Polymer formula design method and device based on NSGA3 algorithm

Abstract

The invention discloses a polymer formula design method and device based on an NSGA3 algorithm, and the method comprises the steps: setting a database, and carrying out the arrangement to obtain an initial data set; dividing the plurality of raw material proportioning fields into a plurality of main raw material proportioning fields and a plurality of secondary raw material proportioning fields according to a set threshold value; converting the plurality of secondary raw material matching fields into category type variable fields to obtain a preprocessed data set; constructing and training each performance prediction model according to the regression prediction model; evaluating individuals of the population through a fitness function to obtain individual fitness; selecting individuals with high individual fitness from the parent population and the filial population as a next-generation population by selecting an operator; processing the next generation population through crossover and mutation operators to obtain a new offspring population; and carrying out iterative selection operation processing on the new offspring population, and obtaining the polymer formula from the last generation population. According to the invention, the raw material ratio optimization efficiency can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of polymer material formula design, and in particular to a polymer formula design method and device based on NSGA3 algorithm. Background Art

[0002] The field of polymer material technology is an extremely rapidly developing field. There are many types of polymer materials, such as rubber, plastic, fiber, coating, adhesive and polymer-based composite materials. The application of polymer materials is also very extensive, and they are widely used in construction, transportation, agriculture, liquid crystal, medical treatment, electrical and electronic industries, etc.

[0003] The preparation process of rubber polymer materials often involves many process conditions, such as raw material ratio, temperature, time and other conditions that will affect the performance of the final product. Therefore, the process optimization process involves a large number of experiments, and the experimental personnel need to cooperate in carrying out the corresponding single factor test and orthogonal test, so the process optimization process will take up a lot of time and materials. Especially for the optimization of raw material ratio, it will involve various variables such as the type of raw materials and the ratio of raw materials. For rubber products that use more types of raw materials, it will greatly increase the experimental cost. The raw material ratio is the most critical factor affecting product performance, so the rapid optimization of the raw material ratio is of great value.

[0004] At present, polymer material products are synthesized from a variety of raw materials in a certain ratio, among which only a small number of raw materials are used as the main raw materials, accounting for a large proportion. The various properties of the synthesized material products are completely different due to the different composition and ratio of the raw materials. Traditionally, the composition and ratio of the raw materials required to synthesize a product based on the various properties of the product are tried based on the experience of engineers and technicians in this field, but this method is time-consuming and labor-intensive, and it is impossible to form the accumulation of experience.

[0005] Therefore, how to invent a polymer material formula design method to improve the efficiency of optimizing the raw material ratio has become an urgent problem to be solved. Summary of the invention

[0006] To this end, the present invention provides a polymer formula design method and device based on the NSGA3 algorithm, which utilizes a machine learning algorithm to respectively construct a mapping relationship between various properties of polymer products and their formulas, and further combines the multi-objective NSGA3 algorithm to realize the reverse design of the product formula.

[0007] In order to achieve the above object, the present invention provides the following technical solution: a polymer formulation design method based on NSGA3 algorithm, comprising:

[0008] By setting up a database, an initial data set is obtained; the initial data set includes several raw material ratio fields and several polymer performance fields;

[0009] Dividing the several raw material ratio fields in the initial data set into several main raw material ratio fields and several secondary raw material ratio fields according to a set threshold; converting the several secondary raw material ratio fields into categorical variable fields to obtain a preprocessed data set;

[0010] Construct each performance prediction model according to the regression prediction model; train each performance prediction model through the preprocessed data set to obtain each trained performance prediction model;

[0011] Designing a fitness function, evaluating individuals of the population by using the fitness function, and obtaining individual fitness;

[0012] By selecting an operation operator, the individuals with high individual fitness are selected from the parent and child populations as the next generation population;

[0013] The next generation population is processed by crossover and mutation operators to obtain a new offspring population; the new offspring population is processed by iterative selection operations until a set generation is reached, and a polymer formula is obtained from the last generation population.

[0014] As a preferred scheme of a polymer formula design method based on the NSGA3 algorithm, in the process of evaluating the individuals of the population through the fitness function to obtain the individual fitness, the individuals are evaluated and predicted through each performance prediction model to obtain each performance corresponding to the individual; each performance of the individual is compared with the target performance, and the individual fitness is obtained through the fitness function.

[0015] As a preferred scheme of a polymer formula design method based on the NSGA3 algorithm, in the process of selecting the next generation population from the parent and offspring populations, uniformly distributed reference points are generated in the target space; the population after the parent and offspring are merged is used, according to the individual fitness, to obtain non-dominated solution sets of different levels through a fast non-dominated sorting strategy; the non-dominated solution set is associated with the reference point, and the same number of individuals as the parent or offspring population are selected to form the next generation population.

[0016] As a preferred scheme of the polymer formula design method based on the NSGA3 algorithm, in the process of processing the next generation population through crossover and mutation operators to obtain the new offspring population, the next generation population is subjected to crossover and mutation processing at the same time, or crossover processing is performed alone, or mutation processing is performed alone.

[0017] As a preferred scheme of a polymer formula design method based on the NSGA3 algorithm, in the process of obtaining a designed polymer formula from the last generation population, the fast non-dominated sorting strategy is used on the last generation population to obtain non-dominated solution sets of different levels; the first level non-dominated solution set is taken out from the non-dominated solution set, and the first level non-dominated solution set is decoded to obtain the polymer formula.

[0018] The present invention also provides a polymer formula design device based on the NSGA3 algorithm, based on the above polymer formula design method based on the NSGA3 algorithm, comprising:

[0019] An initial data set arrangement module is used to arrange and obtain an initial data set by setting a database; the initial data set includes several raw material ratio fields and several polymer performance fields;

[0020] A data set preprocessing module, used for dividing the several raw material ratio fields in the initial data set into several main raw material ratio fields and several secondary raw material ratio fields according to a set threshold; converting the several secondary raw material ratio fields into categorical variable fields to obtain a preprocessed data set;

[0021] Each performance prediction model construction and training module is used to construct each performance prediction model according to the regression prediction model; train each performance prediction model through the preprocessed data set to obtain the trained each performance prediction model;

[0022] An individual fitness acquisition module is used to design a fitness function, and evaluate the individuals of the population through the fitness function to obtain individual fitness;

[0023] A population selection module is used to select individuals with high individual fitness from the parent and child populations as the next generation population by selecting an operation operator;

[0024] The polymer formula acquisition module is used to process the next generation population through crossover and mutation operators to obtain a new offspring population; iteratively select the new offspring population until a set generation is reached, and obtain the polymer formula from the last generation population.

[0025] As a preferred solution of a polymer formula design device based on the NSGA3 algorithm, in the individual fitness acquisition module, in the process of evaluating the individuals of the population through the fitness function to obtain the individual fitness, the individuals are evaluated and predicted through each performance prediction model to obtain each performance corresponding to the individual; each performance of the individual is compared with the target performance, and the individual fitness is obtained through the fitness function.

[0026] As a preferred solution of a polymer formula design device based on the NSGA3 algorithm, in the population selection module, in the process of selecting the next generation population from the parent and offspring populations, uniformly distributed reference points are generated in the target space; the population after the parent and offspring are merged is used, according to the individual fitness, to obtain non-dominated solution sets of different levels through a fast non-dominated sorting strategy; the non-dominated solution set is associated with the reference point, and the same number of individuals as the parent or offspring population are selected to form the next generation population.

[0027] As a preferred solution of a polymer formula design device based on the NSGA3 algorithm, in the polymer formula acquisition module, in the process of processing the next generation population through crossover and mutation operators to obtain the new offspring population, the next generation population is subjected to crossover and mutation processing at the same time, or crossover processing is performed alone, or mutation processing is performed alone.

[0028] As a preferred solution of a polymer formula design device based on the NSGA3 algorithm, in the polymer formula acquisition module, in the process of obtaining the designed polymer formula from the last generation population, the fast non-dominated sorting strategy is used on the last generation population to obtain non-dominated solution sets of different levels; the first level non-dominated solution set is taken out from the non-dominated solution set, and the first level non-dominated solution set is decoded to obtain the polymer formula.

[0029] The present invention has the following advantages: by setting a database, an initial data set is obtained; the initial data set includes several raw material ratio fields and several polymer performance fields; the several raw material ratio fields in the initial data set are divided into several main raw material ratio fields and several secondary raw material ratio fields according to a set threshold; the several secondary raw material ratio fields are converted into categorical variable fields to obtain a preprocessed data set; each performance prediction model is constructed according to a regression prediction model; each performance prediction model is trained by the preprocessed data set to obtain the trained each performance prediction model; a fitness function is designed, and individuals of a population are evaluated by the fitness function to obtain individual fitness; individuals with high individual fitness are selected from parent and child populations as the next generation population by selecting an operation operator; the next generation population is processed by crossover and mutation operation operators to obtain a new child population; the new child population is iteratively selected and processed until a set generation is reached, and a polymer formula is obtained from the last generation population. The present invention does not need to consider the physical and chemical process, but only needs to learn the mapping relationship between raw materials and performance from the data, so as to realize the prediction of raw materials to product performance and further combine the intelligent optimization algorithm to reversely design the formula. Under the condition of the same other process conditions, it can provide more direct data guidance for the experimental optimization of raw material ratio, thereby improving the optimization efficiency of raw material ratio. At the same time, when there are many product target performances that need to be designed, the number of non-dominated solutions increases sharply. It is difficult for the reverse-designed formula of the traditional intelligent optimization algorithm to ensure the uniformity and diversity of the solution. By introducing reference points in the target space to guide the selection of non-dominated solutions, the optimization effect of raw material ratio can be effectively improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the implementation methods of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the implementation methods or the description of the prior art. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other implementation drawings can be derived from the provided drawings without creative work.

[0031] The structures, proportions, sizes, etc. illustrated in this specification are only used to match the contents disclosed in the specification so as to facilitate understanding and reading by persons familiar with the technology. They are not used to limit the conditions under which the present invention can be implemented, and therefore have no substantial technical significance. Any structural modification, change in proportion or adjustment of size shall still fall within the scope of the technical contents disclosed in the present invention without affecting the effects and purposes that can be achieved by the present invention.

[0032] Figure 1A schematic flow chart of a polymer formulation design method based on the NSGA3 algorithm provided in Example 1 of the present invention;

[0033] Figure 2 This is a schematic diagram of the architecture of a polymer formulation design device based on the NSGA3 algorithm provided in Example 2 of the present invention. DETAILED DESCRIPTION

[0034] The following is a description of the implementation of the present invention by specific embodiments. People familiar with the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0035] Example 1

[0036] See also Figure 1 Embodiment 1 of the present invention provides a polymer formulation design method based on the NSGA3 algorithm, comprising the following steps:

[0037] S1. By setting a database, an initial data set is obtained; the initial data set includes several raw material ratio fields and several polymer performance fields;

[0038] S2, dividing the several kinds of raw material ratio fields in the initial data set into several kinds of main raw material ratio fields and several kinds of secondary raw material ratio fields according to a set threshold; converting the several kinds of secondary raw material ratio fields into categorical variable fields to obtain a preprocessed data set;

[0039] S3, constructing each performance prediction model according to the regression prediction model; training each performance prediction model by using the preprocessed data set to obtain each trained performance prediction model;

[0040] S4, designing a fitness function, and evaluating the individuals of the population by using the fitness function to obtain the individual fitness;

[0041] S5, selecting the individuals with high individual fitness from the parent and offspring populations as the next generation population by selecting an operation operator;

[0042] S6. Processing the next generation population through crossover and mutation operators to obtain a new offspring population; performing iterative selection operations on the new offspring population until a set generation is reached, and obtaining a polymer formula from the last generation population.

[0043] In this embodiment, in step S1, an initial data set is obtained by setting a database; the initial data set includes several raw material ratio fields and several polymer performance fields;

[0044] Specifically, by setting up the database, an initial data set including several raw material ratio fields and several polymer performance fields is obtained. For example, there are 97 raw material ratio fields, 4 polymer performance fields, and 150 samples in the initial data set.

[0045] In this embodiment, in step S2, the several raw material ratio fields in the initial data set are divided into several main raw material ratio fields and several secondary raw material ratio fields according to a set threshold; the several secondary raw material ratio fields are converted into categorical variable fields to obtain a preprocessed data set;

[0046] Specifically, several raw material ratio fields in the initial data set are divided into several main raw material ratio fields and several secondary raw material ratio fields according to set thresholds, and the secondary raw material ratio fields are converted into 0 and 1 categorical variable fields respectively to obtain the preprocessed data set.

[0047] For example, the threshold is set to 2%, that is, the ratio below the threshold is the secondary raw material ratio, and the ratio above the threshold is the primary raw material ratio. The non-zero value in the secondary raw material ratio field is set to 1, thereby converting it into a 0, 1 categorical variable field. There are 50 primary raw material ratio fields and 47 secondary raw material ratio fields.

[0048] In this embodiment, in step S3, each performance prediction model is constructed according to the regression prediction model; each performance prediction model is trained by the preprocessed data set to obtain the trained performance prediction model;

[0049] Specifically, a regression prediction model is trained for each polymer property using the preprocessed data set through a machine learning algorithm to obtain each property prediction model.

[0050] For example: the machine learning algorithm uses XGBoost, and the parameters of the four performance prediction models are max_depth=4, learning_rate=0.05, and n_estimators=500.

[0051] In this embodiment, in step S4, a fitness function is designed, and the individuals of the population are evaluated by the fitness function to obtain the individual fitness;

[0052] Specifically, each performance prediction model is used to evaluate and predict the individual to obtain each performance corresponding to the individual; each performance of the individual is compared with the target performance, and the fitness of the individual is obtained through the fitness function. The design principle of the fitness function is that the closer the predicted performance corresponding to the individual is to the target performance, the higher the fitness.

[0053] The expression of the fitness function is:

[0054] Fitness i =(y i -y_pred i ) 2 ,i=1,2,3,4

[0055] In the formula, Fitness i represents the fitness of the i-th target performance; y i Indicates the actual value of the i-th target performance; y_pred i represents the predicted value of the i-th target performance.

[0056] For example: an individual is defined as a 97-dimensional vector, the first 50 dimensions represent the gene length of the main raw material, and the last 47 dimensions represent the 0, 1 binary vector of the secondary raw material.

[0057] In this embodiment, in step S5, the individual with high individual fitness is selected from the parent and child populations as the next generation population by selecting an operation operator;

[0058] Specifically, in the process of selecting the next generation population from the parent and offspring populations, uniformly distributed reference points are generated in the target space; the population after the parent and offspring are merged is used, according to the individual fitness, to obtain non-dominated solution sets of different levels through a fast non-dominated sorting strategy; the non-dominated solution set is associated with the reference points, and the same number of individuals as the parent or offspring population are selected to form the next generation population.

[0059] Among them, the association between the non-dominated solution set and the reference point is that each non-dominated solution belongs to the reference point closest to it, and an individual is selected from the non-dominated solution set associated with each reference point to form the next generation population.

[0060] In the embodiment, in step S6, the next generation population is processed by crossover and mutation operators to obtain a new offspring population; the new offspring population is processed by iterative selection operations until a set generation is reached, and a polymer formula is obtained from the last generation population.

[0061] The individuals in the obtained new offspring population may have undergone both crossover and mutation operations or only one of them. The crossover operation uses two-point crossover, with the crossover probability set to 0.7 and the mutation probability set to 0.2.

[0062] Specifically, the fast non-dominated sorting method is used on the last generation population to obtain non-dominated solution sets of different levels, and then the first-level non-dominated solution set is taken out and then decoded into the designed polymer formula.

[0063] For example: specify the number of iterations to be 500 and the population size to be 100. Decoding the polymer formula means counting the proportion of each raw material in the first 50-dimensional main raw material genes in the 97-dimensional vector representing the individual, and all the secondary raw materials used in the last 47-dimensional secondary raw material vector.

[0064] In a possible embodiment, a specific formulation design example is provided as follows:

[0065] Product performance and formula design are shown in Table 1:

[0066]

[0067]

[0068] Table 1 Product performance and formula design data table

[0069] It can be concluded from Table 1 that the target values ​​of the four product properties are density = 1.175, flexural strength = 80, Vicat softening temperature = 120, and heat deformation temperature = 100. After iterative optimization by the multi-objective NSGA3 algorithm, a series of formulas are obtained, one of which is shown in the table above. Its density prediction value is 1.173, with a relative error of 0.17%, its flexural strength prediction value is 84.1, with a relative error of 5.13%, its Vicat softening temperature prediction value is 120.2, with a relative error of 0.17%, and its heat deformation temperature prediction value is 103.6, with a relative error of 3.60%.

[0070] In summary, the present invention obtains an initial data set by setting a database; the initial data set includes several raw material ratio fields and several polymer performance fields; the several raw material ratio fields in the initial data set are divided into several main raw material ratio fields and several minor raw material ratio fields according to a set threshold; the several minor raw material ratio fields are converted into categorical variable fields to obtain a preprocessed data set; each performance prediction model is constructed according to a regression prediction model; each performance prediction model is trained by the preprocessed data set to obtain the trained each performance prediction model; a fitness function is designed, and individuals of a population are evaluated by the fitness function to obtain individual fitness; individuals with high individual fitness are selected from parent and child populations as the next generation population by selecting an operation operator; the next generation population is processed by crossover and mutation operation operators to obtain a new child population; the new child population is iteratively selected and processed until the set generation is reached, and a polymer formula is obtained from the last generation population. The present invention does not need to consider the physical and chemical process, but only needs to learn the mapping relationship between raw materials and performance from the data, so as to realize the prediction of raw materials to product performance and further combine the intelligent optimization algorithm to reversely design the formula. Under the condition of the same other process conditions, it can provide more direct data guidance for the experimental optimization of raw material ratio, thereby improving the optimization efficiency of raw material ratio. At the same time, when there are many product target performances that need to be designed, the number of non-dominated solutions increases sharply. It is difficult for the reverse-designed formula of the traditional intelligent optimization algorithm to ensure the uniformity and diversity of the solution. By introducing reference points in the target space to guide the selection of non-dominated solutions, the optimization effect of raw material ratio can be effectively improved.

[0071] It should be noted that the method of the embodiment of the present disclosure can be performed by a single device, such as a computer or a server. The method of the present embodiment can also be applied in a distributed scenario and completed by multiple devices cooperating with each other. In the case of such a distributed scenario, one of the multiple devices can only perform one or more steps in the method of the embodiment of the present disclosure, and the multiple devices will interact with each other to complete the described method.

[0072] It should be noted that the above describes some embodiments of the present disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the above embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0073] Example 2

[0074] See also Figure 2 Embodiment 2 of the present invention further provides a polymer formulation design device based on the NSGA3 algorithm, comprising:

[0075] The initial data set arrangement module 001 is used to arrange and obtain the initial data set by setting the database; the initial data set includes several raw material ratio fields and several polymer performance fields;

[0076] The data set preprocessing module 002 is used to divide the several raw material ratio fields in the initial data set into several main raw material ratio fields and several secondary raw material ratio fields according to a set threshold; convert the several secondary raw material ratio fields into categorical variable fields to obtain a preprocessed data set;

[0077] Each performance prediction model construction and training module 003 is used to construct each performance prediction model according to the regression prediction model; train each performance prediction model through the preprocessed data set to obtain the trained each performance prediction model;

[0078] The individual fitness acquisition module 004 is used to design a fitness function, and evaluate the individuals of the population through the fitness function to obtain the individual fitness;

[0079] The population selection module 005 is used to select the individuals with high individual fitness from the parent and child populations as the next generation population by selecting an operation operator;

[0080] The polymer formula acquisition module 006 is used to process the next generation population through crossover and mutation operators to obtain a new offspring population; iteratively select the new offspring population until a set generation is reached, and obtain the polymer formula from the last generation population.

[0081] In this embodiment, in the individual fitness acquisition module 004, in the process of evaluating the individuals of the population through the fitness function to obtain the individual fitness, the individuals are evaluated and predicted through each performance prediction model to obtain each performance corresponding to the individual; each performance of the individual is compared with the target performance, and the individual fitness is obtained through the fitness function.

[0082] In this embodiment, in the population selection module 005, in the process of selecting the next generation population from the parent and offspring populations, uniformly distributed reference points are generated in the target space; the population after the parent and offspring are merged is used, according to the individual fitness, to obtain non-dominated solution sets of different levels through a fast non-dominated sorting strategy; the non-dominated solution set is associated with the reference point, and the same number of individuals as the parent or offspring population are selected to form the next generation population.

[0083] In this embodiment, in the polymer formula acquisition module 006, in the process of processing the next generation population through crossover and mutation operators to obtain the new offspring population, the next generation population is subjected to crossover and mutation processing at the same time, or crossover processing is performed alone, or mutation processing is performed alone.

[0084] In this embodiment, in the polymer formula acquisition module 006, in the process of obtaining the designed polymer formula from the last generation population, the fast non-dominated sorting strategy is used on the last generation population to obtain non-dominated solution sets of different levels; the first level non-dominated solution set is taken out from the non-dominated solution set, and the first level non-dominated solution set is decoded to obtain the polymer formula.

[0085] It should be noted that the information interaction, execution process and other contents between the modules of the above-mentioned system are based on the same concept as the method embodiment in Example 1 of the present application, and the technical effects they bring are the same as those of the method embodiment of the present application. For specific contents, please refer to the description in the method embodiment shown above in the present application, and will not be repeated here.

[0086] Example 3

[0087] Embodiment 3 of the present invention provides a non-transitory computer-readable storage medium, in which a program code for a polymer formulation design method based on the NSGA3 algorithm is stored. The program code includes instructions for executing the polymer formulation design method based on the NSGA3 algorithm of Embodiment 1 or any possible implementation thereof.

[0088] The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that includes one or more available media. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)).

[0089] Example 4

[0090] Embodiment 4 of the present invention provides an electronic device, including: a memory and a processor;

[0091] The processor and the memory communicate with each other via a bus; the memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute a polymer formulation design method based on the NSGA3 algorithm of Example 1 or any possible implementation thereof.

[0092] Specifically, the processor can be implemented by hardware or by software. When implemented by hardware, the processor can be a logic circuit, an integrated circuit, etc.; when implemented by software, the processor can be a general-purpose processor implemented by reading software codes stored in a memory. The memory can be integrated in the processor or can be located outside the processor and exist independently.

[0093] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable systems. The computer instructions can be stored in a computer-readable storage medium, or transmitted from a computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center.

[0094] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing system, they can be concentrated on a single computing system, or distributed on a network composed of multiple computing systems, and optionally, they can be implemented by a program code executable by a computing system, so that they can be stored in a storage system and executed by the computing system, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0095] Although the present invention has been described in detail above by general description and specific embodiments, it is obvious to those skilled in the art that some modifications or improvements can be made to the present invention. Therefore, these modifications or improvements made without departing from the spirit of the present invention all belong to the scope of protection claimed by the present invention.

Claims

1. A polymer formulation design method based on NSGA3 algorithm, characterized in that: include: By setting up a database, an initial data set is obtained; the initial data set includes several raw material ratio fields and several polymer performance fields; Dividing the several raw material ratio fields in the initial data set into several main raw material ratio fields and several secondary raw material ratio fields according to a set threshold; converting the several secondary raw material ratio fields into categorical variable fields to obtain a preprocessed data set; Construct each performance prediction model according to the regression prediction model; train each performance prediction model through the preprocessed data set to obtain each trained performance prediction model; Designing a fitness function, evaluating individuals of the population by using the fitness function to obtain individual fitness; By selecting an operation operator, the individuals with high individual fitness are selected from the parent and child populations as the next generation population; Processing the next generation population by crossover and mutation operators to obtain a new generation population; The new progeny population is subjected to an iterative selection operation until a set generation is reached, and a polymer formula is obtained from the last generation population; The expression of the fitness function is: Fitness i =(y i -y_pred i ) 2 ,i=1,2,3,4 In the formula, Fitness i represents the fitness of the i-th target performance; y i Indicates the actual value of the i-th target performance; y_pred i represents the predicted value of the i-th target performance.

2. A polymer formulation design method based on NSGA3 algorithm according to claim 1, characterized in that: In the process of evaluating the individuals of the population through the fitness function to obtain the individual fitness, the individuals are evaluated and predicted through each performance prediction model to obtain each performance corresponding to the individual; each performance of the individual is compared with the target performance, and the individual fitness is obtained through the fitness function.

3. A polymer formulation design method based on NSGA3 algorithm according to claim 2, characterized in that: In the process of selecting the next generation population from the parent and offspring populations, uniformly distributed reference points are generated in the target space; the population after the parent and offspring are merged is used to obtain non-dominated solution sets of different levels according to the individual fitness through a fast non-dominated sorting strategy; the non-dominated solution set is associated with the reference points, and the same number of individuals as the parent or offspring population are selected to form the next generation population.

4. A polymer formulation design method based on NSGA3 algorithm according to claim 3, characterized in that: In the process of processing the next generation population through crossover and mutation operators to obtain the new offspring population, the next generation population is subjected to crossover and mutation processing simultaneously, or crossover processing is performed alone, or mutation processing is performed alone.

5. A polymer formulation design method based on NSGA3 algorithm according to claim 4, characterized in that: In the process of obtaining the designed polymer formula from the last generation population, the fast non-dominated sorting strategy is applied to the last generation population to obtain non-dominated solution sets of different levels; A first-level non-dominated solution set is taken out from the non-dominated solution set, and the first-level non-dominated solution set is decoded to obtain a designed polymer formula.

6. A polymer formula design device based on NSGA3 algorithm, using a polymer formula design method based on NSGA3 algorithm according to any one of claims 1 to 5, characterized in that: include: An initial data set arrangement module is used to arrange and obtain an initial data set by setting a database; The initial data set includes several raw material ratio fields and several polymer performance fields; A data set preprocessing module, used for dividing the several raw material ratio fields in the initial data set into several main raw material ratio fields and several secondary raw material ratio fields according to a set threshold; converting the several secondary raw material ratio fields into categorical variable fields to obtain a preprocessed data set; Each performance prediction model construction and training module is used to construct each performance prediction model according to the regression prediction model; train each performance prediction model through the preprocessed data set to obtain the trained each performance prediction model; The individual fitness acquisition module is used to design a fitness function, and evaluate the individuals of the population through the fitness function to obtain the individual fitness; the expression of the fitness function is: Fitness i =(y i -y_pred i ) 2 ,i=1,2,3,4 In the formula, Fitness i represents the fitness of the i-th target performance; y i represents the actual value of the i-th target performance; y_pred i represents the predicted value of the i-th target performance; A population selection module is used to select individuals with high individual fitness from the parent and child populations as the next generation population by selecting an operation operator; A polymer formula acquisition module is used to process the next generation population through crossover and mutation operators to obtain a new generation population; The new offspring population is subjected to iterative selection operation processing until a set generation is reached, and a polymer formula is obtained from the last generation population.

7. A polymer formulation design device based on NSGA3 algorithm according to claim 6, characterized in that: In the individual fitness acquisition module, in the process of evaluating the individuals of the population through the fitness function to obtain the individual fitness, the individuals are evaluated and predicted through each performance prediction model to obtain each performance corresponding to the individual; each performance of the individual is compared with the target performance, and the individual fitness is obtained through the fitness function.

8. The polymer formulation design device based on NSGA3 algorithm according to claim 7, characterized in that: In the population selection module, in the process of selecting the next generation population from the parent and offspring populations, uniformly distributed reference points are generated in the target space; the population after the parent and offspring are merged is used to obtain non-dominated solution sets of different levels according to the individual fitness through a fast non-dominated sorting strategy; the non-dominated solution set is associated with the reference points, and the same number of individuals as the parent or offspring population are selected to form the next generation population.

9. The polymer formulation design device based on NSGA3 algorithm according to claim 8, characterized in that: In the polymer formula acquisition module, in the process of processing the next generation population through crossover and mutation operators to obtain the new offspring population, the next generation population is subjected to crossover and mutation processing simultaneously, or crossover processing is performed alone, or mutation processing is performed alone.

10. The polymer formulation design device based on NSGA3 algorithm according to claim 9, characterized in that: In the polymer formula acquisition module, in the process of obtaining the designed polymer formula from the last generation population, the fast non-dominated sorting strategy is used on the last generation population to obtain non-dominated solution sets of different levels; A first-level non-dominated solution set is taken out from the non-dominated solution set, and the first-level non-dominated solution set is decoded to obtain a polymer formula.

Images