A method for estimating kinetic parameters of olefin polymerization reaction

Abstract

The present invention discloses a method for estimating kinetic parameters of olefin polymerization reaction, including: E1 determining the kinetic parameters of single-site reaction; E2 determining the number of active sites and the kinetic parameters of multi-site reaction; E3 outputting the kinetic parameters corresponding to all active sites. E1 includes: selecting a suitable polymerization reaction mechanism and research method; obtaining experimental conditions and initial values of pre-exponential factor, activation energy, and reference temperature in the Arrhenius equation; establishing a single-site reaction kinetic model based on the foregoing method and data; obtaining the kinetic parameters of polymerization reaction and the objective function value by using the Bayesian optimization algorithm based on the experimentally measured data and the single-site reaction kinetic model. E2 includes: assuming that the number of active sites is one more than that in the previous calculation step, establishing a multi-site reaction kinetic model; obtaining the kinetic parameters of polymerization reaction and the objective function value by using the Bayesian algorithm based on the experimentally measured data and the multi-site reaction kinetic model.

Description

Technical Field

[0001] The present invention relates to the technical field of olefin polymerization, and particularly relates to a method for estimating kinetic parameters of olefin polymerization reaction based on Bayesian optimization. Background Art

[0002] Polyolefins generally refer to a general term for a class of thermoplastic resins obtained by polymerizing alone or copolymerizing α-olefins such as ethylene, propylene, 1-butene, 1-pentene, 1-hexene, 1-octene, 4-methyl-1-pentene, and certain cycloolefins. Polyolefins have excellent thermal and mechanical properties and are one of the most important thermoplastic products. The polymerization reaction is the key and core of the entire polyolefin process, and accurate olefin polymerization reaction kinetics is the most important step in constructing a polymerization reaction model that can accurately predict product properties. The polymerization reaction kinetics model is of great significance for both scientific research and industrial engineering.

[0003] There is a complex coupling relationship between the polymerization kinetic parameters and the product targets. Adjusting each kinetic parameter may affect multiple target variables. Therefore, there is a huge workload in fine-tuning the kinetic parameters to match the experimental data by manual or distributed adjustment methods. At present, most of the parameter estimation methods and polymer microstructure modeling methods are based on data fitting of small-scale batch experimental data. However, in the experiment, there are situations such as inaccurate control of temperature, pressure, and concentration, and hysteresis, and the model does not take this change into account, resulting in certain errors in the results.

[0004] Chinese Patent Application CN111724861A provides a fitting calculation method for kinetic parameters of olefin polymerization reactions. It models through the data fitting module of Aspen Plus software and is applicable to the fitting of kinetic parameters for single-active and multi-active site catalysis. For multi-active site kinetics, it first obtains the numerical values of catalyst active sites through deconvolution. This method is only applicable when the temperature, pressure, and concentrations of each reaction component remain constant during the polymerization process. When there are fluctuations in the temperature, pressure, or concentrations of each reaction component during the polymerization process, the deconvolution method cannot obtain the accurate number of active sites because fluctuations in the temperature, pressure, or concentrations of each reaction component during the polymerization process will also cause the polymer dispersity index to deviate from 2, and the deconvolution method assumes that the dispersity index of the polymer obtained from each active site is 2. Currently, the literature reports on the estimation of kinetic parameters of polymerization reactions do not take into account the changes in temperature, pressure, and reaction component concentrations during the polymerization reaction process (for example: Kinetic parameter estimation of HDPE slurry process from molecular weight distribution: Estimability analysis and multistep methodology. AIChE Journal, 2014. 60(10): p. 3442 - 3459; Modeling Propylene Polymerization in a Two-Reactor System: Model Development and Parameter Estimation. Macromolecular Reaction Engineering, 2022: p. 2200027; An Effective Methodology for Kinetic Parameter Estimation for Modeling Commercial Polyolefin Processes from Plant Data Using Efficient Simulation Software Tools. Industrial & Engineering Chemistry Research, 2019. 58(31): p. 14209 - 14226.).

[0005] In view of the fact that the existing methods for estimating the kinetic parameters of olefin polymerization reactions do not consider the shift in molecular weight distribution caused by changes in polymerization conditions during the polymerization process, it is necessary to find a new parameter estimation method to obtain more accurate polymerization kinetic parameters while considering the influence of polymerization condition changes on the molecular weight distribution. Summary of the Invention

[0006] In view of the above technical problems, the present invention provides a method for estimating the kinetic parameters of olefin polymerization reactions. A polymerization kinetic model is established based on the moment method, instantaneous distribution, Arrhenius equation, and phase equilibrium method, and combined with the Bayesian optimization algorithm to improve the parameter estimation and solution speed.

[0007] A method for estimating the kinetic parameters of olefin polymerization reactions includes the following steps:

[0008] E1: Use a single-site reaction kinetic model that considers temperature and pressure changes during the reaction to determine the kinetic parameters, obtaining the kinetic parameters and the objective function value. If the objective function value is less than or equal to the set value, end the step and execute step E3; if the objective function value is higher than the set value, execute step E2;

[0009] E2: Assume that the number of active sites is one more than the number of active sites in the previous step, and use a multi-site catalyst site model to determine the kinetic parameters, obtaining the kinetic parameters and the objective function value. If the objective function value is less than or equal to the set value, end the step and execute step E3; if the objective function value is higher than the set value, execute step E2 again;

[0010] E3: Output the polymerization reaction kinetic parameters corresponding to all active sites;

[0011] Step E1 specifically includes the following steps:

[0012] E101: Determine a suitable polymerization reaction mechanism and research method;

[0013] E102: Obtain the experimental conditions, including the reactor volume, the masses of the solvent, monomer, comonomer, catalyst, cocatalyst, and molecular weight regulator added, as well as the temperature and pressure of the polymerization reaction, and input a set of initial values of the pre-exponential factor, activation energy, and reference temperature of the Arrhenius equation;

[0014] E103: Establish a single active site reaction kinetic model based on the data in E102 and the polymerization reaction mechanism and research method described in E101. Considering the changes in the concentrations of monomers, comonomers, catalysts, and cocatalysts, as well as the change in the reaction rate constant caused by temperature and pressure changes during the reaction, the single active site reaction kinetic model can obtain monomer conversion rate, comonomer conversion rate, monomer consumption rate, comonomer consumption rate, polymer dispersity index, polymer number average molecular weight, polymer weight average molecular weight, and polymer molecular weight distribution curve;

[0015] E104: Based on the experimental measured data and the single active site reaction kinetic model described in E103, use the Bayesian optimization algorithm to obtain the polymerization reaction kinetic parameters and the objective function value. If the objective function value is less than or equal to the set value, end the step and execute step E3; if the objective function value is higher than the set value, execute step E2;

[0016] Step E2 specifically includes the following steps:

[0017] E201: Assume that the number of active sites is one more than that in the previous calculation step. Establish a multi-active site reaction kinetic model based on the data in E102 and the polymerization reaction mechanism and research method described in E101. Considering the changes in the concentrations of monomers, comonomers, catalysts, and cocatalysts, as well as the change in the reaction rate constant caused by temperature and pressure changes during the reaction, the multi-active site reaction kinetic model can obtain monomer conversion rate, comonomer conversion rate, monomer consumption rate, comonomer consumption rate, polymer dispersity index, polymer number average molecular weight, polymer weight average molecular weight, and polymer molecular weight distribution curve;

[0018] E202: Based on the experimental measured data and the multi-active site reaction kinetic model described in E201, use the Bayesian optimization algorithm to obtain the polymerization reaction kinetic parameters and the objective function value. If the objective function value is less than or equal to the set value, end the step and execute step E3; if the objective function value is higher than the set value, execute step E2 again.

[0019] In E104 and E202, the objective function is the error OBJ1 between the monomer consumption curve Rp obtained by the reaction kinetic model E and the experimentally measured data Rp, the error OBJ2 between the molecular weight distribution data MWD obtained by the reaction kinetic model E and the experimentally measured data MWD, or the weighted average OBJ3 of the two;

[0020] The error OBJ1 and the error OBJ2 are mean square error, root mean square error, or mean absolute error;

[0021] Mean Square Error (MSE): The ratio of the sum of the squares of the differences between the predicted values of the reaction kinetics model and the experimentally measured values to the number of data points.

[0022]

[0023] Root Mean Square Error (RMSE): The square root of the ratio of the sum of the squares of the differences between the predicted values of the reaction kinetics model and the experimentally measured values to the number of data points.

[0024]

[0025] Mean Absolute Error (MAE): The ratio of the sum of the absolute values of the differences between the predicted values of the reaction kinetics model and the experimentally measured values to the number of data points.

[0026]

[0027] Where m represents the number of experimental groups, a total of M groups, n represents the number of data points in each group of experiments, and N m is the total number of data points in the m-th group of experiments, and θ m,n represents the experimentally measured value represents the predicted value of the reaction kinetics model.

[0028] The polymerization reaction mechanism and research methods in step E101 include five categories: chain activation, chain initiation, chain growth, chain transfer, and chain termination. Among them, the chain growth reaction includes the Bernoulli model and the end-group model, and the chain transfer reaction includes transfer to β-H, transfer to hydrogen, transfer to monomer, transfer to comonomer, and transfer to cocatalyst;

[0029] The reaction components and products involved in the polymerization reaction mechanism in step E101 include monomers, comonomers, segments of monomers / comonomers, molecular weight regulators, catalysts, cocatalysts, oligomers, and copolymerized or homopolymerized polymers;

[0030] The research methods in step E101 include the moment method, the instantaneous distribution method, the Arrhenius equation, and the phase equilibrium equation, and can obtain monomer conversion, comonomer conversion, monomer consumption rate, comonomer consumption rate, polymer dispersity index (PDI), number-average molecular weight of the polymer, weight-average molecular weight of the polymer, and polymer molecular weight distribution curve.

[0031] The temperature and pressure of the polymerization reaction, the pre-exponential factor, activation energy, and reference temperature of the Arrhenius equation in step E102: The temperature range of the polymerization reaction is 300 - 500K, the pressure range of the polymerization reaction is 1 - 100 bar, the activation energy range is 0.1 - 500 kJ / mol, the pre-exponential factor range is 1e-7 - 1e5, and the reference temperature range is 200 - 600K.

[0032] The method for modeling the reaction kinetics model of a single active site in step E103 is specifically to use the moment method, instantaneous distribution method, Arrhenius equation, and phase equilibrium equation in E101, solve the differential-algebraic equation using a mathematical modeling tool, and perform simulation calculations to obtain results after inputting the data in E102;

[0033] The model input in step E103 includes the reactor volume determined in step E102, the masses of the solvent, monomer, comonomer, catalyst, cocatalyst, and molecular weight regulator added, the pre-exponential factor and activation energy of the Arrhenius equation, the initial value of the reference temperature, and the temperature and pressure of the polymerization reaction.

[0034] The experimental measured data of the polymerization reaction in step E104 includes at least one set of data, and each set of operating data includes at least the mass of the catalyst added, the mass of the cocatalyst added, the mass of the comonomer added, the mass of the monomer added over time, the data of the temperature and pressure of the polymerization reaction changing over time, the molecular weight distribution of the polymer product, the polydispersity index of the polymer, the conversion rate of the comonomer, and the insertion rate of the comonomer. Among them, the molecular weight distribution data is specifically the polymer gel permeation chromatography analysis data.

[0035] In a preferred example, in step E104, using the polymerization reaction mechanism and research method in E101, the data in step E102 is determined and input into the single active site reaction kinetics model established in E103 to calculate the monomer consumption rate, comonomer conversion rate, comonomer insertion rate, molecular weight distribution, and polydispersity index of the polymer. The molecular weight distribution MWD of the polymer obtained from the kinetic model E The error value OBJ2 between the experimental measured molecular weight distribution MWD is used as the objective function. The objective function value is input into the Bayesian optimization algorithm. The Bayesian optimization algorithm is used to continuously obtain new pre-exponential factors and / or activation energies and / or reference temperatures of the Arrhenius equation, and then calculate the new monomer consumption rate, comonomer conversion rate, comonomer insertion rate, molecular weight distribution, and polydispersity index of the polymer through the single active site reaction kinetics model until the Bayesian optimization algorithm reaches the search times and outputs the objective function value. If the objective function value is less than or equal to the set value, step E1 ends and step E3 is executed; if the objective function value is higher than the set value, step E2 is executed.

[0036] For the multi-active site reaction kinetics model described in step E201, its molecular weight distribution data MWD E is obtained by summing the products of the molecular weight distribution MWD corresponding to each active site j j and the mass fraction mf of the polymer yield obtained from each active site j j :

[0037]

[0038] where N J is the number of catalyst active sites.

[0039] In a preferred example, in step E202, the polymerization reaction mechanism and research method in E101 are used to determine the data in step E102, and input into the multi-active-site reaction kinetic model established in E202 to calculate the monomer consumption rate, comonomer conversion rate, comonomer insertion rate, molecular weight distribution, and polymer dispersity index. The polymer molecular weight distribution MWD obtained from the kinetic model E and the error value OBJ2 between the experimentally measured molecular weight distribution MWD are used as the objective function. The objective function value is input into the Bayesian optimization algorithm. The Bayesian optimization algorithm is used to continuously obtain new pre-exponential factors and / or activation energies and / or reference temperatures of the Arrhenius equation, and then calculate the new monomer consumption rate, comonomer conversion rate, comonomer insertion rate, molecular weight distribution, and polymer dispersity index through the multi-active-site reaction kinetic model until the Bayesian optimization algorithm reaches the search times and outputs the objective function value. If the objective function value is less than or equal to the set value, step E2 ends and step E3 is executed; if the objective function value is higher than the set value, step E2 is executed again.

[0040] The monomer can be at least one of ethylene and propylene, the comonomer can be at least one of propylene, 1-butene, 1-hexene, and 1-octene, the catalyst can be at least one of metallocene catalysts and Ziegler-Natta catalysts, the cocatalyst can be at least one of methylaluminoxane, triethylaluminum, and triisobutylaluminum, the molecular weight regulator can be hydrogen, and the solvent can be a linear alkane or cycloalkane with C4-C12, an aromatic hydrocarbon with C6-C9, or a mixed solvent thereof.

[0041] The present invention establishes a reaction kinetic model through copolymerization mechanism, moment method, instantaneous distribution method, Arrhenius equation, and phase equilibrium equation theory using a mathematical modeling tool, and combines the Bayesian optimization algorithm to construct a method for estimating kinetic parameters of olefin polymerization reaction. Its advantages include:

[0042] (1) Considering the situation of polymer molecular weight distribution shift caused by temperature and pressure changes and reaction component concentration changes during the polymerization reaction process, and using the error between the fitting value and experimental value of the molecular weight distribution as the objective function, more accurate kinetic parameters can be obtained through the optimization algorithm.

[0043] (2) The Bayesian optimization algorithm has a higher contraction efficiency than deterministic algorithms, and the solution time of this method is short. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 is a schematic flow chart of the method for estimating kinetic parameters of olefin polymerization reaction of the present invention;

[0045] Figure 2 It is a schematic diagram of the specific operation process of step E1 of the method for estimating the kinetic parameters of the olefin polymerization reaction of the present invention;

[0046] Figure 3 It is a schematic diagram of the specific operation process of step E2 of the method for estimating the kinetic parameters of the olefin polymerization reaction of the present invention;

[0047] Figure 4 It is the ethylene flow rate curve obtained from the kinetic model of the method in Example 1 and the ethylene flow rate curve obtained through experiments;

[0048] Figure 5 It is the molecular weight distribution curve obtained from the kinetic model of the method in Example 1 and the molecular weight distribution curve obtained through experiments;

[0049] Figure 6 It is the molecular weight distribution curve at different times calculated by the kinetic model in Example 1 and the cumulative molecular weight distribution curve;

[0050] Figure 7 It is the experimental temperature change curve of Example 1;

[0051] Figure 8 It is the concentration change curve of ethylene monomer in Example 1;

[0052] Figure 9 It is the fluctuation of the chain growth rate constant in Example

END

[0053] Figure 10 It is the molecular weight distribution curve obtained from the kinetic model of the method in Example 2 and the molecular weight distribution curve obtained through experiments. Detailed implementation manners

[0054] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and not to limit the scope of the present invention.

[0055] As Figure 1 shown, the method for estimating the kinetic parameters of the olefin polymerization reaction of the present invention includes:

[0056] E1: Use a single-site reaction kinetic model considering the temperature and pressure changes during the reaction to determine the kinetic parameters, obtain the kinetic parameters and the objective function value. If the objective function value is less than or equal to the set value, end the step and execute step E3; if the objective function value is higher than the set value, execute step E2;

[0057] E2: Assume that the active site is the number of active sites in the previous step plus one. Determine the kinetic parameters using the multi-active-site catalyst site model to obtain the kinetic parameters and the objective function value. If the objective function value is less than or equal to the set value, end the step and execute step E3; if the objective function value is higher than the set value, execute step E2 again;

[0058] E3: Output the kinetic parameters of the polymerization reaction corresponding to all active sites.

[0059] As Figure 2 shown, step E1 further includes:

[0060] E101: Determine a suitable polymerization reaction mechanism and research method;

[0061] E102: Obtain the experimental conditions, including the reactor volume, the masses of the solvent, monomer, comonomer, catalyst, cocatalyst, and molecular weight regulator added, as well as the temperature and pressure during the polymerization reaction process, and input a set of initial values of the pre-exponential factor, activation energy, and reference temperature of the Arrhenius equation;

[0062] E103: Establish a single-active-site reaction kinetic model based on the data in E102 and the polymerization reaction mechanism and research method described in E101. Considering the changes in the concentrations of the monomer, comonomer, catalyst, and cocatalyst and the change in the reaction rate constant caused by the temperature and pressure changes during the reaction, the single-active-site reaction kinetic model can obtain the monomer conversion rate, comonomer conversion rate, monomer consumption rate, comonomer consumption rate, polymer dispersity index, polymer number-average molecular weight, polymer weight-average molecular weight, and polymer molecular weight distribution curve;

[0063] E104: Based on the experimental measured data and the single-active-site reaction kinetic model described in E103, use the Bayesian optimization algorithm to obtain the polymerization reaction kinetic parameters and the objective function value. If the objective function value is less than or equal to the set value, end the step and execute step E3; if the objective function value is higher than the set value, execute step E2.

[0064] As Figure 3 shown, step E2 further includes:

[0065] E201: Assume that the active site is the number of active sites in the previous calculation step plus one. Based on the data in E102 and the polymerization reaction mechanism and research method described in E101, establish a multi-active-site reaction kinetic model. Considering the changes in the concentrations of monomers, comonomers, catalysts, and cocatalysts and the change in the reaction rate constant caused by temperature and pressure changes during the reaction, the multi-active-site reaction kinetic model can obtain the monomer conversion rate, comonomer conversion rate, monomer consumption rate, comonomer consumption rate, polymer dispersity index, polymer number-average molecular weight, polymer weight-average molecular weight, and polymer molecular weight distribution curve.

[0066] E202: Based on the experimentally measured data and the multi-active-site reaction kinetic model described in E201, use the Bayesian optimization algorithm to obtain the polymerization reaction kinetic parameters and the objective function value. If the objective function value is less than or equal to the set value, end the step and execute step E3; if the objective function value is higher than the set value, execute step E2 again.

[0067] Example 1

[0068] In Example 1 of the present invention, the polymerization reaction mechanism described in step E101 includes five categories: chain activation, chain initiation, chain growth, chain transfer, and chain termination. The chain growth adopts an end-group model, and the chain transfer considers transfer to hydrogen, transfer to monomer, transfer to comonomer, and transfer to cocatalyst. The reaction components and products involved include ethylene monomer, comonomer, monomer / comonomer chain segments, molecular weight regulator, catalyst, cocatalyst, oligomer, copolymer or homopolymer. The polymerization mechanism and components are shown in Table 1, where M j represents the monomer and comonomer, j = 1, 2; P r,j represents the active chain segment ending with j, Al represents the cocatalyst, C represents the catalyst, C * represents the catalyst active site, C d represents the catalyst deactivation site, D r represents the deactivated chain segment. The research method in step E101 includes the moment method, the instantaneous distribution method, the Arrhenius equation, and the phase equilibrium equation.

[0069] Table 1

[0070]

[0071] Table 2 and Table 3 show some of the experimental conditions determined in step E102, and Table 4 shows the initial values of the kinetic parameters determined through literature review and calculation experience.

[0072] Table 2

[0073] Component Function Component Name Monomer Ethylene Comonomer Octene Solvent Hexane Catalyst Dimethylsilyl-tert-butylamine tetramethylcyclopentadienyl titanium dichloride (CGC-Ti) Cocatalyst Methylaluminoxane Molecular Weight Regulator Hydrogen

[0074] Table 3

[0075] Experimental Conditions Value Ethylene Concentration 1.55 mol / L Octene Concentration 0.94 mol / L Catalyst Concentration <![CDATA[25*10 -6 mol / L]]> Cocatalyst Concentration <![CDATA[25*10 -3 mol / L]]> Hydrogen Concentration 0.002 mol / L Reactor Volume 1.2L Reaction Time 600s

[0076] Table 4

[0077]

[0078] Step E103 establishes a kinetic model for single-site polymerization reactions based on the polymerization mechanism, research method, and experimental conditions determined by E101 and E102. The model takes into account the changes in the concentrations of monomers, comonomers, catalysts, and cocatalysts, as well as the reaction rate, caused by temperature and pressure changes during the reaction. The specific process is as follows: By using the Pyomo software package on the Python platform, input the experimental condition data determined by E102, the initial value of the pre-exponential factor of the Arrhenius equation, the activation energy, and the reference temperature. Use the research method of E101 and the polymerization reaction mechanism to establish process constraints, and discretize and solve differential-algebraic equations (DAEs) through the Pyomo software package. The model can obtain monomer conversion, comonomer conversion, monomer consumption rate, comonomer consumption rate, polydispersity index (PDI), number-average molecular weight of the polymer, weight-average molecular weight of the polymer, and polymer molecular weight distribution curve.

[0079] The specific method for performing the optimization calculation in Step E104 is as follows: Based on the Python software platform, combine the kinetic model of single-site polymerization reaction established in E103 and the Bayesian optimization algorithm to establish a kinetic parameter calculation model. Input the experimental data and parameter initial values of Step E102 of the polymerization reaction into the mathematical modeling, and use the mean square error value between the molecular weight distribution MWD of the polymer product obtained from the kinetic model E and the experimentally measured molecular weight distribution MWD as the objective function value. In this embodiment, the reference temperature of the Arrhenius equation is a fixed value. Continuously obtain new pre-exponential factors and activation energies of the Arrhenius equation by using the Bayesian optimization algorithm, and then calculate new monomer consumption rates, comonomer conversions, comonomer insertion rates, molecular weight distributions, etc. through the kinetic model of single-site polymerization reaction until the Bayesian optimization algorithm reaches the search times (1000 search times in this embodiment), and then output the objective function value. In the embodiment of the present invention, according to Step E104 for optimization calculation, the time required for the Bayesian optimization algorithm to search 1000 times is 3152 seconds. The obtained objective function value and the set value of the objective function are shown in Table 5. The objective function value is less than the set value, Step E1 is ended, and Step E3 is executed. The finally obtained kinetic parameters of the single-site reaction are shown in Table 6.

[0080] The ethylene flow rate curve obtained from the kinetic model of the method in this embodiment and the ethylene flow rate curve obtained through experiments are as Figure 4 shown, and the fitting degree is good. The molecular weight distribution curve obtained from the kinetic model of the method in this embodiment and the molecular weight distribution curve obtained through experiments are asFigure 5 As shown, the goodness of fit is good. The PDI of the polymer product obtained from the experiment is 2.3, and the PDI of the polymer product obtained from the kinetic model of the method in this example is 2.29. The molecular weight distribution curves at different times calculated by the kinetic model and the cumulatively obtained molecular weight distribution curves are as Figure 6 shown. As can be seen from Figure 6 , due to large fluctuations in the temperature during the experiment (the experimental temperature change curve is as Figure 7 shown), there are fluctuations in the concentration of ethylene monomer ( Figure 8 shown), and at the same time, the temperature fluctuations also lead to fluctuations in the rate constant ( Figure 9 shows the fluctuations in the chain growth rate constant), resulting in certain differences in the molecular weight distribution at different times (t = 100s, t = 400s, t = 600s) during the polymerization reaction process, and further resulting in the dispersity index of the polymer product not being equal to 2.

[0081] In Comparative Example 1, the same experimental data as in Example 1 were used, and the method of the present invention was applied to obtain kinetic parameters. The difference is that the Bayesian optimization method was not used, but the deterministic algorithm was used to solve. Within the same solution time as in Example 1, the deterministic algorithm could not obtain a feasible solution.

[0082] In Comparative Example 2, the same experimental data as in Example 1 were used, and the method of the present invention was applied to obtain kinetic parameters. The difference is that the temperature fluctuations in the experimental conditions were removed, and it was assumed that the temperature during the experiment was constant. The Bayesian optimization was searched 1000 times, and the objective function value obtained was 0.12. The PDI of the polymer product obtained by the model was 2, with a large error from the experimental measurement value. In Comparative Example 2, the model did not consider the influence of temperature and could not consider the influence of temperature on the molecular weight distribution at different times during the polymerization process when estimating parameters.

[0083] Table 5 Objective function value and objective function set value

[0084] Objective Function Value 0.0074 Objective Function Set Value 0.01

[0085] Table 6 Kinetic parameters of single active site reaction for optimization calculation

[0086]

[0087] Example 2

[0088] In Example 2 of the present invention, taking another polymerization experiment as an example, its polymerization mechanism is the same as that in Table 1, and the reaction system is the same as that in Table 2, only the catalyst and cocatalyst are different. The catalyst is TiCl4, and the cocatalyst is triethylaluminum. In Example 2, after passing through E101 - E104, using the Bayesian optimization algorithm and the single active site kinetic model, after searching 1000 times, the obtained objective function is 0.15, which is greater than the set value of 0.01. Continue to execute E201. Assume that the active site is the number of active sites in the previous calculation step plus one. Based on the data in E102 and the polymerization reaction mechanism and research method in E101, establish a multi - active site reaction kinetic model, considering the changes in the concentrations of monomer, comonomer, catalyst, and cocatalyst and the change in the reaction rate constant caused by temperature and pressure changes during the reaction. The model can obtain monomer conversion, comonomer conversion, monomer consumption rate, comonomer consumption rate, polymer dispersity index (PDI), polymer number - average molecular weight, polymer weight - average molecular weight, and polymer molecular weight distribution curve.

[0089] In step E202, use the Bayesian optimization algorithm and the multi - active site reaction kinetic model to calculate the molecular weight distribution MWD corresponding to each active site j j and the mass fraction mf of polymer production j , obtain the polymerization reaction kinetic parameters and the objective function value. The obtained kinetic parameters are the kinetic parameters corresponding to each active site, and the obtained objective function value is the error between the total molecular weight distribution obtained by the sum of the products of the molecular weight distribution of each active site and the mass fraction of polymer production at that site and the molecular weight distribution of the experimental measured data. If the objective function value is less than or equal to the set value, end step E2 and execute step E3; if the objective function value is higher than the set value, execute step E2 again.

[0090] In Example 2, according to the multi - active site reaction kinetic model and the Bayesian optimization algorithm established in step E2, when the number of active sites is 2, the objective function is 0.034, which is greater than the set value. Therefore, continue to execute step E2. When the number of active sites is 3, the objective function is 0.0089, which is less than the set value, and execute step E3. The objective function values obtained for different numbers of active sites and the set value of the objective function are shown in Table 7. When there are 3 active sites, the mass fractions of polymer production corresponding to each active site are shown in Table 8. Output the polymerization reaction kinetic parameters corresponding to all active sites, as shown in Table 9. The molecular weight distribution curve obtained by the kinetic model of the method in this example and the molecular weight distribution curve obtained by the experiment are as Figure 10 shown, with good fitting degree. The PDI of the polymer product obtained by the experiment is 3.6, and the PDI of the polymer product obtained by the kinetic model of the method in this example is 3.62. The molecular weight distribution of the polymer obtained with 3 active sites is also shown in Figure 10Among them, it is worth noting that for the kinetic fitting problem of multi-active site catalysts, the main difference between this method and the conventional method is that this method does not use the deconvolution method to obtain the number of catalyst active sites. Because the deconvolution method cannot handle the fluctuations of reaction conditions during the polymerization process, there are large errors in obtaining the number of catalyst active sites through the deconvolution method.

[0091] Table 7 Fitting Errors of Different Numbers of Active Centers

[0092] Number of Active Centers Objective Function Value Objective Function Set Value 1 0.15 0.01 2 0.034 0.01 3 0.0089 0.01

[0093] Table 8 Mass Fraction of Polymer Yield Corresponding to Each Active Site

[0094] ​ 1 2 3 ​ 0.149 0.540 0.311

[0095] Table 9 Kinetic Parameters of Multi-Active Site Reaction Optimized by Calculation

[0096]

[0097]

[0098] The method of the present invention can be applied to the olefin polymerization process with homopolymerization and copolymerization mechanisms, and the method of the present invention can obtain more accurate polymerization reaction kinetic parameters, and the combination of the Bayesian optimization algorithm makes the solution speed of this method faster.

[0099] In addition, it should be understood that after reading the above description of the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms also fall within the scope defined by the appended claims of this application.

Claims

1. A method for estimating kinetic parameters of olefin polymerization reaction, characterized in that, It includes the following steps: E1: Determine the kinetic parameters using a single active site reaction kinetic model that takes into account the temperature and pressure changes during the reaction, obtaining the kinetic parameters and the objective function value. If the objective function value is less than or equal to the set value, end the step and execute step E3; if the objective function value is higher than the set value, execute step E2; E2: Assume that the number of active sites is one more than that in the previous step, and determine the kinetic parameters using a multi-active site catalyst site model, obtaining the kinetic parameters and the objective function value. If the objective function value is less than or equal to the set value, end the step and execute step E3; If the objective function value is higher than the set value, execute step E2 again; E3: Output the polymerization reaction kinetic parameters corresponding to all active sites; Step E1 specifically includes the following steps: E101: Determine the suitable polymerization reaction mechanism and research method; E102: Obtain the experimental conditions, including the reactor volume, the masses of the solvent, monomer, comonomer, catalyst, cocatalyst, and molecular weight regulator added, as well as the temperature and pressure of the polymerization reaction, and input a set of initial values of the pre-exponential factor, activation energy, and reference temperature of the Arrhenius equation; E103: Establish a single active site reaction kinetic model based on the data in E102 and the polymerization reaction mechanism and research method described in E101. Considering the changes in the concentrations of the monomer, comonomer, catalyst, and cocatalyst and the reaction rate constant caused by the temperature and pressure changes during the reaction, the single active site reaction kinetic model can obtain the monomer conversion rate, comonomer conversion rate, monomer consumption rate, comonomer consumption rate, polymer dispersity index, polymer number average molecular weight, polymer weight average molecular weight, and polymer molecular weight distribution curve; E104: Based on the experimental measured data and the single active site reaction kinetic model described in E103, use the Bayesian optimization algorithm to obtain the polymerization reaction kinetic parameters and the objective function value. If the objective function value is less than or equal to the set value, end the step and execute step E3; if the objective function value is higher than the set value, execute step E2; Step E2 specifically includes the following steps: E201: Assume that the number of active sites is one more than that in the previous calculation step, and establish a multi-active site reaction kinetic model based on the data in E102 and the polymerization reaction mechanism and research method described in E101. Considering the changes in the concentrations of the monomer, comonomer, catalyst, and cocatalyst and the reaction rate constant caused by the temperature and pressure changes during the reaction, the multi-active site reaction kinetic model can obtain the monomer conversion rate, comonomer conversion rate, monomer consumption rate, comonomer consumption rate, polymer dispersity index, polymer number average molecular weight, polymer weight average molecular weight, and polymer molecular weight distribution curve; E202: Based on the experimental measured data and the multi-active site reaction kinetic model described in E201, use the Bayesian optimization algorithm to obtain the polymerization reaction kinetic parameters and the objective function value. If the objective function value is less than or equal to the set value, end the step and execute step E3; if the objective function value is higher than the set value, execute step E2 again.

2. The method for estimating the kinetic parameters of olefin polymerization reaction according to claim 1, wherein In E104 and E202, the objective function is the monomer consumption curve Rp obtained from the reaction kinetics model E The error OBJ1 between the experimental measured data Rp, the molecular weight distribution data MWD obtained from the reaction kinetics model E The error OBJ2 between the experimental measured data MWD or the weighted average OBJ3 of the two; The error OBJ1 and the error OBJ2 are the mean square error, the root mean square error, or the mean absolute error; The mean square error MSE: the ratio of the sum of the squares of the deviations between the predicted values of the reaction kinetic model and the experimentally measured values to the number of data points, The root mean square error RMSE: the square root of the ratio of the sum of the squares of the deviations between the predicted values of the reaction kinetic model and the experimentally measured values to the number of data points, The mean absolute error MAE: the ratio of the sum of the absolute values of the deviations between the predicted values of the reaction kinetic model and the experimentally measured values to the number of data points, Where m represents the number of experimental groups, with a total of M groups, n represents the number of data points included in each experiment, and N m is the total number of data points included in the m-th experiment, and θ m,n represents the measured experimental value represents the predicted value of the reaction kinetics model.

3. The method for estimating the kinetic parameters of olefin polymerization reaction according to claim 1, characterized in that, The polymerization reaction mechanism and research methods in step E101 include five categories: chain activation, chain initiation, chain growth, chain transfer, and chain termination. Among them, the chain growth reaction includes the Bernoulli model and the end-group model, and the chain transfer reaction includes transfer to β-H, transfer to hydrogen, transfer to monomer, transfer to comonomer, and transfer to cocatalyst; The reaction components and products involved in the polymerization reaction mechanism in step E101 include monomers, comonomers, segments of monomers / comonomers, molecular weight regulators, catalysts, cocatalysts, oligomers, and copolymers or homopolymers; The research methods in step E101 include the moment method, the instantaneous distribution method, the Arrhenius equation, and the phase equilibrium equation, and can obtain monomer conversion, comonomer conversion, monomer consumption rate, comonomer consumption rate, polymer dispersity index, number-average molecular weight of the polymer, weight-average molecular weight of the polymer, and polymer molecular weight distribution curve.

4. The method for estimating the kinetic parameters of olefin polymerization reaction according to claim 1, characterized in that, The temperature and pressure of the polymerization reaction, the pre-exponential factor, activation energy, and reference temperature of the Arrhenius equation in step E102: the temperature range of the polymerization reaction is 300 - 500K, the pressure range of the polymerization reaction is 1 - 100 bar, the activation energy range is 0.1 - 500 kJ / mol, the pre-exponential factor range is 1e-7 - 1e5, and the reference temperature range is 200 - 600K.

5. The method for estimating the kinetic parameters of olefin polymerization reaction according to claim 3, characterized in that, The method for modeling the single active site reaction kinetic model in step E103 is specifically to use the moment method, instantaneous distribution method, Arrhenius equation, and phase equilibrium equation in E101, solve the differential-algebraic equations using a mathematical modeling tool, and perform simulation calculations after inputting the data in E102 to obtain the results; The model input in step E103 includes the reactor volume determined in step E102, the mass of the added solvent, monomers, comonomers, catalyst, cocatalyst, molecular weight regulator, the pre-exponential factor and activation energy of the Arrhenius equation, the initial value of the reference temperature, and the temperature and pressure of the polymerization reaction.

6. The method for estimating the kinetic parameters of olefin polymerization reaction according to claim 1, characterized in that The experimentally measured data of the polymerization reaction in step E104 includes at least one set of data, and each set of operating data includes at least the mass of the added catalyst, the mass of the added cocatalyst, the mass of the added comonomer, the mass of the monomer added over time, the data of the temperature and pressure of the polymerization reaction over time, the molecular weight distribution of the polymer product, the polymer dispersity index, comonomer conversion, and comonomer insertion rate. Among them, the molecular weight distribution data is specifically the polymer gel permeation chromatography analysis data.

7. The method for estimating the kinetic parameters of olefin polymerization reaction according to claim 1, wherein In step E104, using the polymerization reaction mechanism and research method in E101, determine the data in step E102 and input it into the single-site reaction kinetic model established in E103 to calculate the monomer consumption rate, comonomer conversion rate, comonomer insertion rate, molecular weight distribution, and polymer dispersity index, and obtain the polymer molecular weight distribution MWD from the kinetic model. E The error value OBJ2 between the polymer molecular weight distribution MWD obtained from the kinetic model and the experimentally measured molecular weight distribution MWD is used as the objective function. Input the objective function value into the Bayesian optimization algorithm, and continuously obtain new pre-exponential factors and / or activation energies and / or reference temperatures of the Arrhenius equation using the Bayesian optimization algorithm. Then, calculate the new monomer consumption rate, comonomer conversion rate, comonomer insertion rate, molecular weight distribution, and polymer dispersity index through the single-site reaction kinetic model until the Bayesian optimization algorithm reaches the search times and outputs the objective function value. If the objective function value is less than or equal to the set value, end step E1 and execute step E3; if the objective function value is higher than the set value, execute step E2.

8. The method for estimating the kinetic parameters of olefin polymerization reaction according to claim 1, characterized in that, The multi-active-site reaction kinetics model described in step E201, with molecular weight distribution data MWD E is obtained by summing the product of the molecular weight distribution MWD corresponding to each active site j j and the mass fraction mf of the polymer yield obtained at each active site j j : where N J is the number of catalyst active sites.

9. The method for estimating the kinetic parameters of olefin polymerization reaction according to claim 1, characterized in that, In step E202, the polymerization reaction mechanism and research method in E101 are used to determine the data in step E102, and input it into the multi-active-site reaction kinetic model established in E202 to calculate the monomer consumption rate, comonomer conversion rate, comonomer insertion rate, molecular weight distribution, and polymer dispersity index. The polymer molecular weight distribution MWD obtained from the kinetic model E The error value OBJ2 between the polymer molecular weight distribution MWD obtained from the kinetic model and the experimentally measured molecular weight distribution MWD is used as the objective function. The objective function value is input into the Bayesian optimization algorithm. The Bayesian optimization algorithm is used to continuously obtain new pre-exponential factors and / or activation energies and / or reference temperatures of the Arrhenius equation, and then calculate the new monomer consumption rate, comonomer conversion rate, comonomer insertion rate, molecular weight distribution, and polymer dispersity index through the multi-active-site reaction kinetic model until the Bayesian optimization algorithm reaches the search times and outputs the objective function value. If the objective function value is less than or equal to the set value, step E2 ends and step E3 is executed; if the objective function value is higher than the set value, step E2 is executed again.

10. The method for estimating the kinetic parameters of olefin polymerization reaction according to claim 1, wherein, The monomer is at least one of ethylene and propylene, the comonomer is at least one of propylene, 1-butene, 1-hexene, and 1-octene, the catalyst is at least one of metallocene catalysts and Ziegler-Natta catalysts, the cocatalyst is at least one of methylaluminoxane, triethylaluminum, and triisobutylaluminum, the molecular weight regulator is hydrogen, and the solvent is a linear or cyclic alkane having 4 to 12 carbon atoms, an aromatic hydrocarbon having 6 to 9 carbon atoms, or a mixed solvent thereof.

Images