A mechanistic modeling method for catalytic cracking process

Abstract

The present invention relates to the field of catalytic cracking reaction of hydrogenated catalytic light cycle oil, and more specifically, to a method for modeling the mechanism of catalytic cracking process. The method comprises the following steps: step S1, obtaining process data of production equipment; step S2, dividing reaction components according to the principle of kinetic similarity; step S3, setting basic reaction kinetic assumptions for the catalytic cracking process; step S4, based on the basic reaction kinetic assumptions, determining the reaction relationship of each reaction component and constructing a reaction network; step S5, establishing a reaction kinetic model; step S6, selecting an objective function, optimizing the kinetic parameters involved in each reaction, and outputting the optimal kinetic parameter combination; step S7, predicting the yield and properties of related products based on the optimal kinetic parameter combination. The present invention optimizes and solves by improving the gray wolf optimization algorithm, thereby improving the model prediction accuracy, and the constructed mechanism model has good prediction ability.

Description

Technical Field

[0001] The present invention relates to the field of catalytic cracking reaction of hydrocatalytic light cycle oil, and more particularly to a catalytic cracking process mechanism modeling method for producing high-octane gasoline and aromatics using hydrocatalytic light cycle oil as a main raw material. Background Art

[0002] The catalytic cracking unit (FCC) is one of the most common heavy oil lightening processing units and a key means of producing gasoline, diesel, and light olefins. While producing gasoline, FCC produces a large amount of FCCD diesel fuel with high aromatic content and low cetane number, also known as light cycle oil (LCO). This fuel has a high density, poor oxidative stability, and its composition falls significantly short of the high saturated hydrocarbon content, high hydrogen content, and high cetane number required by my country's clean diesel standards for vehicles. Reducing or converting LCO to lower the diesel-to-gasoline ratio can enhance product value, promote structural adjustments within refining and chemical enterprises, and promote environmental protection, making it a research hotspot for refining companies.

[0003] Relevant research shows that the hydrogen consumption of catalytic diesel hydrotreating is too high and the diesel cetane number cannot be significantly improved; while the hydrotreating technology requires major equipment modifications, the investment is not cost-effective, and neither of the above two processing routes can reasonably utilize the rich aromatics in catalytic diesel.

[0004] Combining a hydrogenation unit with a catalytic cracking unit can increase the production of high-octane gasoline or light aromatics. That is, LCO is first subjected to selective hydrogenation saturation and then returned to the catalytic cracking unit for cracking reaction, thereby increasing the production of high-octane gasoline or light aromatics. To a certain extent, it not only solves the outlet of LCO, but also improves the use value of LCO, meeting the market demand of enterprises to reduce the diesel-to-gasoline ratio. It has been widely used in domestic refining and chemical enterprises and has produced good economic benefits.

[0005] The process for producing high-octane gasoline and aromatics using hydrocatalytic light cycle oil as the main raw material has the characteristics of high gasoline selectivity, high octane number, low hydrogen consumption, low investment and flexible operation.

[0006] At present, the research and analysis of this process mainly focuses on the research and analysis of the process reaction path and application effect, as well as the operation status analysis and technological transformation results of related production equipment. However, there is less research on process modeling. However, it is of great significance to establish a mechanism model of the catalytic cracking reaction process of hydrocatalytic light cycle oil and predict the yield and properties of related products in this process. Summary of the Invention

[0007] The purpose of the present invention is to provide a mechanistic modeling method for a catalytic cracking process that uses hydrogenated catalytic light cycle oil as the main raw material to produce more high-octane gasoline and aromatics, thereby solving the problem in the prior art of difficulty in accurately predicting the yield and properties of related products in the catalytic cracking process for producing high-octane gasoline and aromatics.

[0008] In order to achieve the above object, the present invention provides a catalytic cracking process mechanism modeling method, comprising the following steps:

[0009] Step S1, obtaining production device process data;

[0010] Step S2, dividing the reaction components according to the principle of kinetic similarity;

[0011] Step S3: setting basic reaction kinetic assumptions for the catalytic cracking process for model simplification;

[0012] Step S4: Based on the basic reaction kinetics assumptions of step S3, determine the reaction relationship of each reaction component in step S2 and construct a reaction network;

[0013] Step S5: establishing a reaction kinetics model based on reaction components, basic reaction kinetics assumptions and reaction network;

[0014] Step S6: Select an objective function, optimize the kinetic parameters involved in each reaction, and output the optimal kinetic parameter combination;

[0015] Step S7: Based on the optimal kinetic parameter combination, predict the yield and properties of related products.

[0016] In one embodiment, the catalytic cracking process is a catalytic cracking process that uses hydrocatalytic light cycle oil as a main raw material to produce more high-octane gasoline and aromatics;

[0017] The production unit process data of step S1 further includes feed properties, catalyst properties, product yield and operating conditions, etc.

[0018] In one embodiment, the feed further comprises hydrocatalytic light cycle oil (hydrocatalytic diesel) and / or light dirty oil and / or heavy oil;

[0019] The products further include catalytic diesel, stabilized gasoline, light olefins, liquefied gas, dry gas and coke;

[0020] The operating conditions further include reaction temperature, reaction pressure, reactant-to-oil ratio and true weight hourly space velocity, etc.

[0021] In one embodiment, the step S2 further includes performing component classification based on the distillation range and the family composition;

[0022] The component division based on the distillation range and the family composition further includes:

[0023] The heavy oil in the feed is divided into four components;

[0024] The diesel fraction and gasoline fraction are divided according to the PONA group composition.

[0025] In one embodiment, the basic assumptions of the reaction kinetics of step S3 further include:

[0026] All reactions are first-order irreversible reactions;

[0027] The reactions involved are treated as homogeneous reactions;

[0028] The gas flow state in the riser reactor involved in the reaction is isothermal, gas phase, ideal plug flow, and the diffusion within the material particles is ignored;

[0029] The catalyst deactivation problem involved in the reaction is characterized by the time-varying deactivation of the catalyst, and all reactions are carried out on the same acidic active sites;

[0030] The reaction rate was corrected using empirical formulas for heavy aromatic hydrocarbon adsorption deactivation and alkaline nitrogen adsorption deactivation.

[0031] The gas does not generate coke.

[0032] In one embodiment, step S5 further includes:

[0033] Considering the factors affecting catalyst deactivation, basic nitrogen and heavy aromatics adsorption, a reaction kinetic model was established;

[0034] The reaction rate expression of the reaction kinetic model is:

[0035]

[0036] Where a is the mass concentration vector of the reaction component;

[0037] K is the reaction rate constant matrix;

[0038] is the catalyst coking deactivation equation;

[0039] f(A) is the adsorption deactivation equation for heavy aromatics;

[0040] f(N) is the alkaline nitrogen adsorption deactivation equation.

[0041] In one embodiment, the step S6 further includes: minimizing the sum of squares of errors between the model predicted value and the actual value as the objective function of parameter estimation.

[0042] In one embodiment, the step S6 further comprises: optimizing and solving the kinetic parameters involved in each reaction using an improved Grey Wolf optimization algorithm;

[0043] The improved gray wolf optimization algorithm further comprises the following steps:

[0044] Step S601: Set the population size, parameter dimension, maximum number of iterations, parameter variable interval, mutation factor and crossover probability;

[0045] Step S602: Initialize individuals in the population;

[0046] Step S603: Calculate the fitness value of the individuals in the population according to the objective function;

[0047] Step S604: Perform elite reverse learning;

[0048] Step S605: Determine the position of the leader wolf in the wolf pack, calculate the relative distances between the leader wolf and other individuals in the wolf pack, and update the individual positions;

[0049] Step S606: Modify the current optimal solution based on the differential evolution operator;

[0050] Step S607: Determine whether the termination condition of the algorithm optimization is met. If the termination condition is not met, return to step S604 to perform the next round of iterative calculation. If the termination condition is met, the optimization ends and the current optimal solution is output.

[0051] In one embodiment, step S604 further includes: calculating individual fitness according to the objective function, selecting a portion of elite individuals, obtaining an elite reverse solution using a reverse learning method, and updating the individual position by selecting the individual with the higher fitness between the elite reverse solution and the current individual;

[0052] The calculation formula of the elite reverse solution is:

[0053]

[0054] Where, X i,j represents the elite individual of the i-th population in the iterative calculation, x min j and x max j are the dynamic boundaries of the j-th dimension search space, and η is the generalized coefficient.

[0055] In one embodiment, the step S605 further includes:

[0056] Determine the location of the leader of the pack and calculate the relative distances of the other individuals in the pack to the leader;

[0057] Based on the relative distance and approach direction from the leader wolf, the next position of other individuals in the wolf pack is calculated and updated.

[0058] In one embodiment, step S606 further includes: performing mutation, crossover, and selection operations on the current optimal solution based on a differential evolution operator:

[0059] The corresponding calculation formula for the mutation operation is:

[0060] Best_pos'=Best_pos+F*X rand ;

[0061] Among them, Best_pos is the individual with the best current fitness, F is the variation factor, X rand is the difference calculation;

[0062] The crossover operation corresponds to the following calculation formula:

[0063]

[0064] Where cr is the crossover probability and r is a random number between (0,1);

[0065] The selection operation is to select the individual with better fitness from the individual p after mutation and crossover and Best_pos as the optimal solution for this iteration.

[0066] In one embodiment, the optimization termination condition set in step S607 includes: the error between all components and the target yield is less than a first percentage or the current iteration number reaches the maximum iteration number.

[0067] The present invention provides a catalytic cracking process mechanism modeling method. By analyzing the reaction mechanism of the catalytic cracking process with hydrogenated catalytic light cycle oil as the main feed, a reaction kinetic model is constructed. An improved Grey Wolf optimization algorithm is used to optimize and solve the complex and high-dimensional model parameter identification problem. Finally, a catalytic cracking process mechanism model for producing more high-octane gasoline and aromatics is established, achieving accurate prediction of the yield and properties of related products. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] The above and other features, properties and advantages of the present invention will become more apparent through the following description in conjunction with the accompanying drawings and embodiments, in which like reference numerals represent like features throughout, wherein:

[0069] Figure 1 A step diagram of a catalytic cracking process mechanism modeling method according to an embodiment of the present invention is disclosed;

[0070] Figure 2A schematic diagram of a process for catalytic cracking of light cycle oil to produce high-octane gasoline and aromatics is disclosed according to an embodiment of the present invention;

[0071] Figure 3 A schematic diagram of a reaction network for a catalytic cracking process for producing high-octane gasoline and aromatics according to an embodiment of the present invention is disclosed;

[0072] Figure 4 A flow chart of an improved grey wolf optimization algorithm according to an embodiment of the present invention is disclosed. DETAILED DESCRIPTION

[0073] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the invention and are not intended to limit the invention.

[0074] Aiming at the complex high-dimensional parameter identification problem in the mechanism model of the catalytic cracking process of hydrocatalytic light cycle oil, it is a feasible research direction to optimize and solve it through swarm intelligence optimization algorithm, study efficient global optimization search methods and strategies, and improve the model prediction accuracy.

[0075] Figure 1 The following discloses a step diagram of a catalytic cracking process mechanism modeling method according to an embodiment of the present invention, as shown in FIG. Figure 1 As shown, the present invention proposes a catalytic cracking process mechanism modeling method, which specifically includes the following steps:

[0076] Step S1, obtaining production device process data;

[0077] Step S2, dividing the reaction components according to the principle of kinetic similarity;

[0078] Step S3: setting basic reaction kinetic assumptions for the catalytic cracking process for model simplification;

[0079] Step S4: Based on the basic reaction kinetics assumptions of step S3, determine the reaction relationship of each reaction component in step S2 and construct a reaction network;

[0080] Step S5: establishing a reaction kinetics model based on reaction components, basic reaction kinetics assumptions and reaction network;

[0081] Step S6: Select an objective function, optimize the kinetic parameters involved in each reaction, and output the optimal kinetic parameter combination;

[0082] Step S7: Based on the optimal kinetic parameter combination, predict the yield and properties of related products.

[0083] This paper proposes a mechanistic modeling method for the catalytic cracking process, which produces high-octane gasoline and aromatics using hydrocatalytic light cycle oil as the primary feedstock. The method classifies the catalytic cracking process into 18 components, constructs a reaction network, and establishes a process mechanistic model based on reasonable assumptions. The method then uses an improved Gray Wolf optimization algorithm to identify model parameters and predict product yields and properties. Experimental verification shows that the model's predicted results deviate from actual values ​​by less than 2%, demonstrating the robustness of the proposed mechanistic model.

[0084] Figure 2 The schematic diagram of the process of catalytic cracking of light cycle oil to produce high-octane gasoline and aromatics according to one embodiment of the present invention is disclosed. Figure 2 As shown, the catalytic light cycle oil (catalytic diesel) is selectively hydrogenated and saturated in the hydrogenation unit 201, and the condensed ring aromatics are directed hydrogenated and saturated into single-ring aromatics, which greatly increases the potential components that can be cracked;

[0085] After being mixed with heavy crude oil, it enters the catalytic cracking unit 202 again to maximize the ring-opening cracking reaction and produce more high-octane gasoline or light aromatics.

[0086] The products of this process mainly include catalytic diesel (distillation range 200-350℃), stabilized gasoline (distillation range 40-200℃), light olefins (C 4＝ 、C 3＝ 、C 2＝ ), liquefied gas, dry gas and coke.

[0087] The following describes in detail the steps of the catalytic cracking process mechanism modeling method proposed by the present invention, using a catalytic cracking process for producing high-octane gasoline and aromatics as an example. It should be understood that within the scope of the present invention, the above-mentioned technical features of the present invention and the technical features described in detail below (e.g., in the Examples) may be combined and interrelated to form preferred technical solutions.

[0088] Step S1: Acquire process data of the production device.

[0089] Obtaining process data from production units containing information such as feed properties, catalyst properties, product yields, and operating conditions;

[0090] Feedstocks include hydrotreated catalytic diesel, recycled light oil (mainly catalytic diesel and catalytic gasoline), and heavy oil (mainly wax oil). Products include catalytic diesel, gasoline, butene, propylene, liquefied petroleum gas, ethylene, dry gas, and coke.

[0091] Depend on Figure 2 It can be seen that the feed is mainly hydrocatalytic light cycle oil, and also contains light dirty oil (mainly light diesel and gasoline) and heavy oil (mainly wax oil, which may not contain).

[0092] Products include catalytic diesel, stabilized gasoline, light olefins (C 4＝ 、C 3＝ 、C 2＝ ), liquefied gas, dry gas and coke.

[0093] Operating conditions include reaction temperature, reaction pressure, reactant-to-oil ratio, true weight hourly space velocity, etc.

[0094] Some feed property data (four components of heavy oil: SS, SA, SR, SP; diesel PONA: Dp, Do, Dn, Das, Dam; gasoline PONA: Gp, Go, Gn, Ga) and operating conditions (reaction temperature) are shown in Table 1.

[0095] Table 1

[0096]

[0097] Step S2: dividing the reaction components according to the principle of kinetic similarity.

[0098] The reaction systems were grouped and divided into reaction components according to the principle of similar kinetic characteristics.

[0099] In the present embodiment, component division is carried out in conjunction with boiling range and family composition, heavy oil in feed is divided according to four-component composition (SARA), because asphaltene and colloid content are relatively less and mainly produce coking reaction, be merged into one component, namely be divided into heavy oil saturates point (SS), heavy oil aromatic point (SA), heavy oil colloid+asphaltene (SR) three components.Because the content of monocyclic aromatics and polycyclic aromatics in this process diesel aromatics is important investigation index, be divided into separate component.The light olefins (butylene, propylene, ethylene) in gaseous product is important chemical raw material, is also divided into separate component, and all the other products are divided into liquefied gas (C3~C4), dry gas (C1~C2) by carbon number.

[0100] The diesel fraction (distillation range 200-350°C) and gasoline fraction (distillation range 40-200°C) are divided according to the PONA group composition;

[0101] Among them, according to Figure 2 The process mechanism characteristics shown in the figure further subdivide the aromatic components in the diesel fraction into monocyclic aromatic hydrocarbons and polycyclic aromatic hydrocarbons to investigate the changes in the diesel aromatic components before and after the reaction. That is, the gasoline and diesel fractions are divided into:

[0102] Diesel paraffins (Dp), diesel olefins (Do), diesel cycloalkanes (Dn), diesel monocyclic aromatics (Das), diesel polycyclic aromatics (Dam), gasoline paraffins (Gp), gasoline olefins (Go), gasoline cycloalkanes (Gn), gasoline aromatics (Ga); low carbon olefin gas butene (C 4＝ ), propylene (C 3＝ ), ethylene (C 2＝ ) are calculated as single components, the components in the liquefied gas product except butene and propylene are calculated as separate components (LPG), the components in the dry gas product except ethylene are calculated as separate components (DR), and coke is calculated as a separate component, for a total of 18 components.

[0103] Step S3: setting basic reaction kinetic assumptions for the catalytic cracking process for model simplification.

[0104] The basic assumptions of reaction kinetics for the catalytic cracking process to produce more high-octane gasoline and aromatics;

[0105] For the complex gas-solid heterogeneous catalytic reaction in the catalytic cracking process, the basic reaction kinetic assumptions include:

[0106] (1) All reactions are considered to be first-order irreversible reactions;

[0107] (2) Considering the reaction process from a macroscopic perspective, the reactions involved are considered as homogeneous reactions;

[0108] (3) The gas flow state in the riser reactor involved in the reaction is isothermal, gas phase, ideal plug flow, and the diffusion of material particles is ignored;

[0109] (4) The catalyst deactivation problem involved in the reaction is characterized by the time-varying deactivation of the catalyst, and it is assumed that all reactions are carried out on the same acidic active center, that is, the time-varying deactivation of the catalyst affects all reactions to the same extent, is not selective, and is only related to the residence time of the catalyst.

[0110] (5) The adsorption of heavy aromatics and basic nitrogen compounds in the feed involved in the reaction has a serious impact on the reaction results. The reaction rate is corrected using the empirical formulas for the adsorption deactivation of heavy aromatics and basic nitrogen.

[0111] (6) Assume that no coke is generated from the gas.

[0112] Step S4: Based on the basic reaction kinetics assumptions of step S3, the reaction relationship of each reaction component in step S2 is determined to construct a reaction network.

[0113] Determine the reaction relationship of each reaction component and construct a reaction network.

[0114] Figure 3A schematic diagram of a reaction network for a catalytic cracking process for producing high-octane gasoline and aromatics according to an embodiment of the present invention is disclosed. Figure 3 The reaction network shown is constructed based on the basic assumptions of the division of reaction components in step S2 and the kinetics of reaction components in step S3, and mainly includes reactions such as cracking, hydrogen transfer, isomerization, aromatization, and condensation.

[0115] The reaction network specifically includes:

[0116] The saturated fraction in the heavy raw material undergoes cracking reactions to produce diesel fraction, gasoline fraction, and gas. At the same time, coking reactions occur during the reaction process to produce coke components. The reactions of the aromatic fraction and the resinous asphaltene component are similar, with a total of 45 reactions.

[0117] The chain alkanes in the diesel fraction can be cracked to produce chain alkanes and olefins, gas and coke in the gasoline fraction; the diesel olefin components can undergo hydrogen transfer and alkylation reactions to produce diesel cycloalkanes and aromatics, and cracking to produce gasoline olefins, gas olefins and coke; diesel cycloalkanes can also undergo hydrogen transfer reactions to produce diesel aromatics, and cracking reactions to produce gasoline cycloalkanes and olefins, gas and coke; the monocyclic aromatics in diesel aromatics can be dehydrogenated to produce polycyclic aromatic hydrocarbons, and cracking reactions can produce chain alkanes, olefins and aromatics in the gasoline fraction, and at the same time cracking to produce gas and coke; a total of 45 reactions.

[0118] The paraffins and aromatics in the gasoline fraction can both undergo cracking to produce gas components and coke; gasoline olefins can be converted to gasoline cycloalkanes and aromatics, and cracked to produce gas olefins and coke; gasoline cycloalkanes can be converted to gasoline aromatics, gas olefins and coke; a total of 23 reactions.

[0119] Among the gas components, butene can react to produce propylene, ethylene and dry gas; propylene can react to produce ethylene and dry gas; liquefied gas can be converted into ethylene and dry gas, all of which are considered in the reaction network, for a total of 7 reactions.

[0120] In summary, the constructed reaction network contains a total of 120 different types of transformation reactions.

[0121] Step S5: establishing a reaction kinetics model based on reaction components, basic reaction kinetics assumptions and reaction network.

[0122] Considering the factors affecting catalyst deactivation, basic nitrogen and heavy aromatics adsorption, a reaction kinetic model was established;

[0123] Among them, the reaction rate expression of the reaction kinetics model is:

[0124]

[0125] Where a is the mass concentration of the reaction component, kg / kg, and

[0126] a=[SS SA SR D p D o D n D as D am G p G o G n G a C 4＝ C 3＝ LPG C 2＝ DR C];

[0127] K is the reaction rate constant matrix;

[0128] ρ is the gas density, kg / m 3 ;

[0129] S WH is the true weight hourly space velocity, h -1 ;

[0130] is the catalyst coking deactivation equation, and β is the catalyst coking deactivation factor, C C is the catalyst coke content, wt%, and M is the exponential constant;

[0131] f(A) is the adsorption deactivation equation for heavy aromatics, and k A It is the adsorption deactivation factor for heavy aromatics;

[0132] C A is the residual carbon content of the feed oil, m%;

[0133] f(N) is the alkaline nitrogen adsorption deactivation equation, and C N is the alkali nitrogen content of the raw oil, wt%, t C is the catalyst residence time, φ C / O is the reactant-oil ratio, kg / kg.

[0134] The meanings of the variables are shown in Table 2.

[0135] Table 2

[0136]

[0137] Furthermore, based on the contents in steps S2, S3 and S4, the reaction kinetics model constructed is as follows:

[0138]

[0139]

[0140]

[0141]

[0142]

[0143]

[0144]

[0145]

[0146]

[0147]

[0148]

[0149]

[0150]

[0151]

[0152]

[0153]

[0154]

[0155]

[0156] in,

[0157] k is the rate constant of the corresponding conversion reaction process, and all of them conform to the Arrhenius equation, that is,

[0158] k0 and Ea are the pre-exponential factor and activation energy of the corresponding conversion reaction process respectively.

[0159] T is the reaction temperature, and R is the molar gas constant, which is 8.3145 in the SI system.

[0160] From the reaction network constructed in step S4 and the reaction kinetic model constructed in step S5, it can be seen that there are 120 reactions involved in the model, and each reaction involves a reaction rate constant k, that is, 120 reaction rate constants, corresponding to 120 reaction pre-exponential factors and 120 reaction activation energies, a total of 240 parameters, k1~k 120are all reaction rate constants of the corresponding conversion reaction processes, and all conform to the Arrhenius equation.

[0161] Step S6: Select an objective function, optimize the kinetic parameters involved in each reaction, and output the optimal kinetic parameter combination.

[0162] The objective function for kinetic parameter identification is selected. In this embodiment, the minimum sum of squares of the errors between the model predicted value and the actual value is used as the objective function for parameter estimation. The objective function value is represented by F, and the kinetic parameters of each step are determined according to the optimal fitting principle.

[0163]

[0164] Among them, Y i_pre represents the model predicted value of the i-th component, percentage yield; Y i_real Represents the actual value of the i-th component, percentage yield.

[0165] In this embodiment, an improved Grey Wolf Optimization Algorithm is formed by combining elite reverse learning with differential evolution strategy to optimize and solve the kinetic parameters involved in each reaction. The improved Grey Wolf Optimization Algorithm is used to optimize and solve the kinetic parameters in the kinetic model constructed in steps S4 and S5.

[0166] Figure 4 The flowchart of the improved gray wolf optimization algorithm according to an embodiment of the present invention is disclosed. Figure 4 The improved grey wolf optimization algorithm shown in FIG1 introduces elite reverse learning and differential evolution strategies based on the classic grey wolf optimization algorithm to solve complex and high-dimensional dynamic parameter optimization problems.

[0167] The improved gray wolf optimization algorithm further comprises the following steps:

[0168] Step S601: Set the population size, parameter dimension, maximum number of iterations, parameter variable interval, mutation factor and crossover probability;

[0169] Set the population size (pop), parameter dimension (dim), maximum number of iterations (MaxIter), parameter variable interval (ub, lb), mutation factor (F) and crossover probability (cr).

[0170] Step S602: Initialize individuals in the population;

[0171] For each population, individuals in the population are initialized based on the adversarial search. The population initialization formula is:

[0172] X=(ub-lb)×rand+lb

[0173] Where rand is a random number in (0,1), X is all individuals in all populations, and individuals outside the variable interval (ub, lb) are classified as boundaries;

[0174] Step S603: Calculate the fitness value of the individuals in the population according to the objective function;

[0175] In this embodiment, the fitness calculation method is set as:

[0176] fitness=∑F(X);

[0177] That is, the sum of the objective function values ​​of the current population for all inputs is comprehensively evaluated as the fitness value of the individuals in the current population.

[0178] Step S604: Perform elite reverse learning;

[0179] The individual fitness is calculated according to the objective function, and a part of the elite individuals are taken at the same time. The elite reverse solution is obtained by reverse learning. The position of the individual with stronger fitness among the reverse solution and the current individuals is updated.

[0180] Step S605: Determine the position of the leader wolf in the wolf pack, calculate the relative distances between the leader wolf and other individuals in the wolf pack, and update the individual positions;

[0181] Step S606: Modify the current optimal solution based on the differential evolution operator;

[0182] Step S607: Determine whether the termination condition of the algorithm optimization is met. If the termination condition is not met, return to step S604 to perform the next round of iterative calculation. If the termination condition is met, the optimization ends and the current optimal solution is output.

[0183] Furthermore, step S604 includes: in the process of solving the improved gray wolf optimization algorithm, elite reverse learning is to screen elite individuals with high fitness values ​​when solving the reverse solution of the current solution to generate elite individual reverse solutions, which can help improve the convergence speed of the algorithm's early search and the optimization ability in the later stage.

[0184] In this embodiment, the top 30% of individuals in terms of individual fitness are selected to generate elite reverse solutions, thereby improving the algorithm's optimization capability.

[0185] The calculation formula of the elite reverse solution is:

[0186]

[0187] Where, X i,j represents the elite individual of the i-th population in the iterative calculation, x min j and x max jare the dynamic boundaries of the j-th dimension search space, η is the generalized coefficient, and takes a random number between (0,1).

[0188] Furthermore, step S605 includes: determining the position of the leading wolf (α, β, γ) in the wolf pack, and calculating the relative distance (Dα, Dβ, Dγ) between other individuals in the wolf pack and the leading wolf, and calculating and updating the next position of other individuals in the wolf pack based on the relative distance and approach direction to the leading wolf, simulating the surrounding, predation and other behaviors of the wolf pack in hunting.

[0189] The positions of the leading wolves (α, β, γ) are the best, second best, and second best individuals in the current iteration.

[0190] The formula for calculating the relative distance between the other individuals in the wolf pack and the leader wolf is:

[0191]

[0192] Among them, C is a random number in [0,2], X P is the target position, X is the current individual position, and t is the current iteration number;

[0193] The position update method for other individuals is to update the average position after integrating the relative distance and approach direction with the leader wolf (α, β, γ). The calculation formula is:

[0194]

[0195] Wherein, A=a(2r-1) is a coefficient vector;

[0196] is the convergence factor during the algorithm iteration process;

[0197] r is a random number in [0,1].

[0198] That is, in the process of parameter optimization, the process of iterative optimization of parameters for simulating the wolf pack's hunting behaviors such as encirclement and predation of prey can be expressed as:

[0199]

[0200]

[0201] Furthermore, the differential evolution strategy modification in step S606 further includes:

[0202] After this round of iterative calculation is completed, the current optimal solution is differentiated and mutated based on the differential evolution operator, and the best result is taken as the current optimal individual.

[0203] It can be understood that the differential evolution algorithm is formed by simulating the natural principle of "survival of the fittest" of organisms in nature, and includes three parts: mutation, crossover, and selection.

[0204] For high-dimensional, multi-peak local optimal solutions that may appear during the high-dimensional parameter optimization process in this embodiment, differential evolution can help the algorithm improve its ability to escape from the local optimum to a certain extent.

[0205] In the improved grey wolf optimization algorithm solution process, the differential evolution strategy correction in step S606 refers to performing mutation, crossover, and selection correction on the optimal solution in the current iteration. The specific operations are as follows:

[0206] 1) The mutation operation calculation formula is:

[0207] Best_pos'=Best_pos+F*X rand ;

[0208] Among them, Best_pos is the individual with the best current fitness (optimal solution), F is the variation factor, X rand Indicates differential calculation.

[0209] 2) Crossover, crossover operation, generates new offspring individuals according to the crossover probability, and the calculation formula is:

[0210]

[0211] Among them, cr is the crossover probability, r is a random number (0,1)

[0212] 3) Selection: Select the individual with better fitness from the individual p after mutation and crossover and Best_pos as the optimal solution for this iteration.

[0213] Furthermore, the optimization termination condition set in step S607 includes: the error between all components and the target yield is less than a first percentage or the current iteration number reaches the maximum iteration number.

[0214] Preferably, the first percentage is 1%.

[0215] The improved Grey Wolf optimization algorithm inputs include:

[0216] ① The relevant variables involved in the constructed kinetic model, including raw material properties, operating conditions, catalyst properties, and product distribution;

[0217] ②Algorithm parameters, including: population size (pop), parameter dimension (dim), maximum number of iterations (MaxIter), parameter variable interval (ub, lb), mutation factor (F) and crossover probability (cr).

[0218] By improving the Grey Wolf optimization algorithm for parameter identification, the yield and properties of related products are predicted and compared based on the optimal kinetic parameter combination.

[0219] The algorithm output is the optimal solution that meets the conditions for the end of iterative optimization, that is, the current optimal combination of dynamic parameters.

[0220] After repeated experiments, the best results were obtained. The optimization parameters set in this embodiment are shown in Table 3.

[0221] Table 3

[0222] parameter Numerical parameter Numerical Population (pop) 100 Factor of variation (F) 0.3 Parameter dimension (dim) 240 Crossover probability (Cr) 0.6 Maximum number of iterations (MaxIter) 200

[0223] In this embodiment, the optimal kinetic parameter combination results obtained through multiple optimizations are shown in Table 4.

[0224] Table 4

[0225]

[0226]

[0227]

[0228] Step S7: Based on the optimal kinetic parameter combination, predict the yield and properties of related products.

[0229] The product yield was predicted and compared and verified using the optimal kinetic parameter combination obtained in step S6. The results are shown in Table 5.

[0230] The deviation between the predicted value of the product yield by the mechanism model and the actual value does not exceed 2%;

[0231] According to the divided reaction components, the prediction accuracy of product properties is good;

[0232] Moreover, when examining different situations of whether the feed contains heavy oil, the model prediction results performed well. The constructed mechanism model has good predictive ability and good adaptability.

[0233] Table 5

[0234]

[0235]

[0236] The present invention provides a catalytic cracking process mechanism modeling method. It aims to solve the complex high-dimensional parameter identification problem in the catalytic cracking process mechanism model of hydrocatalytic light cycle oil by improving the Grey Wolf optimization algorithm, studying efficient global optimization search methods and strategies, and improving the model prediction accuracy. Through experimental verification, the deviation between the model prediction results and the actual values ​​is less than 2%, and the constructed mechanism model has good predictive ability.

[0237] Although the above methods are illustrated and described as a series of acts for simplicity of explanation, it is to be understood and appreciated that these methods are not limited by the order of the acts, as some acts may occur in a different order and / or concurrently with other acts from those illustrated and described herein or not illustrated and described herein but understandable to those skilled in the art according to one or more embodiments.

[0238] As used in this application and the claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not intended to refer to the singular but may include the plural. Generally speaking, the terms "comprises" and "include" only indicate the inclusion of the steps and elements specifically identified, and these steps and elements do not constitute an exclusive list. A method or apparatus may also include other steps or elements.

[0239] The above embodiments are provided to persons familiar with the art for implementing or using the present invention. Personnel familiar with the art may make various modifications or changes to the above embodiments without departing from the inventive concept of the present invention. Therefore, the scope of protection of the present invention is not limited to the above embodiments, but should be the maximum scope of the innovative features mentioned in the claims.

Claims

1. A catalytic cracking process mechanism modeling method, characterized in that: The catalytic cracking process is a catalytic cracking process for producing high-octane gasoline and aromatics using hydrogenated catalytic light cycle oil as the main raw material. The method comprises the following steps: Step S1, obtaining production device process data; Step S2: dividing the reaction components according to the principle of kinetic similarity, combining the distillation range and the family composition; Step S3: setting basic reaction kinetic assumptions for the catalytic cracking process for model simplification; Step S4: Based on the basic reaction kinetics assumptions of step S3, determine the reaction relationship of each reaction component in step S2 and construct a reaction network; Step S5: establishing a reaction kinetics model based on the reaction components, basic reaction kinetics assumptions, and the reaction network, taking into account the influencing factors of catalyst deactivation, basic nitrogen, and heavy aromatics adsorption; Step S6: selecting an objective function, using an improved Grey Wolf optimization algorithm to optimize and solve the kinetic parameters involved in each reaction, optimizing the kinetic parameters involved in each reaction, and outputting the optimal kinetic parameter combination; Step S7: Based on the optimal kinetic parameter combination, predict the yield and properties of related products.

2. The catalytic cracking process mechanism modeling method according to claim 1, characterized in that: The production unit process data of step S1 further includes feed properties, catalyst properties, product yield and operating conditions.

3. The catalytic cracking process mechanism modeling method according to claim 2, characterized in that: The feed further comprises hydrocatalytic light cycle oil and / or light dirty oil and / or heavy oil; The products further include catalytic diesel, stabilized gasoline, light olefins, liquefied gas, dry gas and coke; The operating conditions further include reaction temperature, reaction pressure, reactant-to-oil ratio and true weight hourly space velocity.

4. The catalytic cracking process mechanism modeling method according to claim 2, characterized in that: The component division in combination with the distillation range and the group composition in step S2 further includes: The heavy oil in the feed is divided into four components; The diesel fraction and gasoline fraction are divided according to the PONA group composition.

5. The catalytic cracking process mechanism modeling method according to claim 2, characterized in that: The basic assumptions of the reaction kinetics of step S3 further include: All reactions are first-order irreversible reactions; The reactions involved are treated as homogeneous reactions; The gas flow state in the riser reactor involved in the reaction is isothermal, gas phase, ideal plug flow, and the diffusion within the material particles is ignored; The catalyst deactivation problem involved in the reaction is characterized by the time-varying deactivation of the catalyst, and all reactions are carried out on the same acidic active sites; The reaction rate was corrected using empirical formulas for heavy aromatic hydrocarbon adsorption deactivation and alkaline nitrogen adsorption deactivation. The gas does not generate coke.

6. The catalytic cracking process mechanism modeling method according to claim 2, characterized in that: In step S5, the reaction rate expression of the reaction kinetics model is: Where a is the mass concentration vector of the reaction component; K is the reaction rate constant matrix; is the catalyst coking deactivation equation; f(A) is the adsorption deactivation equation for heavy aromatics; f(N) is the alkaline nitrogen adsorption deactivation equation.

7. The catalytic cracking process mechanism modeling method according to claim 1, characterized in that: The step S6 further includes: minimizing the sum of squares of errors between the model predicted value and the actual value as the objective function of parameter estimation.

8. The catalytic cracking process mechanism modeling method according to claim 1, characterized in that: In step S6, the improved gray wolf optimization algorithm further includes the following steps: Step S601: Set the population size, parameter dimension, maximum number of iterations, parameter variable interval, mutation factor and crossover probability; Step S602: Initialize individuals in the population; Step S603: Calculate the fitness value of the individuals in the population according to the objective function; Step S604: Perform elite reverse learning; Step S605: Determine the position of the leader wolf in the wolf pack, calculate the relative distances between the leader wolf and other individuals in the wolf pack, and update the individual positions; Step S606: Modify the current optimal solution based on the differential evolution operator; Step S607: Determine whether the termination condition of the algorithm optimization is met. If the termination condition is not met, return to step S604 to perform the next round of iterative calculation. If the termination condition is met, the optimization ends and the current optimal solution is output.

9. The catalytic cracking process mechanism modeling method according to claim 8, characterized in that: The step S604 further includes: calculating individual fitness according to the objective function, selecting a portion of elite individuals, obtaining an elite reverse solution using a reverse learning method, and updating the individual position by selecting the individual with the higher elite reverse solution and the current individual fitness; The calculation formula of the elite reverse solution is: Where, X i,j represents the elite individual of the i-th population in the iterative calculation, x min j and x max j are the dynamic boundaries of the j-th dimension search space, and η is the generalized coefficient.

10. The catalytic cracking process mechanism modeling method according to claim 8, characterized in that: The step S605 further includes: Determine the location of the leader of the pack and calculate the relative distances of the other individuals in the pack to the leader; Based on the relative distance and approach direction from the leader wolf, the next position of other individuals in the wolf pack is calculated and updated.

11. The catalytic cracking process mechanism modeling method according to claim 8, characterized in that: The step S606 further includes: performing mutation, crossover, and selection operations on the current optimal solution based on the differential evolution operator: The corresponding calculation formula for the mutation operation is: Best_pos'=Best_pos+F*X rand ; Among them, Best_pos is the individual with the best current fitness, F is the variation factor, X rand is the difference calculation; The crossover operation corresponds to the following calculation formula: Where cr is the crossover probability and r is a random number between (0,1); The selection operation is to select the individual with better fitness from the individual p after mutation and crossover and Best_pos as the optimal solution for this iteration.

12. The catalytic cracking process mechanism modeling method according to claim 8, characterized in that: The optimization termination condition set in step S607 includes: the error between all components and the target yield is less than a first percentage or the current iteration number reaches the maximum iteration number.

Images