Method and system for prediction of molecular level composition of petrochemical production streams based on multiple models

Abstract

The application relates to a method and system for predicting the molecular level composition of a petrochemical production stream based on multiple models, and relates to the technical field of oil refining and chemical molecular management, and comprises the following steps: determining the molecular composition of raw materials, splicing structural units based on a structural unit-bonding moment matrix framework, and constructing a molecular digital structure topology; combining a preset molecular database, a probability density function and a molecular property predictor to construct a molecular composition model; establishing a molecular level reaction network based on a catalytic reforming reaction mechanism, combining a reaction rate expression to construct a molecular level reaction kinetics model; constructing a reactor model according to mass, energy and momentum transfer equations; constructing inactivation and separation unit models, coupling multiple models in sequence, generating a full-process reaction-separation molecular level prediction model, and predicting the molecular level composition of any stream. The application solves the problems that traditional petrochemical production relies on measured results to adjust process parameters and has operation lag, and through prediction of the molecular composition of a stream, the lag can be eliminated, and the production control precision is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of refinery chemical molecular management, in particular to a method and system for predicting the molecular-level composition of petrochemical production streams based on multiple models. BACKGROUND

[0002] With the upgrading of the petrochemical industry to molecular-level management and intelligentization, accurately grasping the molecular composition of any stream in production has become a key technical requirement for device regulation and optimization.

[0003] Currently, traditional petrochemical production operation methods only rely on measurement results to feedback adjust process parameters, which has the problem of operation lag, not only leading to a long production adjustment period, but also easily causing the production of unqualified products, which not only reduces production efficiency and product qualification rate, but also increases the difficulty of process optimization, and is difficult to adapt to the fine and real-time regulation and control requirements of petrochemical production. SUMMARY

[0004] The present application provides a method and system for predicting the molecular-level composition of petrochemical production streams based on multiple models, which improves the status quo of traditional petrochemical production operations that only rely on measurement results to feedback adjust process parameters, resulting in operation lag, and thus a long production adjustment period or the production of unqualified products.

[0005] The embodiments of the present application disclose the following technical solutions:

[0006] In a first aspect, the embodiments of the present application provide a method for predicting the molecular-level composition of petrochemical production streams based on multiple models, which comprises:

[0007] According to a preset detection scheme, the molecular composition of raw materials in refinery chemical production is determined to obtain the molecular composition of raw materials, and the molecular composition of raw materials is spliced based on a structure unit-key electric matrix framework to construct a molecular digital structure topology;

[0008] According to a preset molecular database and the molecular digital structure topology, a molecular property prediction is performed in combination with a probability density function and a molecular property predictor to construct a molecular composition model;

[0009] A molecular-level reaction network is established based on a catalytic reforming reaction mechanism, and a molecular-level reaction kinetics model is constructed in combination with a reaction rate expression of the reaction system;

[0010] According to a mass transfer equation, an energy transfer equation and a momentum transfer equation, a reactor in refinery chemical production is simulated and modeled to construct a reactor model;

[0011] The inactivation model and the separation unit model are constructed, and the molecular composition model, the molecular level reaction kinetics model, the reactor model, the inactivation model and the separation unit model are coupled in sequence to generate a full-process reaction-separation molecular level prediction model for predicting the molecular level composition of any stream in oil refining and chemical production.

[0012] In a second aspect, the embodiments of the present application provide a multi-model-based prediction system for molecular level composition of streams in petrochemical production, which comprises:

[0013] A raw material molecule digitalization module is configured to determine raw materials in oil refining and chemical production according to a preset detection scheme to obtain raw material molecular composition, splice structural units based on a structural unit-bonding moment matrix framework to construct a molecular digitalization structure topology.

[0014] A molecular composition modeling module is configured to predict molecular properties based on a preset molecular database and the molecular digitalization structure topology, combine a probability density function and a molecular property predictor to construct a molecular composition model.

[0015] A reaction kinetics modeling module is configured to establish a molecular level reaction network based on a catalytic reforming reaction mechanism, and combine a reaction rate expression of a reaction system to construct a molecular level reaction kinetics model.

[0016] A reactor simulation modeling module is configured to simulate a reactor in oil refining and chemical production according to a mass transfer equation, an energy transfer equation and a momentum transfer equation to construct a reactor model.

[0017] A full-process coupling prediction module is configured to construct an inactivation model and a separation unit model, and couple the molecular composition model, the molecular level reaction kinetics model, the reactor model, the inactivation model and the separation unit model in sequence to generate a full-process reaction-separation molecular level prediction model for predicting the molecular level composition of any stream in oil refining and chemical production.

[0018] The one or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0019] The application provides a method and system for predicting molecular-level composition of a petrochemical production stream based on multiple models. The method includes the following steps: constructing a molecular composition model, a molecular-level reaction kinetics model, a reactor model, a deactivation model, and a separation unit model in sequence, and coupling them to realize accurate prediction of the molecular-level composition of any stream in oil refining and chemical industry. First, the molecular composition of the raw material is detected, and a molecular digital structure topology is constructed by using a structure unit-key electric matrix. A molecular composition model is constructed by combining a preset molecular database, a probability density function, and a molecular property predictor. Second, a molecular-level reaction network is constructed according to the catalytic reforming mechanism, and a molecular-level reaction kinetics model is constructed by combining a reaction rate expression. A reactor model is constructed by collecting structure parameters and operation parameters according to the mass, energy, and momentum transfer equations and the equation matching for the reactor type. A deactivation model is constructed according to the catalyst carbon deposition law. A separation unit model is constructed by combining the SRK / PR equation and the flash-distillation column algorithm. Finally, the molecular composition model, the molecular-level reaction kinetics model, the reactor model coupled with the deactivation model, and the separation unit model are coupled in sequence to generate a full-process reaction-separation molecular-level prediction model, which can be used to predict the molecular-level composition of any stream in oil refining and chemical industry.

[0020] The technical solution of the application integrates the steps of digital analysis of the molecular composition of raw materials, quantification of molecular-level reactions, reactor simulation modeling, catalyst deactivation correction, molecular prediction of separation units, and multiple model coupling. The problem of traditional petrochemical production stream prediction, which only focuses on macro components and ignores molecular-level conversion rules and relies on lagging measurement data to adjust the process, is solved. The full-process molecular-level composition is accurately tracked and predicted, the lagging nature of process adjustment is eliminated, and the accuracy of production control is improved. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort.

[0022] Figure 1 The flowchart of the method for predicting the molecular-level composition of a petrochemical production stream based on multiple models provided by the embodiments of the present application is shown in the figure.

[0023] Figure 2 The flowchart of the method for constructing a molecular composition model provided by the embodiments of the present application is shown in the figure.

[0024] Figure 3 The flowchart of the method for constructing a full-process reaction-separation molecular-level prediction model provided by the embodiments of the present application is shown in the figure.

[0025] Figure 4A structural schematic diagram of a prediction system for molecular level composition of petrochemical production streams based on multiple models provided by the embodiments of the present application.

[0026] In the drawings, the components represented by various reference numerals are described as follows:

[0027] Raw material molecular digitization module 01, molecular composition modeling module 02, reaction kinetics modeling module 03, reactor simulation modeling module 04, and whole-process coupling prediction module 05. DETAILED DESCRIPTION

[0028] The present application provides a prediction method and system for molecular level composition of petrochemical production streams based on multiple models, which is used to solve the technical problems that the conventional petrochemical production operation only relies on measurement results to feed back and adjust process parameters, there is operation lag, and then the production adjustment period is too long or unqualified products are produced in the prior art.

[0029] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0030] In the description of the present application, the terms "first", "second" are only used for description purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0031] In the description of the present application, the term "for example" is used to indicate "as an example, illustration or explanation". Any embodiment described as "for example" in the present application is not necessarily interpreted as more preferred or more advantageous than other embodiments. The following description is given in order to enable any person skilled in the art to implement and use the present application. In the following description, details are listed for the purpose of explanation. It should be understood that those skilled in the art can realize the present application without using these specific details. In other examples, well-known structures and processes will not be described in detail to avoid unnecessary details making the description of the present application obscure. Therefore, the present application is not intended to be limited to the shown embodiments, but is consistent with the broadest scope of principles and features disclosed in the present application.

[0032] Embodiment one, as shown in FIG. 1, a prediction system for molecular level composition of petrochemical production streams based on multiple models provided by the embodiments of the present application. Figure 1As shown, the present application provides a multi-model-based prediction method for the molecular-level composition of a petrochemical production stream, which comprises the following steps:

[0033] S110: Determine the raw material in the oil refining and chemical production according to the preset detection scheme to obtain the raw material molecular composition, splice the structural unit based on the structural unit-key electric matrix framework, and construct the molecular digital structure topology;

[0034] In the embodiment of the present application, in the scenario of molecular-level simulation of oil refining and chemical process, in order to accurately obtain the basic information of raw material molecules and realize standardized characterization, the molecular composition of the raw material is determined by multiple instruments, and structured splicing is performed, thereby providing reliable data support for subsequent molecular composition model construction and whole-process prediction.

[0035] Specifically, first, according to the characteristics of the oil refining and chemical raw material, a preset detection scheme with a gas chromatograph detector and an infrared spectrometer as the core is determined, and the raw material is repeatedly determined several times to ensure that the obtained data covers the true distribution of the raw material molecular composition.

[0036] Further, the several determination results are comprehensively evaluated, and after eliminating abnormal data, the comprehensive and accurate raw material molecular composition information is integrated.

[0037] Further, according to the structural unit theory, each type of molecule in the raw material molecular composition is disassembled into basic structural units such as carbon chains and ring structures, and then combined with the molecular chemical properties and the connection rules of the bond electric matrix to orderly splice the basic structural units, thereby forming multiple complete and actual chemical structure-compliant molecular structures, ensuring the authenticity and rationality of the molecular structure.

[0038] Finally, based on the structural unit-key electric matrix framework, the obtained multiple complete molecular structures are digitally converted, and the structural characteristics, bonding relationships and other information of the molecules are converted into a standardized digital matrix form to generate a molecular digital structure topology.

[0039] This step realizes the conversion of the raw material molecules from physical analysis to digital carrier through the technical scheme of "accurate determination-structured splicing-digital characterization", thereby providing unified and computable basic data for subsequent steps such as molecular property prediction and reaction network construction.

[0040] The step S110 in the method provided by the embodiment of the present application comprises:

[0041] The raw material in the oil refining and chemical production is determined several times by using a gas chromatograph detector and an infrared spectrometer, and the raw material molecular composition is obtained by comprehensively evaluating the several determination results;

[0042] According to the structure unit theory, each structure unit in the molecular composition of the raw material is spliced to form a complete molecule by combining chemical properties and bond electric moment matrix connection rules, and a plurality of complete molecular structures are obtained;

[0043] Based on the structure unit-bond electric moment matrix framework, the plurality of complete molecular structures are digitally represented to generate a molecular digital structure topology.

[0044] In the embodiments of the present application, in order to improve the accuracy of predicting the molecular level composition of petrochemical production streams, reliable raw material molecular basic data need to be obtained and standardized digital characterization is completed, which provides core data support for subsequent molecular composition model construction, reaction network building and whole process simulation.

[0045] Firstly, the molecular composition of the raw material in oil refining and chemical production is determined. Considering the complex molecular structure and diverse components of the oil refining and chemical raw material, single detection is easily affected by instrument error and operation environment fluctuation. Therefore, a detection scheme combining gas chromatography detector (GC-FID) and infrared spectrometer is adopted to determine the raw material three times to avoid accidental error of single detection.

[0046] Among them, GC-FID can accurately quantify the molar proportion of different hydrocarbon components in the raw material, and infrared spectrometer can assist in identifying the functional groups in the molecule, and the measurement data of the two form a complement.

[0047] Further, after the determination of the molecular composition of the raw material is completed, the three detection results obtained are subjected to outlier rejection, and then the arithmetic mean is calculated for comprehensive evaluation, and finally the molecular composition of the raw material covering alkanes, cycloalkanes, aromatics and other components is obtained.

[0048] Exemplarily, taking straight-run naphtha raw material detection as an example, in the three GC-FID detections, the first time, the molar fraction of n-pentane is 7.35%, methylcyclohexane is 4.98%, and toluene is 2.36%; the second time, n-pentane is 7.40%, methylcyclohexane is 4.95%, and toluene is 2.40%; the third time, n-pentane is 7.36%, methylcyclohexane is 4.95%, and toluene is 2.38%, the deviation of the three data is less than 1%, and there is no abnormal value.

[0049] Further, after calculating the arithmetic mean, the final molecular composition of the three components in the raw material is determined as n-pentane 7.37% ((7.35%+7.40%+7.36%) / 3=7.37%), methylcyclohexane 4.96% ((4.98%+4.95%+4.95%) / 3=4.96%), and toluene 2.38% ((2.36%+2.40%+2.38%) / 3=2.38%).

[0050] Meanwhile, combined with the detection results of the infrared spectrometer, the components such as 2,6-dimethyloctane and m-xylene in the raw material were identified, forming complete raw material molecular composition data covering alkanes, cycloalkanes and aromatics.

[0051] Further, according to the structural unit theory, each type of molecular composition in the above-mentioned raw material molecular composition is disassembled into a basic structural unit, for example, n-pentane (C5H 12 ) of the alkane type is disassembled into 5-CH2-carbon chain units and 2-CH3 end group units, and n-decane (C 10 H 22 ) is disassembled into 10 continuous-CH2-carbon chain units.

[0052] Further, combined with the molecular chemical properties and the bond electric moment matrix connection rule (i.e., the bond energy and charge distribution between different atoms are represented by matrix elements), the basic structural units are reasonably ordered and spliced according to the chemical bonding.

[0053] For example, 6-CH2-units are spliced into a six-membered ring, and then 1-CH3 unit is connected to any carbon atom of the six-membered ring to form the complete molecular structure of methylcyclohexane; 8-CH2-units and 2-CH3 units are spliced into the straight chain structure of n-decane.

[0054] Further, the splicing mode of the above-mentioned steps is repeated to obtain a plurality of complete molecular structures corresponding to all components in the raw material, while ensuring that each molecular structure conforms to the actual chemical bonding rule and has no unreasonable bonding.

[0055] Finally, the structural unit-bond electric moment matrix hybrid framework is introduced to realize the conversion of the complete molecular structure into a standardized digital form, providing an operable digital carrier for subsequent model calculation.

[0056] Specifically, first, a unique structural unit code is assigned to each complete molecular structure, and then a bond electric moment matrix is constructed, i.e., the atoms in the molecule are taken as the matrix rows / columns, and the matrix element values represent the bond type (1 for single bond, 2 for double bond) and the electronegativity difference between atoms.

[0057] Finally, through the method of “structural unit code + bond electric moment matrix construction”, the structural characteristics, bonding relationship and charge distribution of each complete molecule are converted into standardized digital codes and matrix forms, generating a molecular digital structure topology.

[0058] For example, for methylcyclohexane (C7H 14 ) in the raw material, first, assign the structural unit code: the six-membered ring structural unit code is “R6-01”, and the methyl substituent unit code is “M1-03”.

[0059] Further, a bond matrix is constructed, i.e. taking 7 C atoms (numbered C1-C7) and 14 H atoms (numbered H1-H14) in the molecule as matrix rows / columns, the corresponding matrix element values are all 1 (carbon-carbon single bond) between C1-C6 atoms due to the six-membered ring structure, the corresponding element value between C1 and C7 (methyl C atom) is 1 (carbon-carbon single bond), and the corresponding element value between each C atom and the connected H atom is 1 (carbon-hydrogen single bond), while the C-C electronegativity difference (0) and the C-H electronegativity difference (0.35) are labeled. 14 )Further, a bond matrix is constructed, i.e. taking 7 C atoms (numbered C1-C7) and 14 H atoms (numbered H1-H

[0060] Through the above steps, the ring structure characteristics, substituent position, bond type and electronegativity information of methylcyclohexane are all converted into digital codes and matrices to generate its exclusive molecular digital structure topology.

[0061] Similarly, the same operation is performed on all component molecules in n-pentane, toluene, 2,6-dimethyloctane and other raw materials, and finally a molecular digital structure topology set covering all raw material components is formed, providing accurate digital basic data for subsequent molecular composition model construction, reaction network matching and other steps.

[0062] S120: According to the preset molecular database and the molecular digital structure topology, the molecular property prediction is performed in combination with the probability density function and the molecular property predictor to construct a molecular composition model;

[0063] In the embodiment of the present application, in the scenario of predicting the molecular level composition of petrochemical production streams, in order to convert the molecular digital structure topology into accurate molecular composition data, the initial matching needs to be carried out relying on the preset molecular database, the probability density function is introduced to optimize the initial composition, and the molecular property predictor is used for correction and optimization, etc. steps, to solve the problems of initial matching deviation and insufficient composition prediction accuracy.

[0064] Specifically, first, the initial matching work is carried out relying on the preset molecular database. The preset molecular database covers the digital topology templates of multiple types of molecules such as alkanes, cycloalkanes and aromatic hydrocarbons commonly used in the field of oil refining and chemical industry, and each template is associated with the standard molecular composition information.

[0065] Further, the generated raw material molecular digital structure topology is taken as a retrieval condition, and the most suitable digital topology template is searched in the preset molecular database through a topology feature cosine similarity algorithm. When the feature similarity is ≥95%, the standard composition interval is associated and obtained, and the initial molecular composition is obtained after integration.

[0066] Further, in view of the matching deviation of the initial molecular composition, a Gaussian distribution function is introduced as a probability density function optimization, taking the mole fraction of the initial molecular composition as the mean value, combining the standard deviation calculated from the past multiple batches of detection data of the raw material to simulate the component distribution probability, and screening the high-probability fraction combination with a probability ≥99% to narrow the data fluctuation range.

[0067] Meanwhile, in the optimization process, the probability density function is corrected by using a molecular property predictor, a predicted property is obtained by inputting the candidate composition into the predictor, an error is calculated by comparing the predicted property with a measured property of the raw material, if the error is greater than or equal to 1%, the function parameters are adjusted reversely, the operation is repeated until the prediction error is less than 1%, and an optimized molecular composition is obtained and a molecular composition model is constructed.

[0068] The step realizes the conversion of the molecular composition from initial matching to accurate optimization by the technical scheme of “database matching-function optimization-predictor correction”, and provides basic data support for subsequent molecular-level reaction network construction and whole-process simulation.

[0069] As shown in FIG. 1, the method provided in the embodiment of the present application includes the following steps. Figure 2

[0070] Based on a preset molecular database, an initial molecular composition is obtained according to molecular digital structure topology matching;

[0071] The initial molecular composition is optimized by using a probability density function, and in the optimization process, the probability density function is corrected by using a molecular property predictor until the prediction error is less than 1%, an optimized molecular composition is obtained, and a molecular composition model is constructed, wherein the molecular property predictor is constructed based on deep learning and trained to convergence by using sample data.

[0072] In the embodiment of the present application, in order to convert the molecular digital structure topology into accurate molecular composition data that can support subsequent modeling, hierarchical processing of database matching, function optimization and predictor correction is required to solve the problems of deviation between the initial matching data and the actual raw material molecular distribution and difficulty in considering consistency of molecular macroscopic properties by simply function optimization, and finally a molecular composition model that meets the actual production is constructed.

[0073] Firstly, initial molecular composition matching based on a preset molecular database is carried out to provide a basic data framework for subsequent optimization.

[0074] The preset molecular database is a professional database constructed based on molecular research results and production measured data in the refining and chemical industry, and covers digital topology templates of common molecules such as alkanes, cycloalkanes and aromatic hydrocarbons in typical raw materials such as straight-run naphtha and catalytic cracking gasoline. Each template is associated with corresponding standard molecular composition information, including key parameters such as the conventional molar fraction range of the molecule in the same type of raw material and the proportion of characteristic functional groups.

[0075] ​Specifically, in the process of molecular composition matching, the obtained raw material molecular digital structure topology is taken as the core retrieval condition, and a topology characteristic cosine similarity algorithm is used for matching in the preset molecular database, that is, the characteristic similarity of the to-be-matched topology and each template in the database is calculated, and when the characteristic similarity is greater than or equal to 95%, it is determined that the template is matched, and the standard molecular composition information associated with the template is extracted.

[0076] If there are multiple templates with similar characteristic similarities (for example, the characteristic similarities are 96% and 95.5%, respectively), the basic attributes of the raw material are combined to further filter and finally determine a unique matched template.

[0077] Further, all matched molecules and their composition information are integrated to obtain an initial molecular composition, so as to clearly define the basic distribution range of each type of molecule in the raw material and provide an initial data boundary for subsequent probability density function optimization.

[0078] For example, taking straight-run naphtha as an example, after matching, the initial molecular composition includes n-pentane (molar fraction 7.2%-7.5%), n-hexane (molar fraction 5.1%-5.4%), n-heptane (molar fraction 3.8%-4.1%), cyclohexane (molar fraction 3.5%-3.8%), methylcyclohexane (molar fraction 4.8%-5.0%), benzene (molar fraction 1.2%-1.5%), and toluene (molar fraction 2.3%-2.5%). The composition range of all molecules corresponds to the standard information of the straight-run naphtha template in the preset molecular database, forming a complete initial molecular composition list of the raw material.

[0079] Further, the obtained initial molecular composition is optimized by using a probability density function to narrow the fluctuation range of the initial data and improve the rationality of the composition distribution.

[0080] Wherein, considering that the initial molecular composition is a conventional range value obtained based on the template matching of the preset molecular database, and the molecular distribution of different batches of raw materials in actual production has slight differences, for example, the n-pentane fraction of straight-run naphtha purchased at different times from the same refinery may fluctuate by 0.2%-0.3% due to different crude oil sources, the probability density function is used to simulate the distribution rule of the actual molecular composition.

[0081] Specifically, the Gaussian distribution function is selected as the probability density function, the mean value of the molar fraction of each molecule in the initial molecular composition is taken as the distribution center, the standard deviation is calculated by combining the detection data of multiple batches of the raw material in the past, and the molar fraction probability density curve of each molecule is constructed.

[0082] Furthermore, by analyzing the probability density curve of the mole fraction, a fraction interval with a probability ≥ 99% was selected. The molecular composition within this fraction interval not only conforms to the conventional rules of the database but also closely matches the fluctuation range of actual production, thus obtaining a preliminary set of optimized candidate molecular compositions.

[0083] For example, taking n-pentane and methylcyclohexane in the initial molecular composition of straight-run naphtha as examples, the average initial mole fraction of n-pentane is 7.35%, and the standard deviation calculated based on the data from 120 previous batches is 0.12%. In the constructed mole fraction probability density curve, the fraction range with a probability ≥99% is 7.35% ± 0.15% (i.e., 7.20%-7.50%). The average initial mole fraction of methylcyclohexane is 4.9%, with a standard deviation of 0.08%, corresponding to a fraction range with a probability ≥99% of 4.9% ± 0.1% (i.e., 4.80%-5.00%). After integrating the fraction ranges of all molecules, a preliminary optimized candidate molecular composition set is obtained, such as n-pentane 7.22%-7.48%, methylcyclohexane 4.81%-4.99%, and toluene 2.32%-2.48%, etc.

[0084] Furthermore, the probability density function is iteratively corrected using a molecular property predictor to ensure that the optimized molecular composition accurately reflects the macroscopic properties of the raw materials, thus avoiding the problem of reasonable composition but deviating properties.

[0085] Among them, the molecular property predictor is a prediction model built based on deep learning. By analyzing the input molecular composition data, it explores the nonlinear correlation between molecular composition and macroscopic properties, so as to output macroscopic property prediction values ​​that match the actual characteristics of the raw materials, and must meet the accuracy requirement of prediction error (|predicted value - measured value| / measured value × 100%) < 1%.

[0086] Specifically, the molecular property predictor is based on a fully connected neural network framework, which includes a three-level structure of "input layer-hidden layer-output layer".

[0087] The number of neurons in the input layer is consistent with the number of key molecular types in the raw material molecular composition, and is used to receive the standardized molecular mole fraction data. The hidden layer has 3 layers, each with 128 neurons, and uses the ReLU activation function to enhance the model's ability to extract complex correlation features through nonlinear transformation, so as to avoid the gradient vanishing problem during training.

[0088] Meanwhile, a Dropout layer (with a dropout probability of 0.2) is added after each hidden layer to prevent the model from overfitting; the number of neurons in the output layer corresponds to the types of macroscopic properties to be predicted, and a linear activation function is used to directly output the normalized macroscopic property prediction values, which are then restored to the actual physical quantity units through an inverse normalization operation.

[0089] In the model training stage, first, a training sample set needs to be constructed, that is, a plurality of groups of typical sample data of raw materials in the oil refining and chemical industry are collected, and each group of sample data includes detailed molecular composition and measured values of macroscopic properties.

[0090] Further, the sample data is preprocessed to eliminate abnormal samples, and then the sample set is divided into a training set and a validation set according to a ratio of 8:2, so as to ensure consistency of the two groups of data in terms of raw material type, molecular composition range and macroscopic property distribution, and to ensure the generalization ability of the model.

[0091] Meanwhile, in the training process, the root mean square error (RMSE) of the predicted properties and the measured properties is used as the loss function, the Adam optimizer is used to update the network parameters, the initial learning rate is set to 0.001, and the learning rate is attenuated to 1 / 2 of the previous one every 100 iterations, so as to balance the training speed and convergence accuracy.

[0092] In addition, the model performance is evaluated every 10 iterations using the validation set, and the RMSE trend is recorded. When the RMSE of the validation set decreases by less than 0.005 for 20 consecutive iterations, and the prediction error of the macroscopic properties of all samples is stable and less than 1.5%, the model is determined to be converged, the training is stopped, and a molecular property predictor that can be put into use is obtained.

[0093] For example, the measured density of a group of straight-run naphtha samples is 0.655 g / cm 3 , and the predicted value output by the molecular property predictor is stable between 0.654-0.656 g / cm 3 , with an error of less than 0.3%, which meets the prediction accuracy standard.

[0094] Further, after the construction of the molecular property predictor is completed, the probability density function is iteratively corrected based on the molecular property predictor, so as to ensure that the optimized molecular composition not only meets the probability density function filtered rate interval with a probability of greater than or equal to 99%, but also accurately matches the macroscopic properties of the raw materials.

[0095] Specifically, first, a representative composition data is randomly selected from the candidate molecular composition set obtained by optimizing the probability density function, for example, in straight-run naphtha, the combination of n-pentane 7.36%, methylcyclohexane 4.92%, toluene 2.38% and octane 3.25%, is input into the molecular property predictor after standardization, and the macroscopic property prediction result corresponding to the composition is obtained, for example, the predicted density is 0.650 g / cm 3 , the normal boiling point is 82.2°C, and the refractive index is 1.423.

[0096] Further, the laboratory measured macroscopic property data of the batch of straight-run naphtha raw materials is retrieved, for example, the density measured by a density meter at 20°C is 0.655 g / cm 3, the normal boiling point range is 83.0-83.8℃ (average 83.4℃) measured by distillation apparatus, and the refractive index is 1.426 measured by refractometer.

[0097] At the same time, the prediction error of each macroscopic property is calculated: the density error is |0.650-0.655| / 0.655x100%≈0.76%, the boiling point error is |82.2-83.4| / 83.4x100%≈1.44%, and the refractive index error is |1.423-1.426| / 1.426x100%≈0.21%.

[0098] Among them, since the boiling point error (1.44%) is greater than 1%, the probability density function parameters need to be adjusted in reverse to correct the deviation.

[0099] Specifically, combined with molecular property analysis, the boiling point of the raw material is mainly positively correlated with the mole fraction of high carbon number alkanes, and the predicted boiling point is lower. It is speculated that the fraction of high carbon number alkanes in the candidate composition is lower than the actual value.

[0100] Therefore, the mean of the probability density function distribution of octane is increased from 3.2% to 3.4%, and the standard deviation is expanded from 0.1% to 0.12%. At the same time, the mean of the distribution of nonane is fine-tuned (from 2.1% to 2.2%), and the optimized candidate molecular composition set is regenerated.

[0101] Further, new composition data (n-pentane 7.37%, methylcyclohexane 4.95%, toluene 2.37%, octane 3.38%) are selected from the adjusted candidate molecular composition set and input into the molecular property predictor again to obtain new prediction results: density 0.653 g / cm 3 , boiling point 83.1℃, refractive index 1.425.

[0102] Similarly, the prediction results are compared with the measured values to calculate the errors: density error 0.31% (|0.653-0.655| / 0.655x100%≈0.31%), boiling point error 0.36% (|83.1-83.4| / 83.4x100%≈0.36%), and refractive index error 0.07% (|1.425-1.426| / 1.426x100%≈0.07%). All molecular property errors are less than 1%, so it is determined that the probability density function correction is complete.

[0103] Finally, the molecular composition (n-pentane 7.37%, methylcyclohexane 4.95%, toluene 2.37%, octane 3.38%, etc.) that meets the error requirement after this correction is determined as the final optimized molecular composition, and the optimized fraction of all molecules is integrated to complete the construction of the molecular composition model. This model not only meets the matching requirements of molecular digital structure topology, but also is highly consistent with the actual macroscopic properties of the raw material, and can be used for subsequent molecular-level reaction network construction and whole-process stream prediction.

[0104] S130: Establish a molecular-level reaction network based on the catalytic reforming reaction mechanism, and construct a molecular-level reaction kinetics model in combination with the reaction rate expression of the reaction system;

[0105] In the embodiments of the present application, in order to accurately describe the reaction path and conversion rule of the raw material molecules in the catalytic reforming process, the reaction mechanism needs to be converted into a calculable model framework through the steps of constructing a reaction rule library, building a molecular-level reaction network, and quantifying a kinetics model, thereby providing reaction simulation support for subsequent full-process stream molecular composition prediction.

[0106] Specifically, first, according to the catalytic reforming reaction mechanism in oil refining and chemical production, the reaction rule of the process is combed, and a reaction rule library containing several molecular-level reaction rules is constructed.

[0107] Among them, the catalytic reforming reaction mechanism includes typical reaction types such as dehydrogenation cyclization, isomerization, and hydrocracking. Each reaction rule needs to accurately match the molecular structure characteristics and reaction condition requirements to ensure that the reaction rule can cover the core related elements of molecular reaction.

[0108] Further, a molecular-level reaction network is established based on the constructed reaction rule library.

[0109] Specifically, first, the optimized molecular composition information in the model is extracted to determine the initial reactant set in combination with the obtained molecular composition model, and each molecule in the set is matched according to the reaction rules in the reaction rule library, and the corresponding products are determined when the reaction rules are met.

[0110] Further, the reactants and products are associated and integrated as nodes, and the reaction paths corresponding to the reaction rules are matched as edges to form a complete molecular-level reaction network.

[0111] At the same time, the molecular composition model and the molecular-level reaction network are coupled. After obtaining the molecular composition information through the molecular composition model, each molecule in the information will be matched for reaction according to the rules in the reaction rule library. If the subsequent raw material molecular composition is updated, the molecular-level reaction network can adjust the initial node parameters synchronously, and dynamically optimize the path allocation.

[0112] Finally, based on the reaction rate expression of the reaction system and the constructed molecular-level reaction network, a molecular-level reaction kinetics model is constructed under the condition of computer assistance, taking the structure unit-bond electric matrix framework as the benchmark, so as to provide quantitative support for full-process stream prediction based on the conversion relationship between the molecular reactants and products in the molecular-level reaction network.

[0113] The step S130 in the method provided in the embodiments of the present application includes:

[0114] According to a catalytic reforming reaction mechanism in oil refining and chemical production, a reaction rule base is constructed, wherein the reaction rule base comprises a plurality of reaction rules established based on a molecular level reaction mechanism;

[0115] A molecular level reaction network is established based on the reaction rule base, wherein the molecular composition model and the molecular level reaction network are coupled, after obtaining molecular composition information through the molecular composition model, each molecule in the molecular composition information is matched according to the reaction rules in the reaction rule base, and reacts according to the reaction rules when the reaction rules are satisfied;

[0116] Based on the structure unit-bond electric moment matrix framework, a molecular level reaction kinetics model is constructed under the condition of computer assistance based on a reaction rate expression of a reaction system and the molecular level reaction network, wherein the molecular level reaction kinetics model is used to quantify the conversion relationship of molecular reactants and products in the molecular level reaction network.

[0117] In the embodiments of the present application, in order to accurately simulate the reaction path and conversion rule of the raw material molecules in the catalytic reforming process, a molecular level reaction kinetics model needs to be constructed based on the catalytic reforming reaction mechanism in oil refining and chemical production, and then the abstract reaction mechanism is converted into a calculable model system.

[0118] Firstly, based on the catalytic reforming reaction mechanism in oil refining and chemical production, a reaction rule base is constructed to provide clear rule basis for molecular reaction matching.

[0119] Among them, the catalytic reforming reaction mechanism in oil refining and chemical production takes the molecular structure to determine the reaction activity as the core, covers typical reaction types such as dehydrogenation cyclization, isomerization, hydrocracking, dehydrogenation aromatization, needs to extract reaction rules based on molecular level for the molecular action rule of each type of reaction, and form a rule set covering the main reaction path.

[0120] Exemplarily, the dehydrogenation cyclization reaction rule is clear, straight-chain alkanes need to satisfy the carbon chain length ≥6 carbon atoms, no quaternary carbon atoms, and under the conditions of 380-420℃, 1.0-1.5MPa, platinum-rhenium bimetallic catalyst, can be converted into monocyclic aromatic hydrocarbons with the same carbon number; the isomerization reaction rule provides that straight-chain alkanes (carbon chain ≥5C) can be isomerized into isomers under the conditions of 350-390℃, 1.2-1.6MPa, only the branching degree of the molecular structure is changed, and the carbon number is not changed.

[0121] Further, by systematically analyzing the molecular action rules of dehydrogenation cyclization, isomerization, hydrocracking, dehydrogenation aromatization and other typical reactions, the molecular structure characteristic requirements, reaction condition threshold values and product correlation relationships of each type of reaction are extracted to construct the reaction rule base.

[0122] Further, based on the constructed reaction rule library, a molecular level reaction network is established and coupled with the molecular composition model to construct a dynamic response molecular conversion framework.

[0123] Specifically, the optimized molecular composition information output by the molecular composition model is first taken as the initial reactant node set of the reaction network to determine the molecular identity and initial molar fraction of each node.

[0124] Further, in the manner of "molecule-reaction rule matching", each initial reactant molecule is compared with all reaction rules in the reaction rule library.

[0125] Exemplarily, taking n-heptane molecule as an example, first, the dehydrogenation cyclization reaction rule is matched to check whether it meets the structural requirement of "carbon chain length ≥ 6, no quaternary carbon" and whether it is suitable for the preset reaction condition (temperature 380-420℃, etc.), and if it meets the requirement, it is determined that the molecule can generate toluene (C7-AR-001) through dehydrogenation cyclization reaction, and the reaction path is recorded.

[0126] Further, the isomerization reaction rule is matched to confirm that it can be isomerized into 2-methylhexane, 3-methylhexane and other branched alkanes, forming 2 isomerization reaction paths.

[0127] Finally, the hydrocracking reaction rule is matched, and because the carbon chain length of n-heptane is <10 C, it does not meet the structural requirement of hydrocracking reaction, so this type of reaction path is excluded, and finally n-heptane molecule forms 3 potential reaction paths.

[0128] Further, after repeating the above matching process for all initial reactant molecules, the correlation between "reactant molecule-reaction path-product molecule" is integrated.

[0129] Specifically, taking molecules as nodes (initial reactant nodes, product molecule nodes, and product molecules as intermediate reactant nodes if they can further match reaction rules), and taking reaction paths as directed edges (the direction of the edge is "reactant→product"), a complete molecular level reaction network including initial reactants, intermediate products and final products is built.

[0130] At the same time, the molecular composition model and the molecular level reaction network are deeply coupled to ensure that the change of the molecular composition information can drive the reaction network to adjust in real time.

[0131] Specifically, when the molecular composition model updates the molecular composition information due to the change of raw material batch, the molecular level reaction network will automatically update the molar fraction parameter of the initial reactant node, and re-execute the "molecule-reaction rule matching" process, and then dynamically adjust the molar fraction prediction trend of each intermediate product and final product node.

[0132] Further, when new feedstock molecular composition information is obtained through the molecular composition model, the molecular-level reaction network automatically adds the new molecule as an initial reactant node, matches the corresponding reaction rules, expands the reaction path, and realizes the dynamic linkage of molecular composition and molecular-level reaction network to ensure the real-time and accuracy of the reaction simulation.

[0133] For example, when the molar fraction of n-heptane output by the molecular composition model decreases from 5.2% to 4.8% and a new n-octane (C8 straight-chain alkane) molecule (molar fraction 1.3%) is added in a batch of straight-run naphtha feedstock, the molecular-level reaction network first synchronously updates the molar fraction of the n-heptane node to 4.8%, and after re-matching the rules, the path flow of the dehydrocyclization to generate toluene and isomerization to generate 2-methylhexane decreases with the decrease in the molar fraction, and the initial generation amount of the toluene node is reduced.

[0134] At the same time, the n-octane is added as a new initial reactant node, the dehydrocyclization rule is matched to generate ethylbenzene (C8 monocyclic aromatic hydrocarbon), and the isomerization rule is matched to generate 2-methylheptane (C8 branched-chain alkane), two new reaction paths of “n-octane→ethylbenzene” and “n-octane→2-methylheptane” are added, and the dynamic adaptation of the reaction network to the new molecular composition is realized.

[0135] Further, based on the structural unit-bond electric moment matrix framework, under the condition of computer assistance, a molecular-level reaction kinetics model for quantifying molecular conversion relationships is constructed by combining the reaction rate expression of the reaction system.

[0136] Specifically, first, the digital conversion of molecular information is completed based on the structural unit-bond electric moment matrix framework. That is, the structural characteristics of all node molecules in the molecular-level reaction network are converted into numerical parameters recognizable by the model: for alkane molecules, the number and connection mode of structural units such as methyl and methylene are extracted to generate structural unit codes; for naphthene and aromatic molecules, the ring structure type (such as six-membered ring and five-membered ring) and the position and number of side chains are additionally labeled.

[0137] At the same time, the bond electric moment matrix of each molecule is constructed, and the matrix elements record the electronegativity difference of each chemical bond in the molecule. Then, the structural unit code and the bond electric moment matrix parameter are imported into the computer simulation platform to realize the numerical characterization of the molecular structure.

[0138] Further, the reaction rate expression based on experimental data and reaction mechanism derivation is introduced, and these reaction rate expressions, the molecular information that has been numerically characterized, and the molecular-level reaction network are integrated in the computer simulation platform for rate calculation.

[0139] Specifically, according to the molecular parameters corresponding to the reaction path and the process conditions (temperature, pressure), the reaction rate of each path is calculated in real time, and then the conversion amount of the reactant to the product at different time nodes is quantified by solving the differential equation, and finally the molecular level reaction kinetics model is formed.

[0140] Exemplarily, for the dehydrocyclization reaction path of "n-octane -> ethylbenzene", the rate expression corresponding to the reaction is called first, and then the numerical parameters of n-octane and the process conditions are imported to calculate the reaction rate of the path as 0.0028 mol / (L·min).

[0141] At the same time, combined with the reaction time of 30 min, the conversion amount of n-octane to ethylbenzene through this path within 30 min is quantified as 0.084 mol / L by solving the differential equation, which corresponds to a 1.8% decrease in the molar fraction of n-octane and a 1.8% increase in the molar fraction of ethylbenzene. The molecular level reaction kinetics model formed finally can accurately output such quantitative conversion results, and realize the quantitative simulation of the conversion relationship in the molecular level reaction network.

[0142] S140: According to the mass transfer equation, the energy transfer equation and the momentum transfer equation, a reactor in oil refining and chemical production is simulated and modeled to construct a reactor model;

[0143] In the embodiments of the present application, in order to accurately simulate the mass, energy and momentum transfer processes in the oil refining and chemical reactor, and the real conversion environment of the reduced molecules in the reactor, the transfer equation needs to be adapted according to the type of the reactor to construct a reactor model that fits the actual industrial conditions, and to provide core reaction unit support for subsequent full-process molecular level prediction.

[0144] Specifically, when constructing the reactor model, the type of the reactor in oil refining and chemical production needs to be determined first.

[0145] In the method provided in the embodiments of the present application, the reactor in oil refining and chemical production is one of an axial fixed bed reactor, a radial fixed bed reactor and a radial moving bed reactor.

[0146] Among them, the material flow characteristics and transfer rules of different types of reactors are significantly different, and the appropriate mass transfer equation, energy transfer equation and momentum transfer equation need to be selected to ensure that the equation can accurately describe the transfer behavior in the reactor.

[0147] Exemplarily, the material in the radial fixed bed reactor diffuses along the radial direction, and the axial diffusion can be ignored; the material in the axial fixed bed reactor flows along the axial direction, and the axial mass transfer and heat transfer are the focus; the radial moving bed reactor also needs to consider the influence of catalyst movement on the transfer process.

[0148] Further, for selected reactor types, simulation modeling is conducted based on mass transfer equation, energy transfer equation and momentum transfer equation.

[0149] wherein the mass transfer equation is used to describe the diffusion of molecules, reaction consumption and generation process within the reactor. For a radial fixed-bed reactor, when the axial diffusion is ignored, the mass transfer equation can be expressed as:

[0150] ;

[0151] wherein, is the molar flow rate (mol / s) of component , reflecting the macroscopic rate of material flow; is the radial coordinate of the reactor (m), representing the spatial position of material flow; is the effective bed length of the reactor (m), indicating different positions in the radial direction; is the bulk density of catalyst (mol / m 3 ), quantifying the bulk of catalyst within the reactor per unit volume, reflecting the tightness of catalyst packing; is the reaction rate (mol / (m 3 ·s)) of component , provided by the molecular-level reaction kinetics model, representing the microscopic rate of molecular conversion.

[0152] This equation calculates the change of molar flow rate of component with radial position by differentiating the radial coordinate , accurately depicting the concentration distribution difference of material caused by reaction and diffusion within the radial fixed-bed reactor.

[0153] In addition, the energy transfer equation is used to describe the heat generation, transfer and temperature distribution within the reactor. For a radial fixed-bed reactor, the energy transfer equation is in the form of:

[0154] ;

[0155] wherein, is the temperature within the reactor (K), reflecting the energy state of the reaction system; is the molar reaction heat (kJ / mol) of component involved in the reaction, endothermic positive, exothermic negative, which needs to be calibrated through calorimetry experiment or thermodynamic database; is the molar heat capacity at constant pressure (kJ / (mol·K)) of component , representing the heat storage capacity of the material itself.

[0156] The equation simulates the radial temperature field distribution of the reactor by relating the heat release of the reaction (the sum of the enthalpy changes of molecular reactions) to the heat capacity of the material (the sum of the products of the isobaric specific heat capacity of the components and the flow rate) through the radial temperature change rate.

[0157] Furthermore, the momentum transfer equation is used to describe pressure changes and fluid flow resistance within the reactor. For a radially fixed-bed reactor, the momentum transfer equation can be expressed as:

[0158] ;

[0159] In the formula, The pressure inside the reactor (Pa) reflects the driving force of fluid flow. The catalyst bed porosity (dimensionless) represents the proportion of voids between catalyst particles. The fluid dynamic viscosity (Pa·s) reflects the fluid viscous resistance. The fluid velocity in the empty tower (m / s) is calculated from the material volumetric flow rate and the reactor cross-sectional area, reflecting the macroscopic rate of fluid flow. is the catalyst particle shape factor (dimensionless). The equivalent diameter (m) of the catalyst particles is determined by sieve analysis or laser particle size analyzer. Fluid density (kg / m³) 3 The properties of the material vary significantly with temperature and pressure, requiring calculation using equations of state.

[0160] This equation calculates the pressure loss caused by bed resistance during fluid flow using the radial pressure change rate, reflecting the influence of pressure distribution within the reactor on material flow and reaction contact time.

[0161] Furthermore, after completing the construction of the transfer equation, the reactor model needs to be solved and verified in combination with the actual structure and operating parameters of the reactor to ensure that the reactor model can accurately reproduce the operating state of the industrial reactor.

[0162] First, regarding structural parameters, for radial fixed-bed reactors, it is necessary to obtain the reactor inner diameter, outer diameter, effective bed radial thickness, and catalyst loading height; for axial fixed-bed reactors, the focus is on reactor length, inner diameter, and catalyst bed height; for radial moving-bed reactors, it is necessary to obtain catalyst movement rate, bed radial thickness, catalyst circulation path size, and particle circulation flow rate.

[0163] In addition, operating parameters include feed flow rate, feed temperature, feed pressure, catalyst activity parameters, and material composition information (molecular composition and content). These structural and operating parameters are used as inputs to the transfer equation, thus determining the accuracy of the reactor model output. For example, the feed flow rate affects the empty tower velocity. , and further change the momentum transfer and mass transfer process.

[0164] Further, after completing the transfer equation construction and parameter collection, the reactor model needs to be solved and verified to ensure that the operation state of the industrial reactor can be accurately reproduced.

[0165] Specifically, first, the finite difference method is used to convert the differential equations describing the mass, energy and momentum transfer in the reactor into calculable algebraic equations, and then the existing MATLAB and other tools are programmed to realize iterative solution, and the component molar flow distribution, temperature field distribution and pressure distribution data of different positions in the reactor model output are obtained.

[0166] Further, the industrial actual operation data corresponding to the reactor in oil refining and chemical production are collected, including the molecular composition of the materials at the inlet and outlet of the reactor, the inlet and outlet temperature, the inlet and outlet pressure, and the measured values of the temperature and pressure at the key monitoring points in the bed.

[0167] At the same time, the above parameters calculated by the reactor model are compared with the industrial measured data, if the deviation of the core indicators exceeds the preset threshold, the key parameters in the reactor model are adjusted reversely, the discrete solving and data comparison process are executed again, until the deviation between the calculation results of the reactor model and the industrial actual data is controlled within an acceptable range, and it is ensured that the reactor model can accurately reproduce the real operation state of the industrial reactor.

[0168] Through the above steps, the reactor model constructed can accurately simulate the mass, energy and momentum transfer process in different types of reactors, couple the micro rate of molecular reaction kinetics with the macro transfer process of the reactor, and output the key information such as the molecular composition, temperature and pressure of the post-reaction stream, which provides support for the accurate calculation of the subsequent deactivation model and separation unit model.

[0169] S150: build a deactivation model and a separation unit model, and sequentially connect and couple the molecular composition model, the molecular level reaction kinetics model, the reactor model, the deactivation model and the separation unit model to generate a full-process reaction-separation molecular level prediction model for molecular level composition prediction of any stream in oil refining and chemical production.

[0170] In the embodiments of the present application, in order to accurately reflect the influence of catalyst activity attenuation on the reaction in petrochemical production, and the distribution rule of molecules in the separation link, a specific sub-model needs to be constructed first, and then a full-process prediction ability is formed through ordered coupling of multiple models, which provides complete support for molecular level composition analysis of any stream.

[0171] Specifically, a deactivation model and a separation unit model were constructed separately. When constructing the deactivation model, based on the correlation expression between the catalyst bed position and the coke content, the correlation between the amount of coke deposited and the catalyst activity was used to accurately describe the decay law of catalyst activity over time or feed rate.

[0172] Secondly, when constructing the separation unit model, the structural characteristics of the oil refining and chemical separation unit are combined with the introduction of rigorous molecular thermodynamics and flash distillation tower algorithms. By calculating the gas-liquid phase equilibrium, the molecular orientation in each separation unit can be predicted.

[0173] Furthermore, after the two sub-models are constructed, the molecular composition model, molecular-level reaction kinetics model, reactor model, deactivation model, and separation unit model are connected and coupled according to the actual process sequence of "raw material-reaction-separation" in oil refining and chemical engineering.

[0174] Specifically, the optimized molecular composition output by the molecular composition model is used as the initial input, which is then passed to the molecular-level reaction kinetics model to quantify the molecular transformation relationship, the reactor model to simulate the reaction process, the deactivation model to correct the reaction deviation caused by the decay of catalyst activity, and finally the separation unit model to predict the distribution results of molecules in the separation process.

[0175] Finally, by integrating the coupling and data transfer relationships of the above multiple models, a molecular-level prediction model for the entire reaction-separation process is generated. This model can cover all stages of production and can predict the molecular-level composition of any stream in oil refining and chemical production, providing a data foundation for process optimization and product quality control.

[0176] As attached Figure 3 As shown, step S150 in the method provided in this application embodiment includes:

[0177] Based on the correlation expression between catalyst bed position and coke content, a deactivation model is constructed and coupled with the reactor model. The deactivation model is used to describe the decay law of catalyst activity with time or feed rate.

[0178] Based on the structural characteristics of separation units in oil refining and chemical production, a separation unit model is constructed by combining rigorous molecular thermodynamics and flash distillation column algorithms. The separation unit model is used to predict the molecular orientation in each separation unit by calculating the gas-liquid phase equilibrium.

[0179] In this embodiment of the application, in order to accurately simulate the dynamic impact of catalyst activity decay on the reaction process in petrochemical production, as well as the fine control of molecular distribution by the separation unit, it is necessary to construct specialized sub-models step by step and deeply couple them to correlate the molecular-level behavior of each link, so as to provide core support for the prediction of the molecular composition of the entire process stream.

[0180] In catalytic reactions of petrochemical production, the catalyst will gradually lose activity due to carbon deposition (coke deposition), which directly affects the reaction efficiency and product distribution. Therefore, a deactivation model needs to be built to reflect the activity decay law and cooperate with the reactor model to reduce the real reaction scenario.

[0181] Specifically, first, through industrial side-line experiments and laboratory fixed-bed evaluation, coke content data at different positions in the catalyst bed under different running times and feed loads are collected, combined with catalyst activity detection results, and a correlation expression of catalyst bed position and coke content is fitted to determine the distribution gradient of carbon deposition in the bed.

[0182] Further, based on the correlation expression, a deactivation model is built with time or feed amount as the independent variable, i.e., introducing an activity decay factor to correlate the quantitative relationship between coke deposition and catalyst activity, so that the model can output the activity coefficient of each region in the catalyst bed at any time or any feed amount.

[0183] Further, the deactivation model is coupled with the reactor model to realize the real-time linkage of catalyst activity dynamic decay and molecular conversion in the reactor during the reaction, making the output results of the reactor model more consistent with the actual working conditions of industrial production.

[0184] Specifically, when calculating the reaction kinetics, the reactor model calls the bed activity distribution data output by the deactivation model in real time to correct the reaction rate constant in different regions.

[0185] In catalytic reactions of oil refining and chemical industry, the reaction rate constant is the core parameter that determines the molecular conversion efficiency and product distribution, and its value is directly affected by the catalyst activity.

[0186] The higher the catalyst activity, the larger the reaction rate constant, and the faster the molecular conversion; otherwise, the conversion efficiency decreases. Through the coupling of the deactivation model and the reactor model, the reactor model can obtain the current catalyst activity data of each region and use it as a correction factor to integrate into the calculation process of the reaction rate constant, ensuring that the reaction rate calculation of each region matches the actual activity state of the catalyst in that region.

[0187] Further, when building the separation unit model, the actual structural characteristics of the separation device in oil refining and chemical production need to be considered to ensure that the model matches the industrial equipment well.

[0188] Specifically, first, the core structural parameters of the separation device are obtained, including the number of trays in the distillation column, the material and specifications of the packing, the number and position of the feed inlet, the heat exchange area of the overhead condenser and the bottom reboiler, or the effective volume of the flash tank, the structure of the feed distributor, etc. These parameters are used as the basic input of the separation unit model to build a model framework that matches the actual equipment.

[0189] Further, strict molecular thermodynamics and flash-distillation column algorithm are introduced to accurately calculate the gas-liquid phase equilibrium relationship in the separation unit, so as to ensure that the separation unit model can truly reflect the material separation effect of the industrial separation device.

[0190] In the method provided by the embodiment of the application, the strict molecular thermodynamics and flash-distillation column calculation algorithm adopts one or both of the SRK equation and the PR equation.

[0191] Specifically, for a separation system with a high content of light hydrocarbons, the SRK equation is preferentially used, which has higher calculation accuracy of gas-liquid phase equilibrium parameters of low-carbon alkanes and alkenes; for a separation system containing heavy hydrocarbons and polar components, the PR equation is selected, which can more accurately calculate the fugacity coefficient and phase equilibrium relationship of heavy components; if the separation system contains both light hydrocarbons and heavy hydrocarbons, the two equations can also be combined to calculate the gas-liquid phase equilibrium data of different component intervals in sections.

[0192] Further, in the actual operation of the flash-distillation column algorithm, the calculation needs to be carried out in stages in combination with the process logic of the separation device to realize the whole-process simulation from preliminary phase separation to fine molecular distribution.

[0193] For the flash distillation link, the mixed stream output from the reactor is introduced into the flash module, and the gas-liquid phase equilibrium parameters calculated based on the selected SRK equation or PR equation are used to simulate the rapid phase separation of the material at a set temperature and pressure.

[0194] For example, when the mixed stream enters the flash tank, light molecules are enriched in the gas phase due to their low boiling point and easy vaporization, and heavy molecules remain in the liquid phase due to their high boiling point and difficult vaporization, and the flash module outputs the gas phase and liquid phase streams after preliminary separation, providing pretreated material data for the subsequent distillation link.

[0195] For the distillation column link, the gas phase or liquid phase stream after flash distillation is used as input, and the structure parameters of the distillation column are combined to simulate the gas-liquid mass transfer and heat transfer process through the distillation column algorithm for each plate.

[0196] The distillation column algorithm is based on the plate-by-plate phase equilibrium data calculated by strict molecular thermodynamics, and iteratively calculates the gas phase composition of each plate in the overhead direction (the content of light components gradually increases with the increase of the plate height) and the liquid phase composition of each plate in the bottom direction (the content of heavy components gradually increases with the decrease of the plate height) from the feed plate, until the temperature, pressure and molecular composition of each plate in the column reach a stable state.

[0197] For example, if the light hydrocarbon gas phase stream (containing propane, butane and pentane) after flash distillation is processed, the SRK equation is used to calculate the gas-liquid phase equilibrium, the distillation column has 20 plates, the feed inlet is located at the 10th plate, and the reflux ratio is controlled at 2.5.

[0198] At the same time, the distillation column algorithm can simulate the enrichment of propane molecules in the gas phase at the top of the column (mole fraction of 98% or more), the enrichment of butane molecules in the middle of the column (mole fraction of 92%), and the retention of pentane molecules in the liquid phase at the bottom of the column (mole fraction of 95%), accurately reproducing the separation rule of "light components moving up and heavy components moving down" in the distillation column.

[0199] In addition, if the separation system contains both light hydrocarbons and heavy hydrocarbons, the SRK equation is called for the light hydrocarbon range (C5-C8), the PR equation is called for the heavy hydrocarbon range (C9-C 12 ), and the phase equilibrium data is calculated in sections to ensure the accuracy of the prediction of the molecular trajectory in the whole column.

[0200] Finally, through the deep integration of the above strict molecular thermodynamics and flash-distillation column algorithm, the separation unit model can accurately output the molecular-level composition of each separation stream, providing reliable separation data support for the coupling of the whole process model.

[0201] Further, after the deactivation model and the separation unit model are constructed, the actual process sequence of "raw material molecule analysis → reaction path quantification → conversion in the reactor → catalyst deactivation correction → product separation and purification" in oil refining and chemical production must be strictly followed, and the molecular composition model, the molecular-level reaction kinetics model, the reactor model, the deactivation model, and the separation unit model are sequentially connected and coupled to form a logically closed-loop whole-process reaction-separation molecular-level prediction model.

[0202] Specifically, first, the output of the molecular composition model is used as the initial input of the entire coupling process.

[0203] Among them, the molecular composition model is obtained by matching the pre-set molecular database, optimizing the probability density function, and correcting the deep learning molecular property predictor (error less than 1%), and the optimized molecular composition and corresponding molecular digital structure topology are directly transmitted to the molecular-level reaction kinetics model.

[0204] Further, the molecular-level reaction kinetics model is based on the reaction rule library constructed based on the catalytic reforming reaction mechanism, and the input molecular composition information is used to match and trigger reactions of each molecule according to the reaction rules, and the conversion relationship between the reactants and products is quantified by combining the reaction rate expression of the reaction system, and the molecular conversion path and preliminary reaction rate data are output.

[0205] Further, the preliminary reaction rate data is imported into the reactor model, and the reactor model simulates the mass transfer, heat transfer, and flow process in the reactor according to the mass transfer equation, energy transfer equation, and momentum transfer equation (adapted to axial fixed bed, radial fixed bed, or radial moving bed, etc. reactor types).

[0206] At this time, the deactivation model and the reactor model are in a coupled state, the deactivation model outputs the activity coefficient of each region of the catalyst bed in real time, the reaction rate constant of different regions in the reactor is dynamically corrected, and it is ensured that the composition of the mixed stream after reaction output by the reactor model can truly reflect the actual reaction result under the catalyst activity attenuation.

[0207] Finally, the mixed stream data after reaction is transmitted to the separation unit model, the separation unit model calculates the gas-liquid phase equilibrium and predicts the molecular trajectory based on the device structure framework constructed in advance, the SRK / PR equation and the flash-distillation column algorithm, and outputs the molecular level composition of each separation stream such as the overhead, side draw and bottom.

[0208] Finally, through the orderly coupling and data linkage of the above-mentioned multiple models, the whole-process reaction-separation molecular level prediction model can realize the coverage of the whole link of oil refining and chemical production, not only can trace back the molecular composition characteristics of the raw material stream, but also can accurately predict the molecular conversion dynamics of the reaction intermediate stream, and can determine the molecular level composition of the final separation product stream.

[0209] Through the specific implementation manner described above, the embodiments of the present application achieve the following technical effects:

[0210] The present application proposes a method for predicting the molecular level composition of petrochemical production streams based on multiple models. First, according to a preset detection scheme, a gas chromatograph and an infrared spectrometer are used to detect the oil refining and chemical raw materials multiple times, and the molecular composition of the raw materials is obtained by comprehensive evaluation. Then, based on the structure unit-key electric matrix framework, the structure unit is spliced and the digital structure topology of the molecule is generated. Next, the initial molecular composition is matched by combining the pre-set molecular database, and the molecular composition model is constructed by optimizing the probability density function and correcting the molecular property predictor. Then, according to the catalytic reforming reaction mechanism, the reaction rule library and the molecular level reaction network are constructed, and the molecular level reaction kinetics model is constructed based on the structure unit-key electric matrix framework and the reaction rate expression. Then, according to the mass, energy and momentum transfer equations, the equations are adapted for different reactors, the structure and operation parameters are collected and verified by solving, and the reactor model is constructed. At the same time, the deactivation model is constructed based on the catalyst bed carbon deposition law, and the separation unit model is constructed based on the SRK / PR equation and the flash-distillation column algorithm. Finally, the above models are sequentially coupled to generate a whole-process reaction-separation molecular level prediction model, and the molecular level composition of any stream is predicted.

[0211] The method provided by the embodiments of the present application solves the problem that the traditional petrochemical stream prediction only focuses on macro components, ignores the molecular level conversion rule, and depends on lagging measurement data to adjust process parameters, and cannot dynamically reflect the influence of catalyst deactivation on reaction efficiency, realizes the advance prediction of molecular level composition from raw materials to products in the whole link, eliminates the process adjustment hysteresis, and helps the petrochemical production to change to molecular level fine regulation and control.

[0212] Embodiment two, as shown in the attached Figure 4 Based on the inventive concept of the prediction method of the molecular level composition of the petrochemical production stream provided in embodiment one, the present application also provides a prediction system of the molecular level composition of the petrochemical production stream based on multiple models, which specifically comprises:

[0213] The raw material molecular digitalization module 01 is used to determine the raw material in oil refining and chemical production according to a preset detection scheme to obtain the molecular composition of the raw material, splice the structure units based on the structure unit-key electric matrix framework, and construct a molecular digital structure topology;

[0214] The molecular composition modeling module 02 is used to combine the probability density function and the molecular property predictor to predict the molecular property according to the preset molecular database and the molecular digital structure topology, and construct a molecular composition model;

[0215] The reaction kinetics modeling module 03 is used to establish a molecular level reaction network based on the catalytic reforming reaction mechanism, and combine the reaction rate expression of the reaction system to construct a molecular level reaction kinetics model;

[0216] The reactor simulation modeling module 04 is used to simulate and model the reactor in oil refining and chemical production according to the mass transfer equation, the energy transfer equation and the momentum transfer equation, and construct a reactor model;

[0217] The whole process coupling prediction module 05 is used to construct a deactivation model and a separation unit model, and connect and couple the molecular composition model, the molecular level reaction kinetics model, the reactor model, the deactivation model and the separation unit model in sequence to generate a whole process reaction-separation molecular level prediction model, and predict the molecular level composition of any stream in oil refining and chemical production.

[0218] In one embodiment, the raw material molecular digitalization module 01 is further used to:

[0219] The raw material in oil refining and chemical production is determined several times by using a gas chromatograph detector and an infrared spectrometer, and the molecular composition of the raw material is obtained by comprehensive evaluation based on the determination results; each structural unit in the molecular composition of the raw material is spliced according to the structural unit theory, and a complete molecule is formed by combining chemical properties and bond electric moment matrix connection rules to obtain a plurality of complete molecular structures; and the plurality of complete molecular structures are digitally represented based on the structural unit-bond electric moment matrix framework to generate a molecular digital structure topology.

[0220] In one embodiment, the molecular composition modeling module 02 is further used to:

[0221] Based on the preset molecular database, an initial molecular composition is obtained according to the molecular digital structure topology matching; the initial molecular composition is optimized by using a probability density function, and in the optimization process, the probability density function is corrected by using a molecular property predictor until the prediction error is less than 1%, to obtain an optimized molecular composition, thereby constructing a molecular composition model, wherein the molecular property predictor is constructed based on deep learning and trained to convergence by using sample data.

[0222] In one embodiment, the reaction kinetics modeling module 03 is further used to:

[0223] According to the catalytic reforming reaction mechanism in oil refining and chemical production, a reaction rule library is constructed, wherein the reaction rule library includes a plurality of reaction rules established based on the molecular level reaction mechanism; a molecular level reaction network is established based on the reaction rule library, wherein the molecular composition model and the molecular level reaction network are coupled, after the molecular composition information is obtained by the molecular composition model, each molecule in the molecular composition information will be matched according to the reaction rules in the reaction rule library, and will react according to the reaction rules when the reaction rules are satisfied; based on the reaction rate expression of the reaction system and the molecular level reaction network, a molecular level reaction kinetics model is constructed under the condition of computer assistance with the structural unit-bond electric moment matrix framework as a reference, wherein the molecular level reaction kinetics model is used to quantify the conversion relationship between the molecular reactants and products in the molecular level reaction network.

[0224] In one embodiment, the reactor simulation modeling module 04 further includes:

[0225] The reactor in the oil refining and chemical production is one of an axial fixed bed reactor, a radial fixed bed reactor, and a radial moving bed reactor.

[0226] In one embodiment, the full-process coupling prediction module 05 is further used to:

[0227] Based on the correlation expression of the catalyst bed position and the coke content, a deactivation model is constructed and coupled with the reactor model, wherein the deactivation model is used to describe the attenuation law of catalyst activity with time or feed amount; according to the structural characteristics of the separation device in the oil refining and chemical production, a separation unit model is constructed by combining strict molecular thermodynamics and flash-distillation column algorithm, wherein the separation unit model is used to realize the prediction of the molecular movement in each separation unit by calculating the gas-liquid phase equilibrium.

[0228] Further, the whole-process coupling prediction module 05 further comprises:

[0229] The strict molecular thermodynamics and flash-distillation column calculation algorithm adopts one or both of the SRK equation and the PR equation.

[0230] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. And the above describes the specific embodiments of the present application. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or can be advantageous.

[0231] The above only describes the preferred embodiments of the present application, and does not limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0232] The present application and the drawings are only exemplary descriptions of the present application, and are considered to cover any and all modifications, changes, combinations or equivalents within the scope of the present application. Obviously, those skilled in the art can make various modifications and changes to the present application without departing from the scope of the present application. Thus, if these modifications and changes of the present application belong to the scope of the present application and its equivalent technology, the present application intends to include these modifications and changes.

Claims

1. A method for predicting the molecular-level composition of petrochemical production streams based on multiple models, characterized in that, The methods include: According to the preset detection scheme, the raw materials in oil refining and chemical production are measured to obtain the molecular composition of the raw materials. Based on the structural unit-bond electric matrix framework, the molecular composition of the raw materials is spliced ​​into structural units to construct a molecular digital structure topology. Based on the pre-set molecular database and the digital molecular structure topology, molecular properties are predicted by combining the probability density function and the molecular property predictor, and a molecular composition model is constructed. A molecular-level reaction network was established based on the catalytic reforming reaction mechanism, and a molecular-level reaction kinetic model was constructed by combining the reaction rate expression of the reaction system. Based on the mass transfer equation, energy transfer equation, and momentum transfer equation, a reactor model is constructed for simulation modeling of reactors in oil refining and chemical production. A deactivation model and a separation unit model are constructed, and the molecular composition model, molecular-level reaction kinetics model, reactor model, deactivation model and separation unit model are connected and coupled in sequence to generate a full-process reaction-separation molecular-level prediction model, which can predict the molecular-level composition of any stream in oil refining and chemical production. Specifically, based on a pre-set molecular database and the digital molecular structure topology, molecular properties are predicted using a probability density function and a molecular property predictor, and a molecular composition model is constructed, including: Based on a pre-built molecular database, the initial molecular composition is obtained by topological matching of the digital molecular structure. The initial molecular composition is optimized using a probability density function, and during the optimization process, the probability density function is corrected using a molecular property predictor until the prediction error is less than 1%, thereby obtaining the optimized molecular composition and constructing a molecular composition model. The molecular property predictor is built based on deep learning and trained to convergence using sample data. Among them, a molecular-level reaction network is established based on the catalytic reforming reaction mechanism, and a molecular-level reaction kinetic model is constructed by combining the reaction rate expression of the reaction system, including: A reaction rule library is constructed based on the catalytic reforming reaction mechanism in oil refining and chemical production. The reaction rule library includes several reaction rules established based on molecular-level reaction mechanisms. A molecular-level reaction network is established based on the reaction rule base. The molecular composition model and the molecular-level reaction network are coupled. After obtaining molecular composition information through the molecular composition model, each molecule in the molecular composition information will be matched according to the reaction rules in the reaction rule base, and react according to the reaction rules when the reaction rules are met. Based on the structural unit-bond electrical matrix framework, a molecular-level reaction kinetic model is constructed under computer-aided conditions, based on the reaction rate expression of the reaction system and the molecular-level reaction network. The molecular-level reaction kinetic model is used to quantify the transformation relationship between molecular reactants and products in the molecular-level reaction network.

2. The method for predicting the molecular-level composition of petrochemical production streams based on multiple models according to claim 1, characterized in that, According to a preset detection scheme, the molecular composition of raw materials in oil refining and chemical production is obtained by measuring the raw material molecular composition. Based on the structural unit-bond electrical matrix framework, the molecular composition of the raw material is assembled into structural units to construct a digital molecular structure topology, including: The raw materials in oil refining and chemical production were measured several times using a gas chromatograph and an infrared spectrometer. The molecular composition of the raw materials was obtained by comprehensively evaluating the results of the several measurements. According to the structural unit theory, the structural units in the molecular composition of the raw materials are spliced ​​together, and combined with chemical properties and bond-electric matrix connection rules to form complete molecules, thereby obtaining multiple complete molecular structures. Based on the structural unit-bond electric matrix framework, the multiple complete molecular structures are digitally represented to generate a digital molecular structural topology.

3. The method for predicting the molecular-level composition of petrochemical production streams based on multiple models according to claim 1, characterized in that, The reactor used in the oil refining and chemical production is one of the following: axial fixed bed reactor, radial fixed bed reactor, or radial moving bed reactor.

4. The method for predicting the molecular-level composition of petrochemical production streams based on multiple models according to claim 1, characterized in that, Constructing the inactivation model and the separation unit model includes: Based on the correlation expression between catalyst bed position and coke content, a deactivation model is constructed and coupled with the reactor model. The deactivation model is used to describe the decay law of catalyst activity with time or feed rate. Based on the structural characteristics of separation units in oil refining and chemical production, a separation unit model is constructed by combining rigorous molecular thermodynamics and flash distillation column algorithms. The separation unit model is used to predict the molecular orientation in each separation unit by calculating the gas-liquid phase equilibrium.

5. The method for predicting the molecular-level composition of petrochemical production streams based on multiple models according to claim 4, characterized in that, The rigorous molecular thermodynamics and flash-distillation column calculation algorithm adopts one or both of the SRK equation and PR equation.

6. A multi-model-based prediction system for the molecular-level composition of petrochemical production streams, characterized in that, The system is used to execute the multi-model-based method for predicting the molecular-level composition of petrochemical production streams as described in any one of claims 1-5, and the system comprises: The raw material molecule digitization module is used to determine the raw material molecular composition in oil refining and chemical production according to a preset detection scheme. Based on the structural unit-bond electric matrix framework, the raw material molecular composition is spliced ​​into structural units to construct a molecular digital structure topology. The molecular composition modeling module is used to predict molecular properties and construct a molecular composition model based on a pre-set molecular database and the digital structure topology of the molecules, combined with a probability density function and a molecular property predictor. The reaction kinetics modeling module is used to establish molecular-level reaction networks based on the catalytic reforming reaction mechanism and to construct molecular-level reaction kinetic models by combining the reaction rate expressions of the reaction system. The reactor simulation modeling module is used to simulate and model reactors in oil refining and chemical production based on the mass transfer equation, energy transfer equation, and momentum transfer equation, and to build reactor models. The full-process coupled prediction module is used to construct the deactivation model and the separation unit model, and connect and couple the molecular composition model, molecular-level reaction kinetics model, reactor model, deactivation model and separation unit model in sequence to generate a full-process reaction-separation molecular-level prediction model, which can predict the molecular-level composition of any stream in oil refining and chemical production.

Images