Reaction rate prediction method, device and related equipment based on machine learning

Abstract

The present application provides a method, apparatus, and related equipment for predicting reaction rates based on machine learning. The method comprises: obtaining basic reaction conditions of a multi-functional group molecular model; performing a polymerization reaction simulation on the multi-functional group molecular model based on the basic reaction conditions to obtain first simulated molecular weight distribution data; training a preset neural network based on the first simulated molecular weight distribution data to obtain a polymerization reaction rate prediction model; obtaining molecular weight distribution data to be predicted; and performing reaction rate prediction on the molecular weight distribution data to be predicted based on the polymerization reaction rate prediction model to obtain reaction rate prediction data. The present application obtains first simulated molecular weight distribution data by performing a polymerization reaction simulation with adjustable reaction rate on the multi-functional group molecular model and trains a preset neural network, which is conducive to the polymerization reaction rate prediction model providing a predicted value of the reaction rate coefficient ratio of different types of reaction functional groups during the polymerization reaction.

Description

Technical Field

[0001] The present invention relates to the technical field of polymer material polymerization, and in particular to a reaction rate prediction method, device and related equipment based on machine learning. Background Art

[0002] The molecular weight distribution of polymers is a fundamental property of the material, significantly influencing properties such as dynamic modulus, fracture toughness, glass transition temperature, and viscosity. Therefore, accurately predicting molecular weight distribution before actual synthesis has always been an important research direction in polymer science.

[0003] However, existing simulation methods have limitations when predicting molecular weight distribution. For example, the Flory-Stockmayer theory tends to overlook the formation of ring structures during reactions. Furthermore, traditional functional group molecular models assume that the reaction rates of all monomers are constant throughout the reaction, meaning that the reactivity of the functional groups remains constant throughout the reaction. This makes them unsuitable for complex reactions and only applicable to systems with a low degree of polymerization or catalyst-free systems, making them inapplicable to a wider range of polymer polymerization simulation scenarios. Summary of the Invention

[0004] This application proposes a reaction rate prediction method, device and related equipment based on machine learning, which is applicable to more polymer polymerization simulation scenarios and is conducive to the polymerization reaction rate prediction model providing predicted values ​​of the reaction rate coefficient ratios of different types of reaction functional groups during the polymerization reaction.

[0005] In a first aspect, a reaction rate prediction method based on machine learning is provided, comprising:

[0006] Obtain the basic reaction conditions of multi-functional molecular models;

[0007] Performing a polymerization reaction simulation on the multifunctional molecular model based on the reaction basic conditions to obtain first simulated molecular weight distribution data;

[0008] Training a preset neural network based on the first simulated molecular weight distribution data to obtain a polymerization reaction rate prediction model;

[0009] Obtaining molecular weight distribution data to be predicted;

[0010] The reaction rate is predicted based on the polymerization reaction rate prediction model for the molecular weight distribution data to be predicted to obtain reaction rate prediction data.

[0011] In a second aspect, a reaction rate prediction device based on machine learning is provided, comprising:

[0012] The first acquisition module is used to obtain the reaction basic conditions of the multi-functional group molecular model;

[0013] A first simulation module is used to perform a polymerization reaction simulation on the multifunctional molecular model based on the reaction basic conditions to obtain first simulated molecular weight distribution data;

[0014] A training module, configured to train a preset neural network based on the first simulated molecular weight distribution data to obtain a polymerization reaction rate prediction model;

[0015] A second acquisition module is used to obtain molecular weight distribution data to be predicted;

[0016] The first prediction module is used to perform reaction rate prediction on the molecular weight distribution data to be predicted based on the polymerization reaction rate prediction model to obtain reaction rate prediction data.

[0017] Optionally, in some embodiments of the present application, the device further includes:

[0018] A third acquisition module is used to obtain actual molecular weight distribution data of the multi-functional group molecular model;

[0019] A second prediction module is used to perform reaction rate prediction on the actual molecular weight distribution data based on the polymerization reaction rate prediction model to obtain actual reaction rate data;

[0020] a second simulation module, configured to update the reaction basic conditions based on the actual reaction rate data, and perform a polymerization reaction simulation on the multifunctional molecular model based on the updated reaction basic conditions to obtain second simulated molecular weight distribution data;

[0021] A verification module is used to verify the polymerization reaction rate prediction model according to the actual molecular weight distribution data and the second simulated molecular weight distribution data.

[0022] Optionally, in some embodiments of the present application, the basic reaction conditions include scanning rate ratio parameters and reaction association parameters, the scanning rate ratio parameters include the reaction rate of each functional group in the multi-functional molecular model under different conditions, and the reaction association parameters include the concentration of each functional group in the multi-functional molecular model under different conditions.

[0023] Optionally, in some embodiments of the present application, different functional groups have different concentrations.

[0024] Optionally, in some embodiments of the present application, the first simulation module includes:

[0025] A calculation submodule, configured to calculate the reaction rate of each functional group in the multi-functional group molecular model under different conditions using a preset reaction rate calculation formula;

[0026] The simulation submodule is used to use a Monte Carlo algorithm to simulate the polymerization reaction of the reaction rate of each functional group in the multi-functional molecular model under different conditions based on the reaction basic conditions to obtain the first simulated molecular weight distribution data.

[0027] Optionally, in some embodiments of the present application, the simulation submodule includes:

[0028] a determination unit, configured to determine a target reaction based on the reaction rates of the functional groups in the multi-functional group molecular model under different conditions;

[0029] a simulation unit, configured to perform a polymerization reaction simulation on any two unreacted functional groups in the multi-functional molecular model based on the target reaction;

[0030] The obtaining unit is used to return to calculating the reaction rate of any reaction in the multi-functional molecular model and continue to execute if the reaction degree of the polymerization reaction simulation does not reach the prediction threshold, so as to obtain the first simulated molecular weight distribution data.

[0031] Optionally, in some embodiments of the present application, the multi-functional group molecular model includes functional group A and functional group B, and the reaction rate calculation formula is expressed as:

[0032]

[0033] Where R represents the reaction rate, represents the reaction rate coefficient of the i-th case of functional group A, represents the concentration of functional group A in the ith case, is the reaction rate coefficient of the jth case of functional group B, is the concentration of the jth instance of the B functional group, N represents the number of different instances of the A functional group, and M represents the number of different instances of the B functional group.

[0034] In a third aspect, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned reaction rate prediction method based on machine learning when executing the computer program.

[0035] In a fourth aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned reaction rate prediction method based on machine learning are implemented.

[0036] The present application provides a method, device, computer equipment and storage medium for predicting reaction rate based on machine learning, which obtains the reaction basic conditions of a multi-functional molecular model; performs a polymerization reaction simulation on the multi-functional molecular model based on the reaction basic conditions to obtain first simulated molecular weight distribution data; trains a preset neural network based on the first simulated molecular weight distribution data to obtain a polymerization reaction rate prediction model; obtains molecular weight distribution data to be predicted; and performs reaction rate prediction on the molecular weight distribution data to be predicted based on the polymerization reaction rate prediction model to obtain reaction rate prediction data. In the method for predicting reaction rate based on machine learning provided in the present application, by performing a polymerization reaction simulation with adjustable reaction rate on the multi-functional molecular model under the constraints of the reaction basic conditions, the molecular weight distribution under different reaction conditions, i.e., the first simulated molecular weight distribution data, can be obtained. Then, the preset neural network is trained using the first simulated molecular weight distribution data, which is conducive to the neural network learning the law of molecular weight change of the multi-functional molecular model during the polymerization reaction from the molecular weight distribution under different reaction conditions, thereby being applicable to more polymer polymerization simulation scenarios, and being conducive to the polymerization reaction rate prediction model providing a predicted value of the reaction rate coefficient ratio of different types of reactive functional groups during the polymerization reaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0038] Figure 1 A diagram illustrating the application environment of the reaction rate prediction method based on machine learning provided in an embodiment of the present application;

[0039] Figure 2 A flowchart of a reaction rate prediction method based on machine learning provided in an embodiment of the present application;

[0040] Figure 3 A schematic diagram of a multifunctional molecular model provided in an embodiment of the present application;

[0041] Figure 4 A schematic diagram of a polymerization reaction simulation process provided in an embodiment of the present application;

[0042] Figure 5 A schematic diagram of the structure of a neural network provided in an embodiment of the present application;

[0043] Figure 6A schematic flow chart of a reaction rate prediction method based on machine learning provided in yet another embodiment of the present application;

[0044] Figure 7 A bar chart comparing the structure prediction of the polymerization reaction rate prediction model provided by an embodiment of the present application and the traditional model;

[0045] Figure 8 A bar chart comparing the structure prediction of a polymerization reaction rate prediction model provided by another embodiment of the present application and a traditional model;

[0046] Figure 9 A scatter plot comparing the molecular weight distribution of the polymerization reaction rate prediction model provided in one embodiment of the present application and the traditional model;

[0047] Figure 10 A scatter plot comparing molecular weight distributions of a polymerization reaction rate prediction model provided in another embodiment of the present application and a traditional model;

[0048] Figure 11 A structural block diagram of a reaction rate prediction device based on machine learning provided in an embodiment of the present application;

[0049] Figure 12 This is a structural block diagram of the computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0050] 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 them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0051] In addition, described feature, structure or characteristic can be combined in one or more embodiments in any suitable manner.In the following description, many specific details are provided so as to provide a full understanding of the embodiments of the present application. However, it will be appreciated by those skilled in the art that the technical scheme of the present application can be put into practice without one or more of the specific details, or other methods, components, devices, steps etc. can be adopted. In other cases, known methods, devices, implementations or operations are not shown or described in detail to avoid blurring the various aspects of the application.

[0052] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically separate entities. That is, these functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0053] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may be decomposed, while others may be combined or partially combined. Therefore, the actual execution order may vary depending on the actual situation.

[0054] The reaction rate prediction method based on machine learning provided by the embodiment of the present invention can be applied in Figure 1 In the application environment. Among them, the computer device 110 communicates with the server 120 through the network 130. The computer device 110 can obtain the reaction basic conditions of the multi-functional molecular model; perform polymerization reaction simulation on the multi-functional molecular model based on the reaction basic conditions to obtain first simulated molecular weight distribution data; train a preset neural network based on the first simulated molecular weight distribution data to obtain a polymerization reaction rate prediction model; obtain molecular weight distribution data to be predicted; perform reaction rate prediction on the molecular weight distribution data to be predicted based on the polymerization reaction rate prediction model to obtain reaction rate prediction data, and display it through the computer device 110. In the present invention, by performing polymerization reaction simulation on the multi-functional molecular model under the constraint of the reaction basic conditions, the molecular weight distribution under different reaction conditions, that is, the first simulated molecular weight distribution data, can be obtained. Then, the preset neural network is trained using the first simulated molecular weight distribution data, which is conducive to the neural network learning the law of molecular weight change of the multi-functional molecular model during the polymerization reaction from the molecular weight distribution under different reaction conditions, thereby being applicable to more polymer polymerization simulation scenarios, and being conducive to the polymerization reaction rate prediction model providing a predicted value of the reaction rate coefficient ratio of different types of reaction functional groups during the polymerization reaction. The computer device 110 may be, but is not limited to, various smart phones 110 - 1 , tablet computers 110 - 2 , and notebook computers 110 - 3 . The present invention will be described in detail below through specific embodiments.

[0055] See also Figure 2 As shown, Figure 2 This is a flow chart of a method for predicting reaction rates based on machine learning provided by an embodiment of the present invention. This method can be applied to both a terminal and a server. This embodiment uses the server as an example. The method for predicting reaction rates based on machine learning includes the following steps:

[0056] S101: Obtain the basic reaction conditions of the multifunctional molecular model.

[0057] A multifunctional molecular model refers to a particle containing at least two functional groups. This particle records information about the reactive functional groups it contains. Functional groups can undergo polymerization reactions to link with other polymer monomers. A single polymer monomer can contain one or more functional groups.

[0058] The multi-functional group molecule model includes the molecule name, relative molecular mass, and a hash table. The hash table stores the names of different functional groups on a single molecule and the corresponding number of functional group names, such as "-OH": 3.

[0059] For example, it is assumed that the multi-functional group molecular model includes molecule A, molecule B and molecule C, as shown in Table 1, the type and number of functional groups contained in each molecule, and a set of reaction rates corresponding to each functional group.

[0060] Table 1

[0061]

[0062] In Table 1,<a1, 1, 2> It means that the reaction rate of A molecule with 2 remaining a1 functional groups is 2, and the reaction rate of A molecule with 1 remaining a1 functional group is 1;<b1, 1, 2.2, 3.3> The reaction rate of the b1 functional group of the B molecule with 3 remaining b1 functional groups is 3.3, the reaction rate of the b1 functional group of the B molecule with 2 remaining b1 functional groups is 2.2, and the reaction rate of the b1 functional group of the B molecule with 1 remaining b1 functional group is 1;<b2, 2, 3, 2> Indicates that the reaction rate of the b2 functional group of the B molecule with 3 remaining b2 functional groups is 2, the reaction rate of the b2 functional group of the B molecule with 2 remaining b2 functional groups is 3, and the reaction rate of the b2 functional group of the B molecule with 1 remaining b2 functional group is 2;<c1, 2, 3.4, 5.1, 10> It means that the reaction rate of the C molecule with 4 remaining C1 functional groups is 10, the reaction rate of the C molecule with 3 remaining C1 functional groups is 5.1, the reaction rate of the C molecule with 2 remaining C1 functional groups is 3.4, and the reaction rate of the C molecule with 1 remaining C1 functional group is 2.

[0063] Assuming that the multifunctional molecular model includes reactant 1 and reactant 2, the functional group reaction rules of the multifunctional molecular model are shown in Table 2:

[0064] Table 2

[0065]

[0066] Table 2 shows that in the multifunctional molecular model, molecule A containing an unreacted a1 functional group can react once with molecule B containing an unreacted b1 functional group; molecule B containing an unreacted b2 functional group can react once with molecule C containing an unreacted c1 functional group.

[0067] In one embodiment, the basic reaction conditions include a scanning rate ratio parameter and a reaction association parameter, wherein the scanning rate ratio parameter includes the reaction rate of each functional group in the multi-functional molecular model under different conditions, and the reaction association parameter includes the concentration of each functional group in the multi-functional molecular model under different conditions.

[0068] Among them, different situations can be different parameters under the basic reaction conditions, such as different reaction degrees and reaction rate coefficient ratios between functional groups.

[0069] The polymerization reaction of the proportional coefficient of the reaction rate of the same functional group in different states can be simulated, so as to obtain different distributions of each molecular weight. Finally, the relationship between the reaction rate coefficient ratio of the functional group in different states and the molecular weight distribution can be obtained. Therefore, the proportional coefficient of the reaction rate of the functional group in different states (i.e., different situations) (i.e., the reaction rate coefficient ratio) can be predicted through the molecular weight distribution of a single molecule.

[0070] In one embodiment, different functional groups have different concentrations.

[0071] S102: performing polymerization reaction simulation on the multifunctional molecular model based on the reaction basic conditions to obtain first simulated molecular weight distribution data.

[0072] The basic reaction conditions may include relevant data on various molecular species, functional group reaction rules, reaction rate calculation formulas, and a set total number of reactions or reaction extent. The first simulated molecular weight distribution data may be data obtained by simulating the polymerization reaction of each functional group in the multifunctional molecular model under the basic reaction conditions. The data may include the statistical distribution of the number of molecules with different functional groups in the multifunctional molecular model.

[0073] The reaction basic conditions can include multiple pre-set reaction degrees and multiple reaction ratios. The total number of reactions refers to the total number of times all reaction rules in the multi-functional group molecular model occur.

[0074] Polymerization reaction simulation usually includes simulating the steps of chain initiation, chain growth, chain termination, and possible reactions such as chain transfer, branching and cross-linking in the polymerization process.

[0075] In one embodiment, a mathematical model of the polymerization reaction is established by a machine learning algorithm, and the molecular weight distribution of different reaction rate coefficient ratios of each functional group under different conditions is calculated in combination with the basic reaction conditions, such as the reaction rate constant, the reaction rate coefficient ratio, the reaction degree, etc., to obtain the first molecular weight distribution data.

[0076] In one embodiment, the molecular weight distribution of different reaction ratios of functional groups under different conditions can be predicted by simulating the movement behavior of molecules under reaction conditions, thereby obtaining first molecular weight distribution data.

[0077] Monte Carlo simulations of polymerization reactions with multiple groups of different reaction rate coefficient ratios are performed on the input multifunctional molecular model under different conditions. During the Monte Carlo simulation process, the reaction rate is dynamically adjusted according to a preset reaction rate calculation formula, so that the reaction rate can be calculated more accurately and the accuracy of the reaction rate calculation can be increased, thereby obtaining the first molecular weight distribution data. That is, in one embodiment, the polymerization reaction simulation of the multifunctional molecular model based on the reaction basic conditions is performed to obtain the first simulated molecular weight distribution data, including:

[0078] Calculate the reaction rate of each functional group in the multi-functional group molecular model under different conditions using a preset reaction rate calculation formula;

[0079] A Monte Carlo algorithm is used to simulate the polymerization reaction of the reaction rate coefficient ratios of the functional groups in the multifunctional molecular model under different conditions based on the reaction basic conditions to obtain the first simulated molecular weight distribution data.

[0080] For example, assuming that the multifunctional molecular model includes functional group A and functional group B, the reaction rate calculation formula can be expressed as:

[0081]

[0082] Where R represents the reaction rate, represents the reaction rate coefficient of the i-th case of functional group A, represents the concentration of functional group A in the ith case, is the reaction rate coefficient of the jth case of functional group B, is the concentration of the jth instance of the B functional group, N represents the number of different instances of the A functional group, and M represents the number of different instances of the B functional group.

[0083] In the traditional Monte Carlo algorithm, when the functional group changes between the i-th case and the j-th case during the simulation reaction, and In this embodiment, when the i-th situation and the j-th situation change, the changes will be and New predictions are made to automatically adjust the reactivity of functional groups. In subsequent reaction rate predictions, the polymerization reaction rate prediction model can be used to predict changes in the reaction rate of functional groups, making it applicable to more polymer polymerization simulation scenarios.

[0084] The reaction rate calculation formula for different functional groups is the same, but the concentration is different.

[0085] By improving the Monte Carlo algorithm, that is, adding a reaction rate calculation formula to the Monte Carlo algorithm, the Monte Carlo algorithm is adopted to calculate the reaction rate of each functional group in the multi-functional group molecular model under different conditions using the reaction rate calculation formula, and based on the reaction basic conditions, the polymerization reaction simulation of the reaction rate of each functional group in the multi-functional group molecular model under different conditions is performed to obtain the first simulated molecular weight distribution data.

[0086] In this embodiment, the changing reaction rate is predicted based on the reaction rate calculation formula, so that it can automatically adjust the reactivity of the functional group. Subsequently, machine learning is used to predict the change in the functional group reaction rate from the experimental data, making it applicable to more polymer polymerization simulation scenarios.

[0087] The Gillespie algorithm is a random simulation algorithm based on the Monte Carlo method that is often used in the field of polymer polymerization simulation. At present, the algorithm has the following shortcomings when applied in the field of polymer polymerization simulation: The algorithm and program are based on the activity theory of Flory et al., that is, during the reaction process, the reactivity of the functional group is a constant value throughout the reaction process. It is only applicable to systems with a small degree of polymerization or catalyst-free systems, which limits the scope of application of the algorithm. Therefore, this embodiment can perform Monte Carlo simulations of polymerization reactions with multiple groups of different reaction rate coefficient ratios for the incoming functional groups under different conditions, thereby obtaining the final molecular weight distribution of the system to obtain the first simulated molecular weight distribution data. That is, in one embodiment, the Monte Carlo algorithm is used to simulate the polymerization reaction of the reaction rate of each functional group in the multi-functional group molecular model under different conditions based on the reaction basic conditions to obtain the first simulated molecular weight distribution data, including:

[0088] Determining a target reaction based on the reaction rates of each functional group in the multifunctional molecular model under different circumstances;

[0089] Performing polymerization reaction simulation on any two unreacted functional groups in the multi-functional molecular model based on the target reaction;

[0090] If the reaction degree of the polymerization reaction simulation does not reach the prediction threshold, the calculation returns to the reaction rate of any reaction in the multifunctional molecular model and continues until the reaction degree of the polymerization reaction simulation reaches the prediction threshold to obtain the first simulated molecular weight distribution data.

[0091] For example, suppose a multifunctional molecular reaction requires setting five reaction rate coefficients: k1, k2, k3, k4, and k5. For a molecular system containing this reaction type, the system evolution needs to be simulated at ten evenly spaced reaction levels (10%, 20%, ..., 100%). For each reaction level, the simulation is performed across all parameter combinations to generate multiple sets of molecular distributions. The simulation parameters are set as follows:

[0092] Reaction extent stage: Set 10 reaction endpoints (10% to 100%, in 10% steps). Reaction rate coefficient base: k5 is set to 1 as a reference coefficient. The ratios of k1-k4 relative to k5 are set to 50 gradient values ​​(0.1 to 10, in 0.2 steps). Under these settings, execute the following nested loop process:

[0093] Using a five-layer nested loop structure:

[0094] Outer loop: the reaction degree is cycled from 10% to 100% (10% step, 10 times);

[0095] Second layer: k1 cycles from 0.1 to 5 (step size 0.2, 25 times);

[0096] The third layer: k2 cycles from 0.1 to 5 (step size 0.2, 25 times);

[0097] Fourth layer: k3 cycles from 0.1 to 5 (step size 0.2, 25 times);

[0098] Innermost layer: k4 loops from 0.1 to 5 (step size 0.2, 25 times);

[0099] After each loop completes all iterations, the upper-level variables are updated. Each parameter combination corresponds to a simulation scenario, and each reaction stage executes all four loops independently. The final result is: 10 (reaction stage) × 25^4 (parameter combination) sets of molecular weight distribution data, forming the first molecular weight distribution dataset.

[0100] Molecular weight distribution can be represented in the form of a multidimensional array.

[0101] Illustratively, the molecular weight distribution can be shown as:

[0102] [(0.1,0.2,0.3,...,1.0,1.1...,4.8,4.9,5.0),(0.01,0.02,0.04,......,0.20,0.25,....,0.07,0.04,0.01)], where the first () is the molecular mass and the second () is the probability density of the corresponding molecular mass.

[0103] The reaction ratio refers to the ratio of the number of molecules of different reactive monomers, and the reaction degree refers to the progress of the reaction, that is, how many functional groups in the multifunctional molecular model have reacted. The reaction between different monomers is the reaction of functional groups. The initial reaction ratio of different monomers will polymerize to generate polymers with different molecular weight distributions. For example, Figure 3 As shown in the figure, assuming a multifunctional molecular model includes 1000 functional groups A and 1000 functional groups B, if the reaction degree is set to 90%, it means that 900 A and 900 B undergo polymerization. Different reaction degrees will result in different molecular weight distributions of the generated polymers.

[0104] In one embodiment, in response to user input, basic reaction conditions can be obtained. The basic reaction conditions also include the type and quantity of molecules, the type and quantity of functional groups, the reaction rules of the functional groups, etc. A corresponding number of molecule classes and reaction rule classes can be initialized, and the correspondence between functional groups and reaction rates can be recorded using global variables. For example, the reaction rate corresponding to each functional group is shown in Table 1.

[0105] After obtaining the basic reaction conditions, such as Figure 4 As shown, the reaction rate of each possible reactive functional group in the multifunctional group molecular model is calculated, one reaction is randomly selected from the reaction rates of each possible reactive functional group, namely the target reaction, two monomers with unreacted functional groups are selected to perform a polymerization reaction on the selected target reaction, the two monomers are linked, and the multifunctional group molecular model is updated to determine whether the updated multifunctional group molecular model reaches the expected reaction degree (ie, the prediction threshold). If not, the reaction rate of each possible reaction is continued to be calculated based on the reaction rate calculation formula, and the execution is continued until the reaction degree of the multifunctional group molecular model reaches the expected reaction degree, and the molecular weight distribution of different reaction proportions corresponding to the functional groups under the reaction degree is output, so that the first molecular weight distribution data can be obtained.

[0106] Each molecule records its connections with other molecules. Based on this information, we can calculate the total molecular weight generated, the reaction status of each type of molecule (such as the proportion of unreacted functional groups remaining 1, 2, etc. in the total number of participating molecules, i.e., the K value), the number average molecular weight Mn, the weight average molecular weight Mw, and other data as the first molecular weight distribution data.

[0107] It should be noted that the basic reaction conditions also include environmental factors. In the Monte Carlo simulation algorithm, the values ​​of the reaction rates of monomer functional groups in different states are replaced by multiple variables so that the values ​​can be input and passed. These variables usually represent different reaction conditions or environmental factors, such as temperature, pressure, reactant concentration, etc., which may affect the reaction rate. For example, the reaction rate can be specified explicitly or using the Arrhenius expression. The values ​​under different states refer to the specific numerical values ​​of the reaction rates under these different conditions. For example, the reaction rate of a functional group at high temperature and low temperature will be different, and this difference can be simulated by setting different variable values. In this way, the simulation software can handle various reaction conditions more flexibly, thereby more accurately predicting the behavior and characteristics of the polymerization reaction.

[0108] S103: Training a preset neural network based on the first simulated molecular weight distribution data to obtain a polymerization reaction rate prediction model.

[0109] After the polymerization reaction simulation is completed, the first simulated molecular weight distribution data is used as the characteristic value, and the reaction rate coefficient ratio between functional groups at each reaction degree is used as the target value to train the preset neural network to obtain a polymerization reaction rate prediction model.

[0110] For example, in a polymerization reaction involving only A2+B3 (molecules A and B react with each other, with A having two reactive functional groups and B having three), five k values ​​are calculated: kA0, kA1, kB0, kB1, and kB2. kA0 represents the percentage of A molecules with zero unreacted functional groups compared to the total number of reacted molecules, and the remaining values ​​are similar. The first molecular weight distribution data obtained from the polymerization simulation of molecules A and B is used as the feature value, and the ratio of the reaction rate coefficients between the functional groups of molecules A and B at each reaction level is used as the target value to train a pre-set neural network, resulting in a polymerization reaction rate prediction model.

[0111] The target value is the reaction rate coefficient corresponding to the functional group of molecule A, the reaction rate coefficient corresponding to the functional group of molecule B, the ratio of the reaction rate coefficients of molecules A and B, and the final reaction degree achieved by the reaction.

[0112] Optionally, the preset neural network may be a feedforward neural network, such as Figure 5 As shown, the feedforward neural network includes an input layer, a hidden layer, and an output layer. In this application, the input layer is used to receive the feature values ​​of the external input, convert the feature values ​​into a numerical form that the neural network can process, and pass them to the hidden layer. The hidden layer is the key part of feature extraction and transformation. The hidden layer can have one or more layers, each containing a certain number of neurons. The output layer is used to generate the final output result.

[0113] After the polymerization reaction rate prediction model training is completed, the method further includes:

[0114] Obtaining actual molecular weight distribution data of the multifunctional molecular model;

[0115] Performing reaction rate prediction on the actual molecular weight distribution data based on the polymerization reaction rate prediction model to obtain actual reaction rate data;

[0116] updating the reaction basic conditions based on the actual reaction rate data, and performing a polymerization reaction simulation on the multifunctional molecular model based on the updated reaction basic conditions to obtain second simulated molecular weight distribution data;

[0117] The polymerization reaction rate prediction model is verified based on the actual molecular weight distribution data and the second simulated molecular weight distribution data.

[0118] Actual molecular weight distribution data includes information about each molecule's links to other molecules. This information can be used to calculate the total molecular weight generated, the reaction status of each type of molecule (e.g., the proportion of unreacted functional groups remaining, such as 1, 2, etc., among the total number of participating molecules, i.e., the K value (reaction rate coefficient ratio)), the number-average molecular weight (Mn), and the weight-average molecular weight (Mw). These two parameters are crucial for describing the molecular weight distribution of polymers. The number-average molecular weight (Mn) is often used to describe the average molecular weight of a polymer, while the weight-average molecular weight (Mw) is often used to describe physical properties of a polymer, such as viscosity and mechanical properties, as these properties are generally more strongly influenced by large molecules.

[0119] During the polymerization simulation, the molecular weight of each molecule in the multifunctional molecular model can be recorded to calculate the number average molecular weight and weight average molecular weight.

[0120] The number average molecular weight can be calculated according to the following formula:

[0121]

[0122] in, The molecular weight is The number of molecules.

[0123] The weight average molecular weight can be calculated according to the following formula:

[0124]

[0125] in, The molecular weight is The number of molecules.

[0126] The actual molecular weight distribution data of the multi-functional molecular model can be obtained through experimental means (such as exclusion chromatography); the actual molecular weight distribution data is used as a characteristic value and input into a polymerization reaction rate prediction model to perform reaction rate prediction, and the reaction ratio, reaction degree and reaction rate corresponding to all functional groups corresponding to the actual molecular weight distribution data are obtained; the reaction ratio, reaction degree and reaction rate corresponding to all functional groups, that is, the actual reaction rate data, are updated according to each reaction ratio, reaction degree and reaction rate corresponding to all functional groups in the reaction basic conditions, so that the polymerization reaction simulation of the multi-functional molecular model is performed based on the updated reaction basic conditions to obtain second simulated molecular weight distribution data; a similarity algorithm can be used to calculate the similarity between the second simulated molecular weight distribution data and the actual molecular weight distribution data. If the calculated similarity meets the similarity threshold, it proves that the polymerization reaction rate prediction model predicts accurately. If the calculated similarity does not meet the similarity threshold, the model parameters can be adjusted, the first molecular weight distribution data can be added, the generalization ability of the polymerization reaction rate prediction model can be improved, and more types of polymerization reaction data can be introduced to improve the robustness of the polymerization reaction rate prediction model. After that, after adjusting the parameters and adding data, the polymerization reaction rate prediction model is retrained. Through gradual adjustment and verification, the prediction accuracy of the polymerization reaction rate prediction model can be improved.

[0127] After obtaining the polymerization reaction rate prediction model, the method further includes:

[0128] S104: Obtaining molecular weight distribution data to be predicted.

[0129] The molecular weight distribution data to be predicted refers to the molecular weight distribution data required for reaction rate prediction.

[0130] S105: performing reaction rate prediction on the molecular weight distribution data to be predicted based on the polymerization reaction rate prediction model to obtain reaction rate prediction data.

[0131] Characteristic values ​​can be extracted from the molecular weight distribution data to be predicted to obtain characteristic values, and the reaction rate can be predicted for the characteristic values ​​using a polymerization reaction rate prediction model to obtain the reaction ratio and reaction degree of each functional group corresponding to the molecular weight distribution data to be predicted, as well as the reaction rate corresponding to all functional groups, as reaction rate prediction data.

[0132] For example, it is assumed that the molecular weight distribution data to be predicted includes 5 k values, namely kA0, kA1, kB0, kB1, and kB2, as well as Mw and Mn, wherein kA0 represents the proportion of A molecules with 0 remaining unreacted functional groups in the total number of reacted molecules, kA1 represents the proportion of A molecules with 1 remaining unreacted functional group in the total number of reacted molecules, kB0 represents the proportion of B molecules with 0 remaining unreacted functional groups in the total number of reacted molecules, kB1 represents the proportion of B molecules with 1 remaining unreacted functional group in the total number of reacted molecules, and kB2 represents the proportion of B molecules with 2 remaining unreacted functional groups in the total number of reacted molecules. Therefore, the five k values ​​and Mw, Mn can be extracted from the molecular weight distribution data to be predicted as characteristic values, and the reaction rate is predicted for the characteristic values ​​through a polymerization reaction rate prediction model, so as to obtain the reaction ratio and reaction degree of each functional group corresponding to the molecular weight distribution data to be predicted and the reaction rate corresponding to all functional groups as reaction rate prediction data; through the predicted reaction rate prediction data, a high molecular weight polymer can be synthesized according to the reaction rate prediction data, so that the molecular weight distribution of the synthesized high molecular weight polymer meets expectations, and a high molecular weight polymer with a specific molecular weight and molecular weight distribution can be efficiently synthesized, which not only improves the synthesis efficiency but also reduces the experimental cost and time.

[0133] The polymerization reaction rate prediction model is used to predict the reaction rate coefficient. In industrial production, the reaction rate coefficient can be used to optimize reaction conditions, such as temperature, pressure, concentration, and catalyst selection. In chemical engineering, the reaction rate coefficient is an important basis for reactor design, helping engineers predict reaction behavior within the reactor, select the appropriate reactor type (such as a continuous reactor or a batch reactor), and optimize the reactor's operating conditions. In pharmaceutical synthesis, the reaction rate coefficient can be used to predict the reaction progress and product formation, thereby optimizing the synthesis process.

[0134] In one embodiment, if Figure 6 As shown, a reaction rate prediction method based on machine learning is provided, and the specific process steps are as follows:

[0135] Step 1: Receive input information and initialize the system.

[0136] The specific operations of receiving input information and initializing the system include accepting various types of input information (i.e., basic reaction conditions) to initialize the system. Input information includes relevant data on various molecular species, functional group reaction rules, reaction rate calculation formulas, and the set total number of reactions or reactivity.

[0137] Step 2: Rewrite the Monte Carlo simulation program and add an adjustable reaction rate function.

[0138] The traditional Monte Carlo simulation algorithm for polymerization reactions was rewritten based on the reaction rate calculation formula.

[0139] Step 3: Use the algorithm from Step 2 to simulate a large number of different initial material reaction ratios and reaction degrees for the same polymerization reaction to obtain sufficient data. Each set of reactant ratios and reaction degrees will correspond to the molecular weight distribution of a polymer system after the reaction and the ratio of different monomers. The polymer system can be a type of material system composed of high molecular weight compounds (also called polymers).

[0140] Step 4: Train the preset neural network to obtain the final polymerization reaction rate prediction model.

[0141] The large amount of data in step 3 is used to train a machine learning model (such as a neural network), with molecular weight distribution and reaction degree as input features and reaction rate coefficient ratio as output. A feedforward neural network model is used for training to obtain a model that predicts the reaction rate coefficient ratio based on molecular weight distribution and reaction degree.

[0142] Compared with the traditional Monte Carlo method for simulating the reaction rate of polymer polymerization, which is constant during the reaction process, the polymerization reaction process of this application can more accurately calculate the reaction rate based on the input reaction rate formula. Figure 3 Functional group A and functional group B are shown as a multi-functional molecular model for the reaction experiment, and their reaction rates are used as a factual comparison. The traditional model is predicted based on the Gillespie algorithm, a random simulation algorithm based on the Monte Carlo algorithm; while the present application enables it to automatically adjust the reactivity of the functional group and predict the change in the functional group reaction rate from the experimental data through a machine learning algorithm. It is more suitable for more polymer polymerization simulation scenarios and has distinct flexibility and maneuverability. Refer to Table 3 and Figures 7 and 8 It can be seen that Figure 7 、 8 The comparison shown clearly shows the predicted comparison of one of the basic properties of polymers - structure. The comparison is the ratio of monomers with different numbers of functional groups in each monomer. Figure 3 The multifunctional molecular model shown has two monomers: A has two reactive functional groups, and B has three. KA0 represents the number of A monomers with both functional groups reacting (if 0 is a reactive functional group, there are 0 remaining), KA1 represents the number of A monomers with one reactive functional group, KB0 represents the number of B monomers with three reactive functional groups, and so on. Comparing these numbers yields the proportional data shown in Table 3. Comparison with the final experimental data clearly demonstrates that the polymerization reaction prediction model proposed in this application is more accurate, closer to the actual data, and more reliable in its prediction results than traditional models.

[0143] Table 3

[0144]

[0145] For another basic property of polymers, molecular weight distribution prediction is as follows Figure 9 、 Figure 10 As shown, compared with the traditional model, the data predicted by the polymerization reaction rate prediction model obtained by training the neural network with the first molecular weight distribution data obtained by simulation in this application can be used to predict the molecular weight distribution compared with the traditional Flory-Stockmayer polymer reaction theory, improving its limitations, such as ignoring the formation of ring structures and active assumptions such as functional groups.

[0146] The above is the process of the reaction rate prediction method based on machine learning in this application.

[0147] As described above, the present application provides a method, device, computer equipment and storage medium for predicting reaction rate based on machine learning, by obtaining the reaction basic conditions of a multi-functional molecular model; performing a polymerization reaction simulation on the multi-functional molecular model based on the reaction basic conditions to obtain first simulated molecular weight distribution data; training a preset neural network based on the first simulated molecular weight distribution data to obtain a polymerization reaction rate prediction model; obtaining molecular weight distribution data to be predicted; performing reaction rate prediction on the molecular weight distribution data to be predicted based on the polymerization reaction rate prediction model to obtain reaction rate prediction data. In the method for predicting reaction rate based on machine learning provided in the present application, by performing a polymerization reaction simulation with adjustable reaction rate on the multi-functional molecular model under the constraint of the reaction basic conditions, the molecular weight distribution under different reaction conditions, i.e., the first simulated molecular weight distribution data, can be obtained. Then, the preset neural network is trained using the first simulated molecular weight distribution data, which is conducive to the neural network learning the law of molecular weight change of the multi-functional molecular model during the polymerization reaction from the molecular weight distribution under different reaction conditions, thereby being applicable to more polymer polymerization simulation scenarios, and being conducive to the polymerization reaction rate prediction model providing a predicted value of the reaction rate coefficient ratio of different types of reactive functional groups during the polymerization reaction.

[0148] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0149] In one embodiment, a reaction rate prediction device based on machine learning is provided. The reaction rate prediction device based on machine learning corresponds to the reaction rate prediction method based on machine learning in the above embodiment. Figure 11As shown, the reaction rate prediction device based on machine learning includes:

[0150] The first acquisition module 201 is used to obtain the basic reaction conditions of the multi-functional group molecular model;

[0151] A first simulation module 202 is configured to perform a polymerization reaction simulation on the multifunctional molecular model based on the reaction basic conditions to obtain first simulated molecular weight distribution data;

[0152] A training module 203 is configured to train a preset neural network based on the first simulated molecular weight distribution data to obtain a polymerization reaction rate prediction model;

[0153] The second acquisition module 204 is used to obtain molecular weight distribution data to be predicted;

[0154] The first prediction module 205 is configured to perform reaction rate prediction on the molecular weight distribution data to be predicted based on the polymerization reaction rate prediction model to obtain reaction rate prediction data.

[0155] In the reaction rate prediction method scheme based on machine learning provided in the present application, by obtaining the reaction basic conditions of a multi-functional molecular model; performing a polymerization reaction simulation on the multi-functional molecular model based on the reaction basic conditions to obtain first simulated molecular weight distribution data; training a preset neural network based on the first simulated molecular weight distribution data to obtain a polymerization reaction rate prediction model; obtaining molecular weight distribution data to be predicted; performing reaction rate prediction on the molecular weight distribution data to be predicted based on the polymerization reaction rate prediction model to obtain reaction rate prediction data. In the reaction rate prediction method scheme based on machine learning provided in the present application, by performing a polymerization reaction simulation with adjustable reaction rate on the multi-functional molecular model under the constraint of the reaction basic conditions, the molecular weight distribution under different reaction conditions, i.e., the first simulated molecular weight distribution data, can be obtained. Then, the preset neural network is trained using the first simulated molecular weight distribution data, which is conducive to the neural network learning the law of molecular weight change of the multi-functional molecular model during the polymerization reaction from the molecular weight distribution under different reaction conditions, thereby being applicable to more polymer polymerization simulation scenarios, and being conducive to the polymerization reaction rate prediction model providing a predicted value of the reaction rate coefficient ratio of different types of reactive functional groups during the polymerization reaction.

[0156] Optionally, in some embodiments of the present application, the device further includes:

[0157] A third acquisition module is used to obtain actual molecular weight distribution data of the multi-functional group molecular model;

[0158] A second prediction module is used to perform reaction rate prediction on the actual molecular weight distribution data based on the polymerization reaction rate prediction model to obtain actual reaction rate data;

[0159] a second simulation module, configured to update the reaction basic conditions based on the actual reaction rate data, and perform a polymerization reaction simulation on the multifunctional molecular model based on the updated reaction basic conditions to obtain second simulated molecular weight distribution data;

[0160] A verification module is used to verify the polymerization reaction rate prediction model according to the actual molecular weight distribution data and the second simulated molecular weight distribution data.

[0161] Optionally, in some embodiments of the present application, the basic reaction conditions include scanning rate ratio parameters and reaction association parameters, the scanning rate ratio parameters include the reaction rate of each functional group in the multi-functional molecular model under different conditions, and the reaction association parameters include the concentration of each functional group in the multi-functional molecular model under different conditions.

[0162] Optionally, in some embodiments of the present application, different functional groups have different concentrations.

[0163] Optionally, in some embodiments of the present application, the first simulation module includes:

[0164] A calculation submodule, configured to calculate the reaction rate of each functional group in the multi-functional group molecular model under different conditions using a preset reaction rate calculation formula;

[0165] The simulation submodule is used to use a Monte Carlo algorithm to simulate the polymerization reaction of the reaction rate of each functional group in the multi-functional molecular model under different conditions based on the reaction basic conditions to obtain the first simulated molecular weight distribution data.

[0166] Optionally, in some embodiments of the present application, the simulation submodule includes:

[0167] a determination unit, configured to determine a target reaction based on the reaction rates of the functional groups in the multi-functional group molecular model under different conditions;

[0168] a simulation unit, configured to perform a polymerization reaction simulation on any two unreacted functional groups in the multi-functional molecular model based on the target reaction;

[0169] The obtaining unit is used to return to calculating the reaction rate of any reaction in the multi-functional molecular model and continue to execute if the reaction degree of the polymerization reaction simulation does not reach the prediction threshold, so as to obtain the first simulated molecular weight distribution data.

[0170] Optionally, in some embodiments of the present application, the multi-functional group molecular model includes functional group A and functional group B, and the reaction rate calculation formula is expressed as:

[0171]

[0172] Where R represents the reaction rate, represents the reaction rate coefficient of the i-th case of functional group A, represents the concentration of functional group A in the ith case, is the reaction rate coefficient of the jth case of functional group B, is the concentration of the jth instance of the B functional group, N represents the number of different instances of the A functional group, and M represents the number of different instances of the B functional group.

[0173] In one embodiment, a computer device is provided. The internal structure diagram of the computer device can be as follows: Figure 12 As shown. The computer device includes a processor, memory, network interface, display screen, and input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external server via a network connection. When the computer program is executed by the processor, it implements the functions or steps of a reaction rate prediction method based on machine learning.

[0174] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following steps are implemented:

[0175] Obtain the basic reaction conditions of the multi-functional molecular model; perform polymerization reaction simulation on the multi-functional molecular model based on the basic reaction conditions to obtain first simulated molecular weight distribution data; train a preset neural network based on the first simulated molecular weight distribution data to obtain a polymerization reaction rate prediction model; obtain the molecular weight distribution data to be predicted; perform reaction rate prediction on the molecular weight distribution data to be predicted based on the polymerization reaction rate prediction model to obtain reaction rate prediction data.

[0176] In an embodiment, by performing a polymerization reaction simulation with adjustable reaction rate on a multi-functional molecular model under the constraints of the basic reaction conditions, the molecular weight distribution under different reaction conditions, i.e., the first simulated molecular weight distribution data, can be obtained. Then, the preset neural network is trained using the first simulated molecular weight distribution data, which is beneficial for the neural network to learn the law of molecular weight change of the multi-functional molecular model during the polymerization reaction from the molecular weight distribution under different reaction conditions, thereby being applicable to more polymer polymerization simulation scenarios, and being beneficial for the polymerization reaction rate prediction model to provide a predicted value of the reaction rate coefficient ratio of different types of reaction functional groups during the polymerization reaction.

[0177] In one embodiment, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the following steps are implemented:

[0178] Obtain the basic reaction conditions of the multi-functional molecular model; perform polymerization reaction simulation on the multi-functional molecular model based on the basic reaction conditions to obtain first simulated molecular weight distribution data; train a preset neural network based on the first simulated molecular weight distribution data to obtain a polymerization reaction rate prediction model; obtain the molecular weight distribution data to be predicted; perform reaction rate prediction on the molecular weight distribution data to be predicted based on the polymerization reaction rate prediction model to obtain reaction rate prediction data.

[0179] In an embodiment, by performing a polymerization reaction simulation with adjustable reaction rate on a multi-functional molecular model under the constraints of the basic reaction conditions, the molecular weight distribution under different reaction conditions, i.e., the first simulated molecular weight distribution data, can be obtained. Then, the preset neural network is trained using the first simulated molecular weight distribution data, which is beneficial for the neural network to learn the law of molecular weight change of the multi-functional molecular model during the polymerization reaction from the molecular weight distribution under different reaction conditions, thereby being applicable to more polymer polymerization simulation scenarios, and being beneficial for the polymerization reaction rate prediction model to provide a predicted value of the reaction rate coefficient ratio of different types of reaction functional groups during the polymerization reaction.

[0180] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or computer device can be found in the relevant descriptions of the server side and the client side in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0181] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0182] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0183] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A reaction rate prediction method based on machine learning, characterized in that: The method comprises: Obtain the basic reaction conditions of multi-functional molecular models; Performing a polymerization reaction simulation on the multifunctional molecular model based on the reaction basic conditions to obtain first simulated molecular weight distribution data; Training a preset neural network based on the first simulated molecular weight distribution data to obtain a polymerization reaction rate prediction model; Obtaining molecular weight distribution data to be predicted; Performing reaction rate prediction on the molecular weight distribution data to be predicted based on the polymerization reaction rate prediction model to obtain reaction rate prediction data; The step of performing a polymerization reaction simulation on the multifunctional molecular model based on the reaction basic conditions to obtain first simulated molecular weight distribution data includes: Calculate the reaction rate of each functional group in the multi-functional group molecular model under different conditions using a preset reaction rate calculation formula; Using a Monte Carlo algorithm, based on the reaction basic conditions, a polymerization reaction simulation is performed on the reaction rate of each functional group in the multi-functional molecular model under different conditions to obtain the first simulated molecular weight distribution data; The multifunctional molecular model includes functional group A and functional group B, and the reaction rate calculation formula is expressed as: Where R represents the reaction rate, represents the reaction rate coefficient of the i-th case of functional group A, represents the concentration of functional group A in the ith case, is the reaction rate coefficient of the jth case of functional group B, is the concentration of the jth case of functional group B, N represents the number of different cases of functional group A, and M represents the number of different cases of functional group B; During the polymerization reaction simulation process, when the i-th case and the j-th case change, the reaction rate coefficient of the i-th case of the functional group A and the reaction rate coefficient of the j-th case of the functional group B also change, and the reaction rates of the functional group A and the functional group B are recalculated according to the reaction rate constant calculation formula.

2. The reaction rate prediction method based on machine learning according to claim 1, characterized in that The method further comprises: Obtaining actual molecular weight distribution data of the multifunctional molecular model; Performing reaction rate prediction on the actual molecular weight distribution data based on the polymerization reaction rate prediction model to obtain actual reaction rate data; updating the reaction basic conditions based on the actual reaction rate data, and performing a polymerization reaction simulation on the multifunctional molecular model based on the updated reaction basic conditions to obtain second simulated molecular weight distribution data; The polymerization reaction rate prediction model is verified based on the actual molecular weight distribution data and the second simulated molecular weight distribution data.

3. The reaction rate prediction method based on machine learning according to claim 1, characterized in that The basic reaction conditions include a scan rate ratio parameter and a reaction association parameter. The scan rate ratio parameter includes the reaction rate of each functional group in the multi-functional molecular model under different conditions. The reaction association parameter includes the concentration of each functional group in the multi-functional molecular model under different conditions.

4. The reaction rate prediction method based on machine learning according to claim 3, characterized in that Different functional groups have different concentrations.

5. The reaction rate prediction method based on machine learning according to claim 1, characterized in that The method adopts a Monte Carlo algorithm to simulate the polymerization reaction of the reaction rate of each functional group in the multi-functional molecular model under different conditions based on the reaction basic conditions to obtain the first simulated molecular weight distribution data, including: Determining a target reaction based on the reaction rates of each functional group in the multifunctional molecular model under different circumstances; Performing polymerization reaction simulation on any two unreacted functional groups in the multi-functional molecular model based on the target reaction; If the reaction degree of the polymerization reaction simulation does not reach the prediction threshold, the calculation returns to the reaction rate of any reaction in the multifunctional molecular model and continues until the reaction degree of the polymerization reaction simulation reaches the prediction threshold to obtain the first simulated molecular weight distribution data.

6. A reaction rate prediction device based on machine learning, characterized in that include: The first acquisition module is used to obtain the reaction basic conditions of the multi-functional group molecular model; A first simulation module is configured to perform a polymerization reaction simulation on the multifunctional molecular model based on the reaction basic conditions to obtain first simulated molecular weight distribution data; the polymerization reaction simulation on the multifunctional molecular model based on the reaction basic conditions to obtain first simulated molecular weight distribution data comprises: Calculate the reaction rate of each functional group in the multi-functional group molecular model under different conditions using a preset reaction rate calculation formula; Using a Monte Carlo algorithm, based on the reaction basic conditions, a polymerization reaction simulation is performed on the reaction rate of each functional group in the multi-functional molecular model under different conditions to obtain the first simulated molecular weight distribution data; The multifunctional molecular model includes functional group A and functional group B, and the reaction rate calculation formula is expressed as: Where R represents the reaction rate, represents the reaction rate coefficient of the i-th case of functional group A, represents the concentration of functional group A in the ith case, is the reaction rate coefficient of the jth case of functional group B, is the concentration of the jth case of functional group B, N represents the number of different cases of functional group A, and M represents the number of different cases of functional group B; During the polymerization reaction simulation, when the i-th case and the j-th case change, the reaction rate coefficient of the i-th case of the functional group A and the reaction rate coefficient of the j-th case of the functional group B also change, and the reaction rates of the functional group A and the functional group B are recalculated according to the reaction rate constant calculation formula; A training module, configured to train a preset neural network based on the first simulated molecular weight distribution data to obtain a polymerization reaction rate prediction model; A second acquisition module is used to obtain molecular weight distribution data to be predicted; The first prediction module is used to perform reaction rate prediction on the molecular weight distribution data to be predicted based on the polymerization reaction rate prediction model to obtain reaction rate prediction data.

7. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the reaction rate prediction method based on machine learning are implemented as described in any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the reaction rate prediction method based on machine learning are implemented.

Images