An Adaptive Method for Adjusting Kinetic Parameters of Aggregation Process

Abstract

The present invention discloses an adaptive method for adjusting kinetic parameters of a polymerization process, belonging to the chemical engineering field. The method includes: establishing a microscopic mass model of the polymerization reaction; screening actually estimable kinetic parameters and performing preliminary fitting and solution; classifying the screened parameters according to their influence on the molecular weight distribution and sorting them according to sensitivity; constructing an optimization objective function, and using a particle swarm algorithm to optimize the parameters based on the preliminary optimal values, where different categories of parameters adopt different optimization algorithms. The purpose of the present invention is to adaptively adjust parameters based on small-sample industrial data to replace the complicated model parameter tuning. This method can automatically adjust parameters based on the difference between the calculated molecular weight distribution curve and the on-site industrial data, so that the microscopic mass results under several working conditions are consistent with the on-site data.

Description

Technical Field

[0001] The present invention relates to the chemical industry field, and particularly to an adaptive method for adjusting kinetic parameters in the polymerization process. Background Art

[0002] Polymers are composed of a series of repeating structural units, and their microscopic mass characteristics (such as molecular weight distribution) have an important impact on the intrinsic properties of polymers. However, in the actual production process, the polymerization reaction is usually controlled and guided by macroscopic mass indicators (for example, average molecular weight). However, this method fails to fully reflect the detailed information of the chain structure and is difficult to meet the demand for refined control of high-performance polymer products. In order to accurately obtain the macroscopic and microscopic mass characteristics of polymers, the simulation of the polymerization process has become a key technology. Especially in large-scale industrial production, simulation technology plays a crucial role in product design, process optimization, and operation adjustment. However, the polymerization reaction process model is large in scale and has extremely many parameters, and it is difficult to solve the parameter estimation model. In addition, it is generally difficult to obtain industrial data in large batches, which further increases the difficulty of parameter estimation. Therefore, developing a small-sample and easily solvable parameter adjustment method has become the key to solving this problem. Summary of the Invention

[0003] To solve the problems in the prior art, the present invention proposes an adaptive method for adjusting kinetic parameters in the polymerization process. This method classifies and grades the kinetic parameters and adaptively adjusts the parameters by constructing a weighting function. This method includes kinetic modeling, constructing a correlation function between kinetics and molecular weight distribution, and a two-layer iterative optimization algorithm.

[0004] The present invention is realized by the following technical solutions:

[0005] On the one hand, the present invention proposes an adaptive method for adjusting kinetic parameters in the polymerization process, including:

[0006] 1) Establish a microscopic mass model of the polymerization reaction process;

[0007] 2) Screen the actually estimable kinetic parameters in the polymerization reaction process, and fit the microscopic mass model to solve the initial optimal values of the screened kinetic parameters;

[0008] 3) Classify the screened kinetic parameters according to their influence on the polymer molecular weight distribution, and sort the kinetic parameters in each category from high to low according to sensitivity;

[0009] 4) Construct an optimization objective function, and based on the initial optimal values, perform particle swarm optimization on each kinetic parameter based on the optimization objective function. Different categories of parameters adopt different optimization objective functions.

[0010] On the other hand, the present invention also provides an adaptive polymerization process kinetic parameter adjustment system for implementing the above-mentioned polymerization process kinetic parameter adjustment method.

[0011] The beneficial effects of the present invention are as follows:

[0012] (1) The adaptive kinetic parameter adjustment method proposed by the present invention can perform parameter adjustment and estimation with a relatively small industrial sample data scale. Compared with conventional kinetic parameter estimation methods, the model scale is smaller, which is more conducive to solving.

[0013] (2) The adaptive kinetic parameter adjustment method proposed by the present invention aims to replace the cumbersome manual parameter adjustment, making it more efficient and reliable.

[0014] (3) The adaptive kinetic parameter adjustment method proposed by the present invention is simple to implement, has good accuracy, and is applicable to the simulation of various olefin polymerization reaction processes. Description of the Drawings

[0015] By describing the embodiments of the present invention in combination with the following drawings and attached tables and demonstrating its effects, the purpose, features, and advantages of the present invention can be further understood, where:

[0016] Figure 1 It is a schematic diagram of the adaptive parameter adjustment method;

[0017] Figure 2 It is a comparison between the adaptive parameter adjustment result of a certain industrial grade (Grade 1) and the molecular weight distribution data of one reactor on-site;

[0018] Figure 3 It is a comparison between the adaptive parameter adjustment result of a certain industrial grade (Grade 1) and the molecular weight distribution data of two reactors on-site;

[0019] Figure 4 It is a comparison between the adaptive parameter adjustment result of a certain industrial grade (Grade 1) and the molecular weight distribution data of three reactors on-site.

[0020] Figure 5 It is a comparison between the adaptive parameter adjustment result of a certain industrial grade (Grade 2) and the molecular weight distribution data of one reactor on-site;

[0021] Figure 6 It is a comparison between the adaptive parameter adjustment result of a certain industrial grade (Grade 2) and the molecular weight distribution data of two reactors on-site;

[0022] Figure 7 It is a comparison between the adaptive parameter adjustment result of a certain industrial grade (Grade 2) and the molecular weight distribution data of three reactors on-site. Detailed Embodiments

[0023] The present invention will be described in more detail with reference to the attached drawings of the present invention, so as to describe the specific implementation process of the present invention to those skilled in the art.

[0024] Example 1

[0025] Next, a method for adjusting the kinetic parameters of an adaptive polymerization process provided in this example will be specifically described.

[0026] A method for adjusting the kinetic parameters of an adaptive polymerization process provided in this example is as Figure 1 shown, and includes the following steps:

[0027] 1) Establish a microscopic mass model of the polymerization reaction process;

[0028] 2) Conduct a feasibility estimation of the kinetic parameters;

[0029] 3) Classify and grade the kinetic parameters, and calibrate the sensitivity of their influence on the model;

[0030] 4) Construct an adjustment parameter function that correlates the kinetic parameters with the molecular weight distribution and perform adaptive parameter adjustment.

[0031] Step 1). Construction of the microscopic mass kinetic model

[0032] In this step, an improved microscopic mass kinetic model for the copolymerization reaction of ethylene and butene is established. The model is based on the following assumptions:

[0033] (1) Considering the activation reaction of the cocatalyst with the monomer and comonomer, it is simplified to a linear relationship. It is assumed that the activation rate of the catalyst is positively correlated with the catalyst concentration and temperature;

[0034] (2) It is assumed that there is a non-linear relationship between the rate of the chain growth reaction and the monomer concentration, comonomer concentration, and catalyst activity;

[0035] (3) The chain transfer reaction includes the chain transfer of hydrogen, monomer, and comonomer, and the deactivation of the catalyst is modeled through the interaction between the catalyst and the comonomer.

[0036] On this basis, a general reactor model is established, which involves aspects such as reaction kinetics, thermodynamics, material balance, energy balance, and phase balance, and can simulate the three-phase reaction (gas-liquid-solid) inside the reactor. Its reaction equation is shown in Table 1.

[0037] Table 1 Copolymerization reaction mechanism

[0038] Reaction type Reaction description Catalyst activation #timg##timg# Chain initiation #timg##timg# Chain growth #timg##timg##timg##timg# Transfer to monomer #timg##timg##timg##timg# Transfer to hydrogen #timg##timg# Transfer to cocatalyst #timg##timg# β transfer #timg##timg# Chain deactivation #timg##timg# #timg#

[0039] Where is the potential active site of a certain catalyst. is a cocatalyst, is a monomer, is a comonomer, is a transfer agent, is an active site, is an inactivated active site, is a living polymer with monomer as the active radical chain length of ; is a living polymer with monomer as the active radical chain length of ; is a dead polymer with a chain length of . represents the index of active sites, ranging from 1 to 5. is the kinetic rate constant of different reactions, where is the reaction rate of cocatalyst activation; is the reaction rate of comonomer activation; is the chain initiation rate of the active site and monomer A; is the chain initiation rate of the active site and monomer B; is the chain growth rate, representing the growth reaction rate between the living polymer with monomer A or B as the activity and monomer A or B; is the chain transfer rate, representing the reaction rate of the living polymer with monomer A or B as the activity transferring to monomer A or B; and represent the reaction rate of the living polymer with monomer A / B as the activity transferring to the transfer agent; and represent the reaction rate of the living polymer with monomer A / B as the activity transferring to the cocatalyst; and represent the rate of self-transfer ( transfer) of the living polymer with monomer A / B as the activity; is the reaction rate of the inactivation of the active site and the living polymer with monomer A / B as the activity.

[0040] Step 2). Adaptive estimation and screening of kinetic parameters

[0041] The kinetic mechanism of the polymerization reaction is the core component of the polymerization process model. It reveals the basic characteristics of the process and directly affects various performance indicators of the final polymer product (such as melt index, polydispersity index, average molecular weight, and molecular weight distribution, etc.).

[0042] In this step, by analyzing the sensitivity of reaction kinetic parameters and studying the confidence intervals, it is determined which parameters can be effectively adjusted adaptively. The specific steps of the screening process for parameter estimability analysis are as follows: First, pre-select the parameters. In this stage, those parameters with relatively small sensitivities are removed from the subset of parameters to be estimated; then, perform parameter estimation to obtain the optimal estimated values of the parameters within the current subset; finally, conduct post-optimization analysis. In this stage, for those parameters with relatively large confidence intervals of the optimal values, they are also removed from the subset of parameters to be estimated. By continuously looping through the above steps, the parameters with lower credibility are gradually removed until the maximum confidence interval ratio of the remaining parameters in the subset of parameters to be estimated is lower than the set threshold. At this time, the obtained parameter set is the actual estimable parameter set.

[0043] This step is specifically implemented through the following sub-steps:

[0044] Parameter screening based on sensitivity analysis: Use the microscopic mass model to perform sensitivity analysis on all possible kinetic parameters (such as the chain growth rate constant , the chain termination rate constant , etc.). Through numerical simulation using the model in step 1, calculate the sensitivities of each kinetic parameter, and the sensitivities are used to describe the magnitude of the influence of the parameters on the polymer molecular weight distribution. Screen the kinetic parameters according to the sensitivities, retain the parameters with greater influence on the polymer molecular weight distribution, such as catalyst concentration, reaction temperature, hydrogen concentration, etc., and remove the parameters with smaller sensitivities to ensure that the computing resources are concentrated on the key parameters. The sensitivity is calculated according to the following formula:

[0045]

[0046] where, the is the sensitivity of the th kinetic parameter corresponding to the th active site, is the probability density parameter of the polymer molecular weight distribution at the th active site, represents the partial derivative, and ln represents the logarithm. The sensitivity on the logarithmic scale is calculated in the formula, which can make the comparison of kinetic parameters on different scales more objective.

[0047] Suppose the sensitivity of a certain parameter is the largest among all parameters, and its value is . In this embodiment, retain the parameters whose sensitivities are not less than .

[0048] Parameter Screening Based on Estimability Analysis: To improve the reliability of parameter estimation, the present invention further determines which kinetic parameters can be adaptively adjusted by gradually analyzing the retained reaction kinetic parameters and applying error analysis and confidence interval analysis. In this process, the Bayesian estimation method is used to evaluate the uncertainty of the parameters, and the untrustworthy parameters (confidence interval ratio exceeding 20%) are excluded. The definition of the confidence interval ratio is:

[0049]

[0050] where represents the optimal estimated value of the parameter, represent the upper and lower limits of the confidence interval of the parameter, respectively.

[0051] Preliminary Parameter Estimation: For the set of actually estimable parameters obtained after two screenings, the least squares method (LSM) is used to fit the microscopic mass model to obtain the preliminary optimal kinetic parameter values as the initial parameter values for subsequent optimization.

[0052] Step 3). Classification and Sensitivity Ranking of Kinetic Parameters

[0053] To improve the accuracy of adaptive estimation, the present invention proposes a multi-step adaptive parameter estimation method. By classifying and grading the kinetic parameters, the complex non-linear ill-posed problem is successfully transformed into multiple simple sub-problems with low non-linearity, thus effectively avoiding the problem of initial value sensitivity and reducing the solution difficulty. This method not only improves the accuracy of parameter estimation but also can better adapt to the complex situations in the polymerization process at industrial scale.

[0054] Specifically, the kinetic parameters screened in step 2) are classified according to their influence on the molecular weight distribution (MWD) and graded according to sensitivity.

[0055] In this embodiment, the parameters are divided into four categories:

[0056] : Positively correlated with the molecular weight, affecting the position of the molecular weight distribution;

[0057] : Negatively correlated with the molecular weight, affecting the position of the molecular weight distribution;

[0058] : Positively correlated with the molecular weight, affecting the shape of the molecular weight distribution;

[0059] : Negatively correlated with the molecular weight, affecting the shape of the molecular weight distribution.

[0060] where and The parameters of this type play a decisive role in the average molecular weight of the polymer, usually including catalyst concentration, reaction temperature, etc.; and The parameters of this type mainly affect the width of the molecular weight distribution or the sharpness of the distribution, usually including hydrogen concentration, comonomer ratio, etc.

[0061] After classification, the parameters in each category are sorted from high to low according to sensitivity, that is is the relatively most sensitive parameter, while is the relatively least sensitive parameter.

[0062] Step 4). Adaptive parameter adjustment

[0063] One of the core innovations of the present invention is the adaptive parameter adjustment method, which dynamically adjusts the key parameters in the polymerization reaction process based on the above kinetic model and correlation function to optimize the target molecular weight distribution. In this step, an optimization objective function is first constructed, and then the particle swarm optimization algorithm is used to optimize each kinetic parameter, and different types of parameters are designed with different optimization algorithms. The specific process is as follows:

[0064] Construction of the optimization objective function: For and type parameters, by calculating the difference between the molecular weight distribution of the initial parameter values and the target molecular weight distribution, an optimization objective function is established:

[0065]

[0066] Among them, is the set of kinetic parameters to be optimized, is the current molecular weight distribution, is the target molecular weight distribution, represents the square of the two-norm.

[0067] For and type parameters, by calculating the standard deviation of the molecular weight distribution and the target molecular weight distribution, an optimization objective function is established:

[0068]

[0069] Where and are the standard deviations of calculating the molecular weight distribution and the target molecular weight respectively.

[0070] In the objective function, the molecular weight distribution Obtained by simulating through a microscopic quality model based on the parameters to be estimated; based on the classified parameter set, the functional correlation between the molecular weight distribution and the parameters to be estimated is as follows:

[0071]

[0072]

[0073]

[0074] where is the chain length, is the molecular weight distribution curve calculated based on the adaptive estimation algorithm, is the molecular weight distribution sub-curve at the active site , is the nominal molecular weight distribution curve at the active site ; , represent the kinetic parameters affecting the position of the molecular weight distribution, , represent the kinetic parameters affecting the size of the molecular weight distribution, , , , are the indices (consecutive integers starting from zero) of various parameter sets, is the probability density parameter of the molecular weight distribution, , , , are respectively , , and adaptive estimation weighting functions, and the definition of the adaptive estimation weighting function is as follows:

[0075]

[0076]

[0077]

[0078]

[0079] where and are the coefficients controlling the weight decay rate ( takes 0.5 - 2, takes 0.3 - 1.0).

[0080] The molecular weight distribution curve consists of multiple active sites Sub-curves on are composed of, and each sub-curve is a nominal curve obtained through parameter transformation.

[0081] Design of the optimization algorithm: Based on the above optimization objective function, the Particle Swarm Optimization (PSO) algorithm is used to solve the optimization problem of the actually estimable parameter set. In the PSO algorithm, each parameter in the actually estimable parameter set is taken as a particle, the kinetic parameters to be optimized are taken as the positions of the particles, and the amplitude and direction of parameter update are taken as the velocities of the particles. The update formula of the particle swarm is as follows:

[0082]

[0083] where is the velocity of the particle at time t and t + 1, is the position of the particle at time t, is the historical optimal position of the particle, is the global optimal position, is the inertia weight, , are the acceleration constants, , are random numbers.

[0084] For and types of parameters, the objective function of each particle is to minimize the position error, that is, to minimize the difference between the molecular weight distribution and the target distribution; for and types of parameters, the objective function is to minimize the shape error of the molecular weight distribution, that is , where and are the standard deviations of the calculated curve and the target curve respectively.

[0085] Parameter update and feedback adjustment: In each optimization iteration of the PSO algorithm, according to the particle swarm optimization result, the parameter values of the microscopic mass kinetic model in the initial stage of the reaction are adjusted, and the polymerization process is simulated through the microscopic mass kinetic model. Through real-time feedback, the optimization objective is continuously reduced until the error between the molecular weight distribution and the target value is less than the predetermined threshold.

[0086] This method can adapt to the changes of reaction conditions in the reactor in real time. After each optimization iteration, the attenuation coefficients , , , in the weight functions and , thus ensuring the stability of the polymerization process and the quality of the final polymer.

[0087] Example 2

[0088] In this example, a three-reactor series system for producing polyethylene slurry is involved. The system includes three identical continuous stirred tank reactors (CSTRs). In this system, the three reactors are connected in sequence, namely the first reactor, the second reactor, and the third reactor. The liquid-phase material of the previous reactor is transferred to the next reactor or output as a product after flash evaporation and condensation treatment. In this example, n-hexane (C6H14) is used as the solvent, and gaseous raw materials such as ethylene, butene, and hydrogen are used, and a Ziegler-Natta catalyst is used for the olefin copolymerization process.

[0089] This technology covers the basic reaction types involved in olefin polymerization using Ziegler-Natta catalysts, including catalyst activation, chain initiation, polymer chain growth, chain transfer, and catalyst deactivation. Considering the complexity of the Ziegler-Natta copolymerization reaction mechanism and the strong coupling and nonlinear characteristics of its reaction process, these factors often make the calculation process difficult to converge. In the simplified model proposed in the present invention, the catalyst activation reaction only involves the activation of cocatalyst, monomer, and comonomer; the chain growth reaction only considers the chain extension of monomer and comonomer; the chain transfer reaction includes chain transfer to monomer, hydrogen, cocatalyst, and self-chain transfer.

[0090] This simplified model not only ensures the high accuracy of the model but also significantly reduces the complexity of modeling. In the Ziegler-Natta catalyst system, ethylene and comonomers such as butene react in the polymerization reactor to produce copolymers with different chain lengths and compositions. Establishing a reactor model based on an accurate reaction mechanism is crucial for determining the composition and molecular weight distribution of the copolymer. This reactor model includes comprehensive considerations of polymerization reaction kinetics, thermodynamics, material balance, energy balance, and phase balance. Considering the particularity of the copolymer system, separate mass balances for different chain lengths are also required. In the modeling process, moment models are often used to represent the relationship between different chain lengths to simplify the complexity of the model.

[0091] This algorithm is completed based on the model in the Python environment. First, a copolymerization process model of ethylene and butene is established, and its equation is shown as follows.

[0092] Polymer chain growth equation:

[0093]

[0094]

[0095] Where and Represents the generation rate of vinyl polymer chains and butenyl polymer chains with a chain length of , and represents the concentration of vinyl polymer chains with a chain length of -1 and ; and represents the concentration of butenyl polymer chains with a chain length of -1 and ;

[0096] Dead polymer formation equation:

[0097]

[0098] where represents the generation rate of dead polymer chains with a length of , and represent the chain transfer rate constants of ethylene chains and butene chains, respectively;

[0099] Catalyst consumption equation:

[0100]

[0101] where is the consumption rate of the catalyst, is the activation rate constant of the catalyst and the cocatalyst, is the activation rate constant of the catalyst and butene monomer, is the cocatalyst concentration, is the catalyst concentration;

[0102] Cocatalyst consumption equation:

[0103]

[0104] where is the consumption rate of the cocatalyst, is the chain transfer rate constant of the cocatalyst and the active center, , are the concentrations of vinyl active centers and butenyl active centers;

[0105] Active chain initiation equation:

[0106]

[0107] where is the generation rate of the initial active chains, , are the initiation rate constants of the monomers, is the initial active chain concentration, , , , are the chain transfer rate constants, is the hydrogen concentration, is the active chain deactivation rate constant;

[0108] Dead catalyst formation equation:

[0109]

[0110] where, is the formation rate of the dead catalyst, , are the deactivation rate constants of the vinyl and butenyl active centers;

[0111] Monomer consumption equation:

[0112]

[0113]

[0114] where, is the consumption rate of ethylene monomer, is the consumption rate of butene monomer, , , , are the chain transfer rate constants of the monomer and the active center;

[0115] Hydrogen consumption equation:

[0116]

[0117] where, is the consumption rate of hydrogen, is the zero-order moment of the living polymer with monomer A at the active site, is the zero-order moment of the living polymer with monomer B at the active site.

[0118] After that, adaptive estimation and screening of kinetic parameters are carried out. The screened kinetic parameters are classified and sorted from high to low sensitivity as follows:

[0119] (positively correlated with molecular weight, affecting position): None.

[0120] (negatively correlated with molecular weight, affecting position): , , , , , 。

[0121] (Positively correlated with molecular weight, affecting the shape): , , , 。

[0122] (Negatively correlated with molecular weight, affecting the shape): , , 。

[0123] Subsequently, adaptive parameter adjustment is performed. In the parameter optimization process of parameter adjustment, the particle swarm optimization of different types of parameters is as follows: For and type parameters, that is, negatively correlated parameters related to the position of molecular weight distribution:

[0124] Particle swarm initialization and objective function: Initialize the position and velocity of the particles. The objective function of each particle is to minimize the position error, that is, to minimize the difference between the molecular weight distribution and the target distribution.

[0125] Particle swarm iterative search: Update the particle position and velocity through multiple iterations to search for the global optimal solution, and finally obtain the optimized type parameters.

[0126] For and type parameters (positively and negatively correlated with the shape of molecular weight distribution respectively):

[0127] Particle swarm initialization and objective function: Initialize the position and velocity of the particles. The objective function is to minimize the shape error of the molecular weight distribution.

[0128] Particle swarm iterative search: Similarly, through multiple iterations, update the particle position and velocity, search for the global optimal solution, and finally obtain the optimized and type parameters.

[0129] After completing the optimization of X and Y type parameters, update the optimized parameters into the reaction kinetics model. Based on the new optimized parameters, re-solve the model to verify whether the optimized molecular weight distribution meets the expectations. Compare the target molecular weight distribution with the actually calculated molecular weight distribution. If the error meets the set threshold, the optimization process ends and the final optimization result is output. If the error is greater than or equal to the set threshold, return to the parameter adjustment module to re-adjust the parameters and continue the optimization until the error is less than the predetermined threshold.

[0130] Before meeting the target error threshold, this method adopts a multi-step optimization method to further improve the accuracy of the model by gradually fine-tuning the parameters. During the optimization process, according to the sensitivity analysis results, the weights of each objective function are dynamically adjusted to ensure that the final optimization result can minimize the error. According to this method, the adaptive parameter adjustment is carried out on the data of a certain industrial process, and the results are as Figures 2 - 7 shown. Figure 2 , Figure 5 are respectively the comparison between the adaptive tuning results of two different ethylene-butene copolymers (with different industrial grades) and the molecular weight distribution data of one kettle on site; Figure 3 , Figure 6 are respectively the comparison between the adaptive tuning results of two different ethylene-butene copolymers and the molecular weight distribution data of two kettles on site; Figure 4 , Figure 7 are respectively the comparison between the adaptive tuning results of two different ethylene-butene copolymers and the molecular weight distribution data of three kettles on site. Based on this solution, the parameter adjustment based on multiple groups of industrial data is realized, and the molecular weight distribution of the model is consistent with the industrial test data.

[0131] Example 3

[0132] In this embodiment, an adaptive kinetic parameter adjustment system for the polymerization process is also provided, and this system is used to implement the above embodiment. The terms "module", "unit", etc. used below can be a combination of software and / or hardware that can realize a predetermined function. Although the system described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware is also possible.

[0133] The adaptive kinetic parameter adjustment system for the polymerization process includes:

[0134] A model construction module for establishing a microscopic mass model of the polymerization reaction process;

[0135] A parameter screening module for screening the actually estimable kinetic parameters in the polymerization reaction process;

[0136] An initial value solving module for solving the preliminary optimal values of the screened kinetic parameters;

[0137] A classification and sorting module for classifying and sorting the screened kinetic parameters;

[0138] A parameter adjustment module for constructing an optimization objective function and optimizing each kinetic parameter based on the optimization objective function.

[0139] For the system embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the partial descriptions of the method embodiments, and the implementation methods of the remaining modules will not be elaborated here. The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of the present invention. A person of ordinary skill in the art can understand and implement it without creative work.

[0140] The embodiments of the system of the present invention can be applied to any device with data processing capabilities, and the device with data processing capabilities can be a device or apparatus such as a computer. The system embodiments can be implemented by software, or by hardware or a combination of software and hardware. Taking software implementation as an example, as a logically meaningful device, it is formed by the processor of any device with data processing capabilities reading the corresponding computer program instructions in the non-volatile memory into the memory for operation.

[0141] The above-described embodiments only represent several implementation manners of the present invention. The descriptions are relatively specific and detailed, but should not be construed as limiting the scope of the patent of the present invention. For those of ordinary skill in the art, without departing from the concept of the present invention, several variations and improvements can still be made, and these all belong to the protection scope of the present invention.

Claims

1. An adaptive method for adjusting kinetic parameters of an aggregation process, characterized in that, include: 1) Establish a microscopic quality model of the polymerization process; 2) Screen the kinetic parameters that can be actually estimated during the polymerization reaction, and fit the micro-mass model to solve the preliminary optimal values ​​of the kinetic parameters after screening; 3) Classify the kinetic parameters obtained by the screening according to their effects on the molecular weight distribution of the polymer, and sort the kinetic parameters in each category from high to low in terms of sensitivity; 4) Construct an optimization objective function. On the basis of the preliminary optimal value, perform particle swarm optimization on the sorted dynamic parameters based on the optimization objective function. Different optimization objective functions are used for different categories of parameters. In step 2), the kinetic parameters that can be actually estimated during the screening polymerization reaction are specifically: 2.1) Numerical simulation based on the micro-mass model is performed to calculate the sensitivity of all kinetic parameters that affect the molecular weight distribution of the polymer and eliminate the kinetic parameters with less sensitivity; The sensitivity calculation formula is: ； Among them, the is the th sensitivity of the th kinetic parameter corresponding to the th active site, is the probability density parameter of the polymer molecular weight distribution on the th active site, represents the partial derivative, and ln represents the logarithm; 2.2) Using the Bayesian estimation method to perform uncertainty assessment on the retained kinetic parameters, obtain the confidence intervals of the kinetic parameters, and eliminate the kinetic parameters whose confidence interval ratio exceeds a threshold. The confidence interval ratio is the ratio of the width of the confidence interval of the kinetic parameter to the optimal estimate of the kinetic parameter, and the optimal estimate of the kinetic parameter is obtained by maximum a posteriori estimation.

2. The adaptive method for adjusting kinetic parameters of an aggregation process according to claim 1, wherein In step 2), the actually estimable kinetic parameter is selected from the group consisting of a chain growth rate constant, a chain termination rate constant, a chain transfer rate constant, an activation rate constant and a deactivation rate constant.

3. The adaptive method for adjusting the kinetic parameters of the aggregation process according to claim 1, characterized in that, In step 1), the microscopic mass model of the polymerization reaction process includes a plurality of equations for solving the generation or consumption rate of living polymers, dead polymers, catalysts, and reactant monomers according to the concentration of substances.

4. The adaptive aggregation process kinetic parameter adjustment method according to claim 1, wherein In step 2.1), the kinetic parameters with a sensitivity lower than 1% of the maximum sensitivity value were eliminated.

5. The adaptive method for adjusting kinetic parameters of an aggregation process according to claim 1, characterized in that, In step 2), the least square method is used to fit the micro-quality model.

6. The adaptive method for adjusting kinetic parameters of an aggregation process according to claim 1, characterized in that In step 3), the kinetic parameters obtained by screening are classified according to their effects on the molecular weight distribution of the polymer, specifically into four categories: The kinetic parameters that are positively correlated with the molecular weight and affect the positions of the molecular weight distribution are classes; The kinetic parameters related to the molecular weight negatively and affecting the positions of the molecular weight distribution are types; The kinetic parameters that are positively correlated with the molecular weight and affect the shape of the molecular weight distribution are types; The kinetic parameters that are negatively correlated with the molecular weight and affect the shape of the molecular weight distribution are types.

7. The adaptive method for adjusting the kinetic parameters of the aggregation process according to claim 6, characterized in that, In step 4), for and the kinetic parameters of the class, the optimization objective function is: ； Among them, is the set of kinetic parameters screened in step 2), is the estimated molecular weight distribution curve, is the target molecular weight distribution curve, represents the square of the two-norm; For and the kinetic parameters of the class, the optimization objective function is: ； Wherein, and are respectively the estimated molecular weight distribution curve and the target molecular weight standard deviation; Estimated molecular weight distribution curve The calculation formula is as follows: ； ； ； Among them, is the chain length, is the estimated molecular weight distribution curve, is the molecular weight distribution sub-curve on the active site , is the nominal molecular weight distribution curve on the active site , , , and represent class, class, class and class kinetic parameters, , , , are the indexes of the kinetic parameters in each category, is the probability density parameter of the molecular weight distribution on the th active site, , , , are respectively , , and adaptive estimation weighting functions.

8. The adaptive method for adjusting kinetic parameters of an aggregation process according to claim 1, characterized in that, In step 4), the particle swarm optimization is performed on each kinetic parameter based on the optimization objective function, specifically: each kinetic parameter in the actual estimable parameter set is taken as a particle, the kinetic parameter value to be optimized is taken as the particle position, the amplitude and direction of the kinetic parameter update is taken as the particle speed, and the particle swarm is iteratively updated based on the optimization objective function.

9. An adaptive polymerization process kinetic parameter regulation system for implementing the polymerization process kinetic parameter regulation method according to claim 1, characterized in that The system includes: Model building module, used to build microscopic quality models of polymerization processes; Parameter screening module, used to screen the kinetic parameters that can actually be estimated during the polymerization reaction; Initial value solving module, used to solve the initial optimal value of the kinetic parameters after screening; A classification and sorting module is used to classify and sort the kinetic parameters obtained by screening; The parameter adjustment module is used to construct an optimization objective function and optimize various kinetic parameters based on the optimization objective function.

Images