGE algorithm to obtain multiple glass fiber components that meet the constraint conditions.

[0018] Furthermore, the method further comprises:

[0019] The multi-objective optimization model is solved by repeatedly using the multi-objective optimization algorithm based on the AGE algorithm to obtain the optimal content range of each component of the glass fiber.

[0020] Furthermore, the method preferably has one or more of the following:

[0021] Preferably, the performance parameter data include modulus, density and molding temperature.

[0022] Preferably, the machine learning model is a neural network machine learning model.

[0023] Preferably, the constraints include restrictions on the glass fiber modulus, density and molding temperature.

[0024] Preferably, the multi-objective optimization model is solved by using a multi-objective optimization algorithm based on the AGE algorithm to obtain multiple glass fiber components that meet the constraints, that is, the glass fiber component solution x that meets the constraints * The specific process includes:

[0025] The first step is to set the parameters of the AGE algorithm, including the population size N and the update number A;

[0026] The second step is to randomly generate a glass fiber composition solution set P with a population size of N;

[0027] The third step is to generate the next generation of glass fiber composition solution set Q through crossover and mutation;

[0028] Step 4: Combining the above two glass fiber composition solution sets, an optimized glass fiber composition solution set F is obtained;

[0029] Step 5: Generate the next generation glass fiber composition solution set P with a population size of N under the AGE framework;

[0030] Repeat steps 3 to 5 until the preset update algebra A is reached, and the final glass fiber composition solution x that meets the constraints is obtained. * .

[0031] Preferably, the above-mentioned two glass fiber component solution sets are integrated to obtain an optimized glass fiber component solution set F, which specifically includes:

[0032] Take the union of the glass fiber composition solution set P and the glass fiber composition solution set Q;

[0033] sorting the union in descending order;

[0034] Normalization processing is performed to obtain the optimized glass fiber composition solution set F.

[0035] Preferably, under the AGE framework, the next generation glass fiber composition solution set P with a population size of N is generated by calculating the survival score ss.

[0036] Preferably, the method of repeatedly performing the multi-objective optimization algorithm based on the AGE algorithm to solve the multi-objective optimization model to obtain the optimal content range of each component of the glass fiber includes:

[0037] Repeatedly solve the multi-objective optimization model using the multi-objective optimization algorithm based on the AGE algorithm until the preset number threshold n is reached, and n sets of glass fiber composition solutions x are obtained. * ;

[0038] Solve x for n groups of glass fiber components * Mathematical statistics are performed to generate the optimal content range of each component of glass fiber.

[0039] Preferably, the mathematical statistics is achieved by statistically analyzing the mass ratios of the components.

[0040] On the other hand, a glass fiber is provided, wherein the content of each component of the glass fiber is within the optimal content range of each component of the glass fiber.

[0041] Furthermore, the glass fiber preferably comprises one or more of the following:

[0042] Preferably, the components of the glass fiber include SiO2, Al2O3, CaO, MgO, Li2O, K2O, Na2O, Fe2O3, TiO2, ZrO2 and CeO2, and the rest are impurities.

[0043] Preferably, the mass percentage of each component of the glass fiber satisfies: TiO2+ZrO2≥1, Li2O+Na2O+K2O≤1.4, MgO+CaO≥12.6, Al2O3+CaO≤30.2.

[0044] In another aspect, a system is provided, comprising:

[0045] A computer, comprising:

[0046] Computer-based processors, and

[0047] a computer-based memory storing computer-executable instructions that, when executed by a computer-based processor, cause the computer-based processor to generate an optimal content range for each component of the glass fiber;

[0048] One or more machines configured to implement a machine learning model for predicting various properties of a glass fiber material;

[0049] First computer-based software for constructing a multi-objective optimization model; and

[0050] A second computer-based software for solving a multi-objective optimization model using a multi-objective optimization algorithm based on the AGE algorithm.

[0051] Furthermore, the system also includes:

[0052] One or more machines are configured to drive the second software to run repeatedly for several times to obtain the optimal content range of each component of the glass fiber.

[0053] Furthermore, the system also includes:

[0054] A third computer-based software is used to drive the second software to repeatedly run several times until a preset number threshold n is reached, to obtain n sets of glass fiber composition solutions x * ;

[0055] Used to solve x for n groups of glass fiber components * A fourth computer-based software that performs mathematical statistics generates an optimal content range for each component of the glass fiber.

[0056] The system preferably includes one or more of the above methods.

[0057] Compared with the prior art, the present invention has the following significant advantages:

[0058] (1) The multi-objective optimization algorithm proposed based on AGE takes into account both the diversity among population members and the degree of proximity of each population member to the ideal point, which can improve the efficiency of obtaining the next generation of population.

[0059] (2) Compared with other traditional multi-objective algorithms, such as NSGA-III, the multi-objective optimization algorithm proposed based on AGE does not require any assumptions about the Pareto front, which enables the algorithm to be used in multi-objective optimization problems with more objective functions and constraints;

[0060] (3) The multi-objective optimization algorithm proposed based on AGE uses a fast algorithm to estimate the geometry of the Pareto front. Its computational time complexity is lower than that of the non-dominated sorting algorithm in the NSGA-III algorithm, which greatly reduces the time to obtain the optimal solution.

[0061] (4) The tensile modulus of the glass fiber of the glass fiber composition obtained by the present invention reaches above 95 GPa, and the specific modulus is greater than 3.68, that is, it has the characteristics of high modulus and high specific modulus. The glass fiber does not contain Y2O and La2O3 rare metal oxides, and the raw material cost is low, and it can be used in the production of large wind turbine blades.

[0062] The present invention is further described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] FIG1 is a flow chart of a method for designing a high modulus, high specific modulus glass fiber composition according to one embodiment.

[0064] FIG2 is a diagram showing the updating process of the AGE algorithm under different parameters in one embodiment.

[0065] FIG3 is a box plot of the experimental modulus values ​​of glass fibers in one embodiment.

[0066] FIG4 is a box plot of experimental specific modulus values ​​of glass fibers according to an embodiment.

[0067] FIG5 is a box plot showing the time taken to calculate the glass fiber composition according to one embodiment.

[0068] Implementation Method

[0069] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0070] It should be noted that if the embodiments of the present invention involve directional indications (such as up, down, left, right, front, back, etc.), the directional indications are only used to explain the relative position relationship, movement status, etc. between the various components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indications will also change accordingly.

[0071] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present invention, the descriptions of "first", "second", etc. are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features specified as "first" and "second" may explicitly or implicitly include at least one of such features. In addition, the technical solutions between the various embodiments can be combined with each other, but this must be based on the fact that ordinary technicians in this field can implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0072] In one embodiment, in conjunction with FIG1 , a method for designing a glass fiber composition is provided, the method comprising the following steps:

[0073] Collect data on various performance parameters of glass fiber materials, including modulus, density, and molding temperature, and build a neural network machine learning model to predict its performance;

[0074] Based on the machine learning model, the objective function is constructed according to the requirements of modulus and specific modulus, and the constraints are constructed according to the glass fiber modulus, density and molding temperature to build a multi-objective optimization model;

[0075] Solving the multi-objective optimization model using a multi-objective optimization algorithm based on the AGE algorithm to obtain multiple glass fiber compositions that meet the constraint conditions;

[0076] The multi-objective optimization model is solved by repeatedly using the multi-objective optimization algorithm based on the AGE algorithm to obtain the optimal content range of each component of the glass fiber.

[0077] Furthermore, in one embodiment, the multi-objective optimization model and constraints are:

[0078] Multi-objective optimization model: This model ensures that each performance parameter of the designed glass fiber reaches the set optimal target, and the optimal target corresponding to each performance parameter satisfies certain balance constraints, so that the overall performance of the glass fiber is optimized;

[0079] The restriction conditions include: the data values ​​of each performance parameter calculated by the machine learning model respectively meet their corresponding preset restriction ranges.

[0080] Furthermore, in one embodiment, the multi-objective optimization model is solved by a multi-objective optimization algorithm based on the AGE algorithm to obtain multiple glass fiber components that meet the constraints, that is, the glass fiber component solution x that meets the constraints. * The specific process includes:

[0081] The first step is to set the parameters of the AGE algorithm, including the population size N and the update number A;

[0082] The second step is to randomly generate a glass fiber composition solution set P with a population size of N;

[0083] The third step is to generate the next generation glass fiber component solution set Q by crossover and mutation; preferably, the next generation glass fiber component solution set Q is generated by using a simulated binary crossover method and a polynomial mutation method;

[0084] The fourth step is to combine the above two glass fiber composition solution sets to obtain the optimized glass fiber composition solution set F. Specifically:

[0085] Take the union of the glass fiber composition solution set P and the glass fiber composition solution set Q;

[0086] Arranging the union in descending order; preferably, arranging the union in descending order by a fast non-dominated sorting method;

[0087] Perform normalization processing to obtain the optimized glass fiber composition solution set F;

[0088] Step 5: Under the AGE framework, the next generation glass fiber composition solution set P with a population size of N is generated by calculating the survival score ss. The calculation formula of the survival score ss is as follows:

[0089] Among them, F1 is the first non-dominated plane, S∈F1,

[0090] diversity(S, F1)=min||C n (S)-C(T)|| p

[0091] proximity(S)=||C n (S)||

[0092] Where n is the number of oxides in the glass fiber composition, S,T∈F1C(·) is the objective function, and proximity represents the target vector C n (S) is the distance from the ideal point, diversity is the minimum distance of other solutions in F1, ||·|| p Represents the geometric p-norm, and the calculation formula of p is as follows:

[0093] Where M is the number of objective functions, Z i is the i-th coordinate in the target space;

[0094] Repeat steps 3 to 5 until the preset update algebra A is reached, and the final glass fiber composition solution x that meets the constraints is obtained. * The specific process includes:

[0095] Repeatedly solve the multi-objective optimization model using the multi-objective optimization algorithm based on the AGE algorithm until the preset number threshold n is reached, and n sets of glass fiber composition solutions x are obtained. * ;

[0096] Solve x for n groups of glass fiber components * Mathematical statistics are performed to generate the optimal content range of each component of glass fiber.

[0097] In one embodiment, a high modulus and high specific modulus glass fiber is provided based on the above design method. The production process of the glass fiber is as follows:

[0098] S1. Obtain data on the modulus, density, and molding temperature properties of SiO2-based multi-element oxide glass systems from a database, train the data using a neural network model, and develop a machine learning model for calculating the modulus, density, and molding temperature.

[0099] S2. In actual engineering production, in addition to the requirements for the modulus of glass fiber, its specific modulus must also be as large as possible. The objective function is determined based on this requirement.

[0100] S3. After determining the objective function, it is necessary to set constraints on the modulus, molding temperature, and density of the glass fiber to determine the search space.

[0101] S4. Set the parameters of the AGE algorithm: population size N and update number A. Setting these two parameters too large will result in a long calculation time, while setting them too small will cause the algorithm to fail and thus fail to calculate the optimal solution. Therefore, the present invention sets the population size N and update number A by calculating the average variance of each oxide in each formula within the population. The specific calculation formula is as follows:

[0102] Among them, Var(x i ) is the variance of the i-th oxide in the glass fiber formula in the population, and n is the number of oxides. As can be seen from Figure 2, when the population size is 150, the variance decreases slowly, and even when the update algebra reaches 140, it does not approach 0, indicating that the population size is too small, more update algebras are required, and it takes a long time; when the population size is between 160-240, the update algebra starts to drop rapidly from about 20 generations, until the variance starts to slowly decrease at about 80 generations, and finally approaches 0 at about 120 generations. When the population size is 250, it can be found that the larger the population size, the fewer update algebras are required for the variance to quickly decrease to a stable stage, but the larger the population size, the higher the calculation time. Therefore, the population size is set between 160-240, and the update algebra is set at 80-120 for the best. The present invention sets the population size N=200 and the update algebra A=100.

[0103] S5. Use the AGE algorithm to solve this multi-objective optimization model. During the solution process, as the number of update generations increases, the variance of all solutions in the population gradually decreases, and finally all solutions will converge to a consistent solution x that satisfies the constraints. * Repeat the AGE algorithm many times to obtain a series of glass fiber composition solutions x *Calculated by mass, the composition ranges are as follows: SiO2: 55%-60%, Al2O3: 18%-25%, CaO: 0.1%-7.8%, MgO: 11%-18%, Li2O: 0%-0.3%, K2O: 0.1%-0.5%, Na2O: 0.1%-1%, Fe2O3: 0.2%-1%, TiO2: 0.05%-4%, ZrO2: 0.05%-4%, CeO2: 0%-0.3%, with the remainder being impurities. The mass percentages of TiO2 and ZrO2 satisfy the following: TiO2+ZrO2≥1, and the mass ratios of Li2O, Na2O, and K2O satisfy the following: Li2O+Na2O+K2O≤1.4. The mass percentage of MgO and CaO satisfies: MgO+CaO≥12.6; the mass percentage of Al2O3 and CaO satisfies: Al2O3+CaO≤30.2, and the components do not contain rare metal oxides La2O3 and Y2O3.

[0104] The tensile modulus of the glass fiber of the present invention reaches above 95 GPa, and the specific modulus is greater than 3.68. The glass fiber does not contain Y2O3 and La2O3 rare metal oxides, and the raw material cost is low. It can be used in the production of large wind turbine blades.

[0105] The glass fiber with high modulus and high specific modulus can be prepared according to the following method:

[0106] (1) Calculate the mass of various raw materials required based on the formula composition of glass fiber with high modulus and high specific modulus;

[0107] (2) Weighing each raw material and transporting it to a mixing tank for mixing. After mixing evenly, transport it to the kiln head silo to obtain a batch material;

[0108] (3) The mixed material in the kiln head silo is fed into the tank kiln. In the tank kiln, the mixed material is clarified and homogenized under high temperature conditions of 1450℃-1600℃, and high-quality glass liquid enters the drawing operation channel;

[0109] (4) The glass liquid in the drawing operation channel is cooled to 1270-1350℃, flows out through the platinum bushing, and is quickly drawn into glass fiber by the drawing machine.

[0110] As a specific example, the glass fiber is subjected to density, modulus, specific modulus and molding temperature performance tests.

[0111] The experimental performance data of the glass fibers with high modulus and high specific modulus designed using the AGE multi-objective optimization algorithm in Examples 1-10 are shown in Table 1.

[0112] The glass fiber performance data of Comparative Examples 1-10 based on the single-objective optimization algorithm are shown in Table 2.

[0113] The glass fiber performance data of Comparative Examples 11-20 based on the NSGA-III algorithm are shown in Table 3.

[0114] The glass fiber performance data of Comparative Examples 21-30, which are not within the composition range of the present invention, are shown in Table 4.

[0115] Table 1 Data table of Examples 1-10

[0116] Table 2 Data table of comparative examples 1-10 based on single-objective optimization algorithm

[0117] Table 3 Data table of comparative examples 11-20 based on NSGA-III algorithm

[0118] Table 4 Data table of comparative examples 21-30

[0119] The method proposed in the present invention was compared with the other two methods (single-objective optimization (traditional method 1) and the optimization method based on the NSGA-III algorithm (traditional method 2)). The calculated formula was experimentally made into glass fiber and its performance was tested. The results are shown in Tables 2 and 3. It can be seen from the table that the glass fiber formula screened by the present invention is superior to the other two traditional methods in terms of the median of the maximum value of the modulus and specific modulus, and the calculation time is also the lowest among the three methods.

[0120] Figures 3 to 5 show the box plots of the modulus, specific modulus and time consumption of the formulas obtained using different algorithms. It can be seen from the figures that the median of the modulus of the present invention is 96.0 GPa, the median of the modulus of the single-objective optimization model is 92.9 GPa, and the median of the modulus of the optimization method based on the NSGA-III algorithm is 92.9 GPa; the median of the specific modulus of the present invention is 3.726, the median of the specific modulus of the single-objective optimization model is 3.590, and the median of the specific modulus of the optimization method based on the NSGA-III algorithm is 3.585; the median time consumption of the present invention is 18.4 s, the median time consumption of the single-objective optimization model is 20.8 s, and the median time consumption of the optimization method based on the NSGA-III algorithm is 26 s; the above results also indicate that the single-objective optimization and the NSGA-III algorithm cannot calculate a formula greater than 95 GPa under the same target performance and constraints. The glass fiber designed by the present invention has a specific modulus of ≥3.68, a modulus of ≥95GPa, and a molding temperature between 1271-1340°C.

[0121] Finally, data outside the scope of the present invention are compared, as shown in Table 4. The modulus and specific modulus in Comparative Examples 21-30 are both low, failing to meet the requirements of the present invention; compared to the comparative examples, the lack of the addition of La2O3 and Y2O3 rare metal oxides in the examples greatly reduces production costs.

[0122] In comparative example 21, TiO2+ZrO2=0.4 is less than 1. Although the density can be reduced compared with the embodiment, the modulus is only 92.2 GPa and the specific modulus is only 3.63. The content of TiO2 and ZrO2 is too low, which will result in a lower modulus of the glass fiber.

[0123] Li2O, Na2O and K2O are common alkali metal oxides in glass fiber. + and Na + Mainly plays the role of disconnecting the network, Li + Taking Comparative Example 23 as an example, the sum of Li2O+Na2O+K2O is greater than 1.4, and the modulus and specific modulus are difficult to meet the requirements.

[0124] Although the addition of Al2O3 can increase the modulus and molding temperature of the glass fiber, in Comparative Example 30, Al2O3+CaO=31 is greater than 30.2, the modulus is 93.8 GPa, and the specific modulus is 3.673; at the same time, in Comparative Example 28, MgO+CaO≤12.6, the modulus is only 87 GPa, and the specific modulus is 3.473. Both the modulus and the specific modulus do not meet the requirements.

[0125] The embodiments or aspects thereof may be implemented in the form of hardware, firmware, or software. If implemented in software, the software may be stored on any non-transitory computer-readable medium that is configured to enable a processor to load the software or a set of instructions thereof. The processor then executes the instructions and is configured to operate or cause the apparatus to operate in the manner described herein.

[0126] In addition, firmware, software, routines, or instructions may be described herein as performing certain actions and / or functions of a data processor. However, it should be appreciated that such descriptions are included herein only for convenience, and that such actions are actually caused by a computing device, processor, controller, or other device executing the firmware, software, routines, instructions, etc.

[0127] It should be understood that the flow charts, block diagrams, and network diagrams may include more or fewer elements, be arranged differently, or be represented differently. However, it should be understood that certain implementations may dictate the number of block diagrams and network diagrams, as well as the number of block diagrams and network diagrams that illustrate the execution of an embodiment implemented in a particular manner.

[0128] Accordingly, other embodiments may also be implemented in various computer architectures, physical, virtual, cloud computers, and / or some combination thereof, and thus, the data processors described herein are for example purposes only and not as limitations of the embodiments.

[0129] While the invention has been particularly shown and described with reference to particular embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the scope of the invention as encompassed by the appended claims.

Claims

1. A method for designing glass fiber composition, characterized in that: The method comprises the following steps: Collect data on various performance parameters of glass fiber materials and build a machine learning model to predict their performance; Based on the machine learning model, the objective function is constructed according to the requirements of modulus and specific modulus, and the constraints are constructed according to the glass fiber modulus, density and molding temperature to build a multi-objective optimization model; The multi-objective optimization model is solved by using a multi-objective optimization algorithm based on the AGE algorithm to obtain a plurality of glass fiber components that meet the constraint conditions; Repeating the multi-objective optimization algorithm based on the AGE algorithm to solve the multi-objective optimization model, to obtain the optimal content range of each component of the glass fiber; The multi-objective optimization model is solved by using a multi-objective optimization algorithm based on the AGE algorithm to obtain multiple glass fiber components that meet the constraint conditions, that is, the glass fiber component solution x that meets the constraint conditions * The specific process includes: The first step is to set the parameters based on the AGE algorithm, including the population size N and the update number A; The second step is to randomly generate a glass fiber component solution set P with a population size of N; The third step is to generate the next generation glass fiber component solution set Q through crossover and mutation; Step 4: Combining the above two glass fiber component solution sets, an optimized glass fiber component solution set F is obtained; The fifth step is to generate the next generation glass fiber component solution set P with a population size of N by calculating the survival score ss under the AGE framework; Repeat steps 3 to 5 until the preset update algebra A is reached, and the glass fiber composition solution x that finally meets the constraints is obtained. * ; Among them, the population size N and the update algebra A are set by calculating the average variance of each oxide in each formula within the population. The specific calculation formula is as follows: Among them, Var(x i ) is the variance of the ith oxide in the glass fiber formula in the population, and n is the number of oxides; Among them, the calculation formula of survival score ss is as follows: Among them, F1 is the first non-dominated plane, S∈F1, diversity(S,F1)=min||C n (S)-C(T)|| p proximity(S)=||C n (S)|| Where n is the number of oxides in the glass fiber component, S, T∈F1, C(·) is the objective function, and proximity represents the target vector C n (S) is the distance from the ideal point, diversity is the minimum distance of other solutions in F1, ||·|| p Represents the geometric p-norm, and the calculation formula of p is as follows: Where M is the number of objective functions, Z i is the i-th coordinate in the target space.

2. The glass fiber composition design method according to claim 1, characterized in that: The performance parameter data include modulus, density and molding temperature.

3. The glass fiber composition design method according to claim 1, characterized in that: The machine learning model is a neural network machine learning model.

4. The glass fiber composition design method according to claim 1 or 2, characterized in that: The constraints include limitations on glass fiber modulus, density, and molding temperature.

5. The glass fiber composition design method according to claim 1, characterized in that: The above two glass fiber component solution sets are combined to obtain an optimized glass fiber component solution set F, which specifically includes: Taking the union of the glass fiber component solution set P and the glass fiber component solution set Q; Arrange the union in descending order; Normalization processing is performed to obtain the optimized glass fiber component solution set F.

6. The glass fiber composition design method according to claim 1, characterized in that: The method of repeatedly performing the multi-objective optimization algorithm based on the AGE algorithm to solve the multi-objective optimization model to obtain the optimal content range of each component of the glass fiber includes: Repeat the multi-objective optimization algorithm based on the AGE algorithm to solve the multi-objective optimization model until a preset number of times threshold n is reached to obtain n groups of glass fiber composition solutions x * ; Solve x for n groups of glass fiber components * Mathematical statistics are performed to generate the optimal content range of each component of glass fiber.

7. The method for designing glass fiber composition according to claim 6, characterized in that: The mathematical statistics are achieved by statistically analyzing the mass ratios of the components.

8. Glass fiber according to the method according to any one of claims 1 to 7, characterized in that: The content of each component of the glass fiber is within the optimal content range of the components of the glass fiber; the components of the glass fiber include SiO2, Al2O3, CaO, MgO, Li2O, K2O, Na2O, Fe2O3, TiO2, ZrO2 and CeO2, and the rest are impurities.

9. The glass fiber according to claim 8, characterized in that The mass percentage of each component of the glass fiber satisfies: TiO2+ZrO2≥1, Li2O+Na2O+K2O≤1.4, MgO+CaO≥12.6, Al2O3+CaO≤30.

2.

10. A system for implementing the method according to any one of claims 1 to 7, characterized in that: include: A computer, the computer comprising: Computer-based processors, and A computer-based memory storing computer-executable instructions, which, when executed by a computer-based processor, cause the computer-based processor to generate an optimal content range of each component of the glass fiber; One or more machines configured to implement a machine learning model that predicts various properties of a fiberglass material; a first computer-based software for constructing a multi-objective optimization model; and A second computer-based software for solving a multi-objective optimization model using a multi-objective optimization algorithm based on the AGE algorithm.

11. The system according to claim 10, characterized in that The system also includes: One or more machines are configured to drive the second software to run repeatedly for several times to obtain the optimal content range of each component of the glass fiber.

12. The system according to claim 11, characterized in that The system also includes: A third computer-based software for driving the second software to repeatedly run for a number of times until a preset number threshold n is reached, to obtain n sets of glass fiber composition solutions x * ; Used to solve x for n groups of glass fiber components * A fourth computer-based software that performs mathematical statistics generates an optimal content range for each component of the glass fiber.

13. The system according to claim 12, characterized in that The mathematical statistics are achieved by statistically analyzing the mass ratios of the components.

14. The system according to claim 10, characterized in that The performance parameter data include modulus, density and molding temperature.

15. The system according to claim 10, characterized in that The machine learning model is a neural network machine learning model.

16. The system according to claim 10, characterized in that The constraints include limitations on glass fiber modulus, density, and molding temperature.

17. The system according to claim 10, characterized in that The multi-objective optimization model is solved by using a multi-objective optimization algorithm based on the AGE algorithm to obtain multiple glass fiber components that meet the constraint conditions, that is, the glass fiber component solution x that meets the constraint conditions * The specific process includes: Set the parameters based on the AGE algorithm, including population size N and update number A; Randomly generate a glass fiber component solution set P with a population size of N; Generate the next generation glass fiber component solution set Q through crossover and mutation; Combining the above two glass fiber component solution sets, an optimized glass fiber component solution set F is obtained; Under the AGE framework, the next generation of glass fiber composition solution set P with a population size of N is generated; Repeat steps 3 to 5 until the preset update algebra A is reached, and the glass fiber composition solution x that finally meets the constraints is obtained. * .

18. The system according to claim 17, characterized in that The above two glass fiber component solution sets are combined to obtain an optimized glass fiber component solution set F, which specifically includes: Taking the union of the glass fiber component solution set P and the glass fiber component solution set Q; Arrange the union in descending order; Normalization processing is performed to obtain the optimized glass fiber component solution set F.

19. The system according to claim 18, characterized in that In the AGE framework, the glass fiber component solution set P of the next generation population size N is generated by calculating the survival score ss.

20. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Citation Information

Patent Citations

  • Multi-objective optimization method for injection molding process parameters of glass fiber reinforced plastics

    CN112115579A

  • Accurate design method and system for ore fiber components

    CN114707368A

  • Method for predicting fatigue performance of chopped fiber composite material based on machine learning and progressive damage

    CN115547431A

  • Glass fiber component design method and generated glass fiber

    CN116227368A

Cited By

  • Multi-system glass material modulus analysis method

    CN119274723A

  • Preparation method of nylon glass fiber composite flexible hinge

    CN120588405A

  • Glass fiber molecular structure-modulus performance relation prediction method based on graph neural network

    CN121054112A