A process simulation optimization method and system for a molecular grade atmospheric and vacuum distillation unit

Abstract

The present application relates to the technical field of crude oil molecular level physical analysis, and specifically discloses a process simulation optimization method and system for a molecular level crude oil atmospheric and vacuum distillation unit, which breaks through the limitations of traditional virtual component simulation by constructing a full distillation molecular library and a molecular level composition model, establishing accurate mapping of virtual components and molecules, and having the following advantages: 1) realizing two-way conversion between the molecular level and the macro level, improving the prediction accuracy of product molecular composition by accurately quantifying the mapping rules and the reliability of the mapping rules; 2) accurately predicting the macro properties and molecular composition of each side product by combining real-time data dynamic regulation, realizing process visualization, and enhancing the stability of the unit operation; 3) relying on a multi-dimensional process scheme matrix and parallel computing, efficiently screening the optimal scheme, taking into account the product yield and quality, and adapting to the downstream process requirements, which promotes the atmospheric and vacuum distillation unit to move from macro simulation to molecular level fine optimization, and significantly improves the process control efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of crude oil molecular level physical analysis, in particular to a molecular level atmospheric and vacuum distillation unit process simulation optimization method and system. BACKGROUND

[0002] The existing technology of molecular level atmospheric and vacuum distillation unit process simulation optimization has gradually realized industrial application. By integrating DCS and LIMS system data, using molecular level model correction, virtual component division and non-ideal cutting simulation methods, the simulation accuracy and production efficiency of the device separation process are improved. Main tools such as Aspen HYSYS (Aspen process simulation software) and domestic control technology model, combined with molecular dynamics, machine learning and real-time optimization algorithm, support molecular level reaction path analysis and closed-loop control, but still face challenges such as high computational complexity, high data acquisition cost and limited model generalization ability. The current development direction is AI-driven molecular design, digital twin integration and industrial standardization, which gradually becomes a key technical support for oil refining enterprises to reduce costs and increase efficiency.

[0003] For example, the Chinese invention patent with the publication number CN109423331B discloses a petroleum component separation system and its separation method. The first adsorption tower top outlet of the separation system is connected with a first evaporator, a first cooling tank and a rectifying tower in sequence, the first adsorption tower bottom outlet is connected with the material inlet of the lower part of the second adsorption tower, and the upper part of the first adsorption tower is further provided with a first steam purging inlet; the second adsorption tower top outlet is connected with a second evaporator, a second cooling tank and a rectifying tower in sequence, the second adsorption tower bottom outlet is connected with a third cooling tank, a third evaporator, an extractor and a rectifying tower in sequence, and the upper part of the second adsorption tower is further provided with a second steam purging inlet; the third evaporator is further provided with an inlet of a third solvent, and the extractor is further provided with an inlet of a fourth solvent.

[0004] For example, the patent number CN108279251B discloses a method and device for simulating the separation process of petroleum molecules, which comprises the following steps: (1) establishing a molecular library containing the molecules of the whole fraction of the petroleum sample, and predicting the properties of the molecules in the molecular library; (2) calculating and optimizing the content of the molecules in the molecular library, and establishing a composition model which can represent the molecular composition of the petroleum sample; (3) dividing the molecular composition of the petroleum sample in the composition model into several virtual components according to the properties of the molecules in the separation process, and saving the mapping rules of mutual conversion; (4) calculating the properties of the virtual components according to the molecular composition of the virtual components, and performing thermodynamic calculation on each virtual component as a pure component; (5) obtaining the molecular composition of the separation products in the separation process by backstepping the mapping rules. The scheme provided by the present application can provide the molecular level information of the petroleum sample and its fractions, and the real behavior of the molecules is described in the process simulation calculation. The above-mentioned technology at least has the following technical problems: crude oil is a complex mixture composed of tens of thousands of different molecules, and the molecular composition (boiling point, molecular weight, polarity, sulfur content, etc.) of different crude oils (such as high-sulfur crude oil, heavy crude oil, light crude oil, etc.) differs greatly, so it is difficult for a simple mapping rule to adapt to the molecular characteristics of all crude oils, resulting in unstable mapping accuracy. SUMMARY

[0005] In order to solve the technical problem of unstable mapping accuracy in the prior art, the embodiments of the present application provide a process simulation optimization method and system for a molecular grade atmospheric and vacuum distillation unit. The technical scheme is as follows:

[0006] On the one hand, a process simulation optimization method for a molecular grade atmospheric and vacuum distillation unit is provided, which comprises the following steps: step one, constructing a molecular library containing the molecules of the whole fraction of the petroleum sample, and predicting and optimizing the molecules in the molecular library, thereby establishing a molecular grade composition model, and based on the initialized mapping rule, mapping and converting the virtual components and the molecules in the molecular grade composition model with each other, thereby backstepping the molecular composition of the separation products in the crude oil separation process, monitoring and obtaining the process parameters of the mapping and conversion, and classifying the accuracy of the current mapping rule; step two, realizing the simulation of the mapping and conversion process through the atmospheric and vacuum distillation unit simulation model, thereby predicting the macroscopic properties and molecular grade composition of each side product of the actual atmospheric and vacuum distillation unit, and realizing the visualization and dynamic regulation of the working condition in combination with the real-time data of the actual atmospheric and vacuum distillation unit; step three, generating a multi-dimensional process scheme matrix through the atmospheric and vacuum distillation unit optimization model, and screening the target process scheme based on the parallel calculation of the strict mechanism model, thereby realizing the process simulation optimization of the molecular grade atmospheric and vacuum distillation unit.

[0007] In another aspect, a process simulation optimization system for a molecular atmospheric and vacuum distillation unit is provided, which comprises: a molecular library construction and mapping module, configured to construct a molecular library containing molecules of all distillation fractions of a petroleum sample, and to predict and optimize the molecules in the molecular library, thereby establishing a molecular composition model, and to map and convert virtual components and the molecules in the molecular composition model based on an initialized mapping rule, thereby back-calculating the molecular composition of the separated products in a crude oil separation process, monitoring and obtaining parameters in the mapping and conversion process, and classifying the accuracy of the current mapping rule; a simulation prediction and dynamic regulation module, configured to simulate the mapping and conversion process through an atmospheric and vacuum distillation unit simulation model, thereby predicting the macroscopic properties and molecular composition of each side product of the actual atmospheric and vacuum distillation unit, and realizing working condition visualization and dynamic regulation in combination with real-time data of the actual atmospheric and vacuum distillation unit; and a multi-scheme optimization and screening module, configured to generate a multi-dimensional process scheme matrix through an atmospheric and vacuum distillation unit optimization model, and to screen a target process scheme based on parallel calculation of a strict mechanism model, thereby realizing process simulation optimization of the molecular atmospheric and vacuum distillation unit.

[0008] The technical scheme provided by the embodiments of the present application has at least the following beneficial effects:

[0009] (1) The process simulation optimization method and system for a molecular atmospheric and vacuum distillation unit provided by the present application break through the limitations of traditional simulation based on virtual components, and realize accurate prediction and optimization of the whole chain from the molecular composition of crude oil to product properties through fine modeling and dynamic regulation at the molecular level. The core value lies in that: the composition model with bidirectional mapping of molecules and virtual components accurately displays the composition details of various molecules in the stream by mapping and converting the molecular composition of the stream before and after process simulation; and the yield and properties of the products are calculated based on a strict mechanism model, and after calibration by actual device operation data, the yield and properties of the products can be accurately predicted in time by pulling the current working condition, and different process schemes can be quickly predicted based thereon, thereby providing timely data support for process adjustment of the device. Ultimately, the macroscopic process parameters are associated with the microscopic characteristics at the molecular level, thereby providing a new technical path for intelligent and fine operation of the atmospheric and vacuum distillation unit.

[0010] (2) The construction of a full-fraction molecular library and a molecule-virtual component mapping system lays a data foundation for molecular simulation. The core value lies in that: the bidirectional mapping design of molecules and virtual components solves the problem that virtual components in traditional models cannot reflect the details of molecules, realizes molecular-level conversion of the composition of the stream before and after process simulation, and can accurately display the composition details of various molecules in the stream; and dynamic evaluation and classification (high / medium / low accuracy) of the mapping rule ensure the reliability of the conversion process, avoid distortion of the composition information due to mapping deviation, and guarantee the richness and accuracy of the molecular composition description from the source, thereby providing high-quality basic data for subsequent simulation and optimization.

[0011] (3) Through the combination of simulation model and real-time data, the dynamic visualization and precise control of the molecular grade atmospheric and vacuum distillation unit are realized. The core value lies in: relying on the strict mechanism model, the yield and macro properties (such as density, viscosity) of each side product can be accurately calculated and predicted based on the molecular composition, and after the calibration of the actual device operation data, the accurate prediction results can be directly output by pulling the current working condition, solving the problem of large lag or deviation of traditional simulation prediction; at the same time, the comparison and analysis of real-time data and simulation results can quickly identify the abnormal working condition, and the dynamic control is realized combined with the molecular composition details, so that the operator can intuitively master the influence of molecular level changes on the product.

[0012] (4) Through multi-dimensional process scheme optimization and parallel computing, the global optimization of the molecular grade atmospheric and vacuum distillation unit process is realized. The core value lies in: based on the high-precision prediction ability of the strict mechanism model, the candidate schemes in the multi-dimensional process scheme matrix can be simulated quickly, and the product yield, molecular composition and macro properties corresponding to each scheme can be accurately output, solving the problem of high cost and low efficiency of traditional trial and error method; at the same time, parallel computing technology speeds up the scheme screening process, and can provide timely and reliable data support for device process adjustment in a short time. BRIEF DESCRIPTION OF DRAWINGS

[0013] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. 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 creative labor.

[0014] Figure 1 is a flowchart of a process simulation optimization method of a molecular grade atmospheric and vacuum distillation unit provided by the embodiments of the present application;

[0015] Figure 2 is a structural schematic diagram of a process simulation optimization system of a molecular grade atmospheric and vacuum distillation unit provided by the embodiments of the present application;

[0016] Figure 3 is a simple diagram of a molecular level mapping method provided by the embodiments of the present application;

[0017] Figure 4 is a simple diagram of a flash evaporation process provided by the embodiments of the present application;

[0018] Figure 5 is a mathematical model diagram of an N-stage tower plate of a distillation column provided by the embodiments of the present application. DETAILED DESCRIPTION

[0019] The technical solutions in the present application will be described below with reference to the drawings.

[0020] In the embodiments of the present application, the words such as "example", "for example" are used to represent an example, illustration, or description. Any embodiment or design scheme described as "example" in the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. Rather, the word "example" is intended to present the concept in a specific manner. In addition, in the embodiments of the present application, the meaning expressed by "and / or" can be both, or can be one of the two.

[0021] In the embodiments of the present application, sometimes the subscript such as W1 may be written in the form of non-subscript such as W1, and the meanings expressed thereby are consistent when the difference is not emphasized.

[0022] In order to make the technical problems, technical schemes and advantages to be solved by the present application more clear, the following will be described in detail in combination with the drawings and specific embodiments.

[0023] The embodiments of the present application provide a process simulation optimization method of a molecular grade atmospheric and vacuum distillation unit, which can be realized by a process simulation optimization system of a molecular grade atmospheric and vacuum distillation unit, such as a process simulation optimization method flow chart of a molecular grade atmospheric and vacuum distillation unit as shown in the figure. Figure 1 The processing flow of the method can include the following steps: step one, constructing a molecular library containing molecules of full distillation fraction of a petroleum sample, and predicting and optimizing the molecules in the molecular library, thereby establishing a molecular grade composition model, based on the initialized mapping rule, mutually mapping and converting the virtual components and the molecules in the molecular grade composition model, thereby backstepping the molecular composition of the separated products in the crude oil separation process, monitoring and obtaining the mapping and conversion process parameters, and classifying the accuracy of the current mapping rule; step two, realizing simulation of the mapping and conversion process through an atmospheric and vacuum distillation unit simulation model, thereby predicting the macroscopic properties and molecular grade composition of each side product of the actual atmospheric and vacuum distillation unit, and simultaneously realizing working condition visualization and dynamic regulation and control in combination with real-time data of the actual atmospheric and vacuum distillation unit; step three, generating a multi-dimensional process scheme matrix through an atmospheric and vacuum distillation unit optimization model, and simultaneously screening out a target process scheme based on parallel calculation of a strict mechanism model, thereby realizing process simulation optimization of the molecular grade atmospheric and vacuum distillation unit.

[0024] It needs to be explained that the above-mentioned construction contains the molecular library of the molecules of the whole distillation section of the petroleum sample, refers to the way of coupling the structure-oriented lumping method and the bond electric moment matrix method to construct the molecular library containing the molecules of the whole distillation section of the petroleum sample, the total number of molecules is about 20,000, and the molecular types include paraffin, naphthene, aromatic hydrocarbon, mercaptan, sulfide, thiophene, pyridine, pyrrole, phenol and carboxylic acid; and the group contribution method is used to predict the properties of the molecules in the molecular library; the predicted properties include density, boiling point, critical temperature, critical pressure, critical volume, eccentric factor, etc.; based on the determination results of the crude oil sample, the content of the molecules in the above-mentioned molecular library is calculated and optimized, so that the properties of the molecular library are close to the macroscopic properties of the crude oil sample, thereby establishing a composition model that can represent the molecular composition of the petroleum sample, i.e. the molecular level composition model.

[0025] As Figure 3 The molecular level mapping method of the composition model provided by the embodiment of the present application is shown in the diagram, according to one or more properties of the molecules of the raw material in the separation process, the molecular composition of the raw material in the composition model is divided into several virtual components; at the same time, the mapping rules of the mutual transformation of the virtual components and the molecules in the composition model are saved; the properties of the virtual components are calculated according to the molecular composition of the virtual components; and the virtual components are used as pure components for thermodynamic calculation of the separation process; based on the obtained thermodynamic properties of the virtual components, the separation process simulation is carried out, and through the reverse mapping rule, the molecular set composition details of the separation products in the separation process are obtained.

[0026] Specifically, the accuracy of the current mapping rule is classified, and the specific classification process is as follows: first, the molecular level composition model mapping accuracy coefficient is obtained by analyzing the mapping transformation process parameters, and compared with the first classification threshold and the second classification threshold preset in the simulation database; the first classification threshold is used to distinguish the molecular level composition model mapping accuracy coefficient critical value of high-precision mapping and medium-precision mapping, and the second classification threshold is used to distinguish the molecular level composition model mapping accuracy coefficient critical value of medium-precision mapping and low-precision mapping.

[0027] When the molecular-level composition model mapping accuracy coefficient is not less than the first classification threshold, the accuracy of the current mapping rule is classified as high-precision mapping, and the core parameters of the current mapping rule are recorded as a reference template of the same type of crude oil sample; the core parameters include molecular library characteristic parameters, virtual component division parameters, mapping rule logic parameters, and verification and performance parameters; the reference template is stored in the form of a database entry, each entry contains the above four-dimensional parameters, and is associated with the corresponding crude oil type label (such as Middle East light crude oil), mapping rule version number and verification timestamp, facilitating retrieval and traceability; when a new crude oil enters the device, it is matched with the same type of crude oil label first (such as density), and the corresponding reference template is automatically retrieved; then based on the molecular library fine-tuning parameters (such as fine-tuning carbon number distribution range) of the new crude oil, the adaptive mapping rule is quickly generated without building from scratch.

[0028] When the molecular-level composition model mapping accuracy coefficient is between the first classification threshold and the second classification threshold, the accuracy of the current mapping rule is classified as medium-precision mapping, and the current mapping rule is optimized; when the molecular-level composition model mapping accuracy coefficient is less than the second classification threshold, the accuracy of the current mapping rule is classified as low-precision mapping, and the current mapping rule is adjusted; the molecular-level composition model mapping accuracy coefficient is reacquired and marked as the final value of the molecular-level composition model mapping accuracy, so as to determine whether to issue a warning for the current mapping rule.

[0029] In a specific embodiment, the present application constructs a full-fraction molecular library and a dynamic mapping system, lays a solid foundation for molecular-level simulation, effectively solves the problem of insufficient virtual component information through bidirectional mapping design, and ensures the lossless transmission of stream molecular composition details in the simulation process; the dynamic evaluation and classification (high / medium / low precision) mechanism of the mapping rule significantly improves the reliability of composition conversion and avoids mapping distortion, thereby ensuring the richness and accuracy of the basic data from the source.

[0030] Specifically, the current mapping rule is optimized, and the specific optimization process is as follows: taking the average of the first classification threshold and the second classification threshold, marking it as the mapping accuracy median, and comparing the molecular-level composition model mapping accuracy coefficient with the mapping accuracy median; the mapping accuracy median is used as an intermediate judgment reference for the molecular-level composition model mapping accuracy coefficient, for dividing different levels of mapping rule optimization range; the setting of this mapping accuracy median avoids over-optimization of high-precision mapping rules, and ensures targeted strengthening of medium and low-precision mapping rules, thereby ensuring optimization efficiency while improving the adaptability of dynamic adjustment of mapping rules.

[0031] When the mapping accuracy coefficient of the molecular-level composition model is not lower than the mapping accuracy median value, based on the mapping accuracy coefficient of the molecular-level composition model and the first classification threshold, a high-zone deviation value is obtained, and a single-level optimization is performed based on the high-zone deviation value, that is, the density weight increase value of the virtual component is directly matched based on the high-zone deviation value to improve the mapping accuracy of the virtual component density; the above-mentioned obtaining of the high-zone deviation value refers to subtracting the first classification threshold from the mapping accuracy coefficient of the molecular-level composition model; the above-mentioned directly matching of the density weight increase value of the virtual component based on the high-zone deviation value, the specific matching process is: simulating a preset high-zone deviation value and density weight increase value corresponding relationship table in the database, according to the calculated high-zone deviation value, the corresponding density weight increase value can be directly matched from the high-zone deviation value and density weight increase value corresponding relationship table; the obtained density weight increase value is added to the original density weight of the virtual component, and the adjusted virtual component density weight is obtained. In the mapping rule, the boiling point weight usually accounts for the highest proportion, which may cover up the subtle differences of density and other properties. By increasing the density weight, the influence weight of multiple properties can be balanced, the mapping deviation caused by single property dominance can be avoided, the matching of molecules and virtual components can be more comprehensive, and when the mapping accuracy coefficient of the molecular-level composition model is not lower than the mapping accuracy median value, it indicates that the current mapping rule is reliable as a whole, and complex adjustment is not needed. By directly matching the density weight increase value, the gap with the first classification threshold can be quickly narrowed, and the mapping accuracy coefficient of the sub-level composition model can be efficiently pushed into the high-precision interval without changing the core mapping logic.

[0032] When the mapping accuracy coefficient of the molecular-level composition model is lower than the mapping accuracy median value, double optimization is performed, and the specific process is: the first optimization refers to obtaining a low-zone deviation value based on the mapping accuracy coefficient of the molecular-level composition model and the second classification threshold, and matching a boiling range interval reduction coefficient based on the low-zone deviation value, so as to narrow the boiling range interval of the virtual component. The larger the low-zone deviation value is, the smaller the boiling range interval reduction coefficient is, and the narrower the virtual component boiling range is. This adjustment refines the division accuracy of the virtual component, so that the physical and chemical properties (such as boiling point) of the molecules in the same boiling range interval are closer, and the averaging error caused by the wide interval is reduced; at the same time, the mapping resolution is improved, the cross-classification of different boiling range molecules is avoided, the average properties of the virtual component more accurately represent the actual molecular characteristics, the consistency between the model predicted molecular composition and the real molecular distribution of the crude oil is significantly improved, and finally the mapping accuracy coefficient of the molecular-level composition model is directly increased. The second optimization refers to introducing a correction term of the group contribution method to refine the molecular density prediction model.

[0033] The low zone deviation value refers to the molecular level composition model mapping accuracy coefficient minus the second classification threshold value; the low zone deviation value is matched to obtain the boiling range interval reduction coefficient, and the specific matching process is as follows: the boiling range interval reduction coefficients corresponding to each low zone deviation value interval in the simulation database are stored, the obtained low zone deviation value is input into the simulation database, the simulation database can match the corresponding low zone deviation value interval, and then the boiling range interval reduction coefficient corresponding to the interval is the required reduction coefficient. The obtained boiling range interval reduction coefficient is multiplied by the original boiling range interval, and the obtained result is the boiling range interval that needs to be adjusted; the boiling range interval reduction coefficient is less than 1, indicating that the boiling range interval of the virtual component needs to be reduced by a certain proportion.

[0034] The correction term of the introduced group contribution method includes adjacent group interaction correction, ring structure tension correction, heteroatom and functional group synergy correction, and branched chain steric hindrance correction. This refinement significantly improves the density prediction accuracy of complex molecules (such as polycyclic aromatic hydrocarbons), making the density matching of molecules and virtual components more accurate, reducing the misclassification of molecules due to density prediction deviation, and forming synergy with the optimization of boiling range interval reduction to improve the molecular level composition model mapping accuracy coefficient.

[0035] Further, the current mapping rule is adjusted, and the specific adjustment process is as follows: based on the molecular level composition model mapping accuracy coefficient and the second classification threshold value, a low limit deviation value is obtained, and compared with a preset invalid deviation value in the simulation database; when the low limit deviation value is lower than the invalid deviation value, the division basis of the virtual component is redefined; when the low limit deviation value is not lower than the invalid deviation value, the mapping algorithm of the current mapping rule is replaced.

[0036] The low limit deviation value refers to the second classification threshold value minus the molecular level composition model mapping accuracy coefficient; the invalid deviation value is a critical threshold value preset in the simulation database, which is used to judge whether the molecular level composition model mapping rule is invalid. When the low limit deviation value is greater than or equal to the invalid deviation value, the molecular level composition model mapping rule is recorded as invalid, otherwise, it is not counted as invalid; the redefined division basis of the virtual component refers to integrating molecular structure characteristics (such as ring number) and macroscopic properties (such as viscosity) to construct a composite division standard, for example, taking "boiling range + ring number + sulfur content" as the new basis of diesel fraction virtual component, so that the virtual component more accurately corresponds to a specific molecular set, and the attribute interval of the virtual component is more in line with the real distribution law of crude oil molecules, providing a more scientific basic framework for the subsequent mapping rule, reducing the mapping deviation caused by classification logic defects from the source, especially suitable for molecular level simulation of heavy and complex crude oil fractions.

[0037] The mapping algorithm for replacing the current mapping rule specifically includes changing from linear matching to a nonlinear model (such as a neural network), introducing probabilistic matching (calculating the probability of molecules belonging to each virtual component), or using ensemble learning (integrating the results of multiple algorithms), optimizing the matching process of molecules and virtual components, improving the mapping accuracy under the existing division basis by upgrading the algorithm logic, solving the molecular classification deviation caused by the limitations of the algorithm (such as the traditional linear algorithm cannot capture the complex correlation between attributes), quickly improving the mapping efficiency and accuracy without changing the core definition of virtual components, and has low iteration cost and strong adaptability, which can flexibly cope with the changes in molecular characteristics of different crude oils.

[0038] Specifically, the mapping conversion process parameters are monitored and obtained, and the specific analysis process is as follows: molecular attribute distribution deviation factors, yield conservation deviation factors, and key molecular type proportion deviation factors are extracted from the mapping conversion process parameters as core evaluation parameters, the action intensity coefficients of each factor are preset in the simulation database to quantify their weight contribution values to the molecular-level composition model mapping accuracy coefficient, and finally a weighted average fusion algorithm is used to synthesize the molecular-level composition model mapping accuracy coefficient.

[0039] The above-mentioned molecular attribute distribution deviation factor represents the ratio of the molecular attribute distribution deviation value of the mapping conversion process to its defined value; the above-mentioned yield conservation deviation factor represents the ratio of the yield conservation deviation value of the mapping conversion process to its defined value; and the above-mentioned key molecular type proportion deviation factor represents the ratio of the key molecular type proportion deviation value of the mapping conversion process to its defined value.

[0040] The molecular-level composition model mapping accuracy coefficient represents the mutual mapping conversion accuracy between virtual components and molecules in the molecular-level composition model, and the specific evaluation method is as follows:

[0041] ;

[0042] ;

[0043] ;

[0044] ;

[0045] In the formula, MMAC is a molecular-level composition model mapping accuracy coefficient, MADF is a mapping conversion process molecular attribute distribution deviation factor, MAD is a mapping conversion process molecular attribute distribution deviation value, DMAD is a preset defined molecular attribute distribution deviation value in the simulation database, YCDF is a mapping conversion process yield conservation deviation factor, YCD is a mapping conversion process yield conservation deviation value, DYCD is a preset defined yield conservation deviation value in the simulation database, KTDF is a mapping conversion process key molecular type proportion deviation factor, KTD is a mapping conversion process key molecular type proportion deviation value, DKTD is a preset defined key molecular type proportion deviation value in the simulation database, gs is an action intensity coefficient corresponding to the preset molecular attribute distribution deviation factor in the simulation database, ge is an action intensity coefficient corresponding to the preset yield conservation deviation factor in the simulation database, and gt is an action intensity coefficient corresponding to the preset key molecular type proportion deviation factor in the simulation database.

[0046] It should be explained that the above-mentioned molecular attribute distribution deviation value is a measure of the deviation of the overall distribution of the molecular attributes of the virtual component after mapping conversion from the corresponding attribute distribution in the actual molecular-level composition, which is obtained by calculating the attribute distribution difference (such as KL divergence) between the actual molecules and the virtual component after mapping; the above-mentioned yield conservation deviation value reflects the deviation of the “total yield conservation” principle before and after mapping conversion, and in the molecular-level simulation, the total yield needs to be strictly conserved, and the yield conservation deviation value is the absolute difference between the total yield of the virtual component after mapping and the total yield of the actual molecular-level composition, which is obtained by calculating the absolute difference between the total yield before and after mapping; the above-mentioned key molecular type proportion deviation value measures the total proportion of the key molecular type (such as polycyclic aromatic hydrocarbon) in the virtual component after mapping from the total proportion of the corresponding key molecular type in the actual molecular-level composition, which is obtained by calculating the absolute difference between the total proportion of the key molecular type in the actual molecules and the virtual component after mapping.

[0047] The above-mentioned defined molecular attribute distribution deviation value represents the maximum value of the molecular attribute distribution deviation value within a specified range; the above-mentioned defined yield conservation deviation value represents the maximum value of the yield conservation deviation value within a specified range; and the above-mentioned defined key molecular type proportion deviation value represents the maximum value of the key molecular type proportion within a specified range.

[0048] If the molecular attribute distribution deviation value is large (the overall attribute distribution deviates from the actual value), it will lead to the misclassification of the key molecular type (which usually has specific attributes) into non-target virtual components, thereby increasing the key molecular type proportion deviation value; at the same time, the mismatch of the attribute distribution may be accompanied by errors in molecular counting or weight allocation, indirectly causing the total yield to be not conserved, thereby causing the yield conservation deviation value to rise.

[0049] The action intensity coefficient corresponding to the molecular property distribution deviation factor indicates that when the molecular property distribution deviation factor changes by a unit amplitude, the molecular-level composition model mapping accuracy coefficient will change by a corresponding amplitude. The action intensity coefficient corresponding to the yield conservation deviation factor indicates that when the yield conservation deviation factor changes by a unit amplitude, the molecular-level composition model mapping accuracy coefficient will change by a corresponding amplitude. The action intensity coefficient corresponding to the key molecular type proportion deviation factor indicates that when the key molecular type proportion deviation factor changes by a unit amplitude, the molecular-level composition model mapping accuracy coefficient will change by a corresponding amplitude. The simulation database stores the mapping relationship between the molecular property distribution deviation factor and the action intensity coefficient corresponding thereto, the mapping relationship between the yield conservation deviation factor and the action intensity coefficient corresponding thereto, and the mapping relationship between the key molecular type proportion deviation factor and the action intensity coefficient corresponding thereto. For example, the molecular property distribution deviation factor, the yield conservation deviation factor, and the key molecular type proportion deviation factor are input into the simulation database. The simulation database generates the action intensity coefficient corresponding to the molecular property distribution deviation factor, the action intensity coefficient corresponding to the yield conservation deviation factor, and the action intensity coefficient corresponding to the key molecular type proportion deviation factor based on the preset mapping rule, and the numerical range of each type of action intensity coefficient is strictly controlled between 0 and 1.

[0050] The greater the molecular property distribution deviation factor, the higher the proportion of the actual deviation of the molecular property distribution relative to the acceptable defined value, the more significant the deviation of the molecular property distribution in the mapping conversion process from the true situation, and the lower the molecular-level composition model mapping accuracy coefficient. The greater the yield conservation deviation factor, the higher the proportion of the actual deviation of the yield conservation relative to the defined value, the worse the consistency of the yield before and after mapping, and the lower the molecular-level composition model mapping accuracy coefficient. The greater the key molecular type proportion deviation factor, the higher the proportion of the actual deviation of the key molecular type proportion relative to the defined value, the more obvious the deviation of the proportion of the key molecule in the mapping result from the true situation, and the lower the molecular-level composition model mapping accuracy coefficient.

[0051] Specifically, whether to perform early warning on the current mapping rule is determined by comparing the molecular-level composition model mapping accuracy final value with the first classification threshold. When the molecular-level composition model mapping accuracy final value is not lower than the first classification threshold, it is determined that no early warning is performed on the current mapping rule, and the core parameters of the current mapping rule are recorded as a benchmark template of the same type of crude oil sample. The specific process of recording the core parameters of the current mapping rule as the benchmark template of the same type of crude oil sample is the same as that described above.

[0052] When the molecular level composition model mapping accurate final value is lower than the first classification threshold value, it is judged that the current mapping rule is warned, the key characteristics of the current crude oil and the mismatch mode of the mapping rule are recorded, and are marked as an adaptive blind area, and the target function adaptability of the optimization model is checked.

[0053] It needs to be explained that the above-mentioned warning of the current mapping rule means that the warning information is sent to authorized personnel in an information manner; the above-mentioned recording of the key characteristics of the current crude oil and the mismatch mode of the mapping rule means that the key characteristics of the current crude oil (such as the peak value of the sub-attribute distribution of the crude oil) are extracted, and the mismatch mode of the mapping rule is analyzed: for example, a certain type of high-sulfur crude oil cannot always reach the first classification threshold value through the existing mapping rule, which is marked as an adaptive blind area, and data for subsequent development of a targeted mapping rule (such as a "boiling point + sulfur content" double-factor grouping for high-sulfur crude oil) is accumulated to avoid repeated trial and error; the above-mentioned checking of the target function adaptability of the optimization model means that if the target (such as the maximum light oil yield) of the atmospheric and vacuum distillation unit optimization model conflicts with the mapping accuracy (such as excessive simplification of molecular grouping in pursuit of yield), a "mapping deviation penalty term" needs to be added to the target function, and when the mapping accuracy is lower than the first classification threshold value, the optimization weight is reduced to prioritize the basic accuracy of the model.

[0054] Further, the process simulation optimization of the molecular level atmospheric and vacuum distillation unit is realized, and the specific analysis process is as follows: the molecular level atmospheric and vacuum distillation unit includes a molecular level composition model, an atmospheric and vacuum distillation unit simulation model, and an atmospheric and vacuum distillation unit optimization model; the atmospheric and vacuum distillation unit simulation model, with the molecular level atmospheric and vacuum distillation separation model as the core, supplemented by a thermodynamic calculation module, a single equilibrium stage phase equilibrium module, and a distillation column unit model, together constitutes a complete separation process simulation system, which is used to realize the separation process of the actual atmospheric and vacuum distillation unit.

[0055] It needs to be explained that the above-mentioned thermodynamic calculation module means that a molecular level thermodynamic calculation module is developed based on a cubic equation of state; the SRK equation and the PR equation, which are selected and improved in the present application, are more successful, the SRK equation has higher accuracy in calculating gas-liquid equilibrium of pure hydrocarbons and hydrocarbon mixture systems, and the PR equation has higher accuracy in calculating molar volume of liquid phase, and the basic form of the equation is as follows:

[0056] SRK equation of state:

[0057] ;

[0058] ;

[0059] ;

[0060] ;

[0061] PR equation of state:

[0062] ;

[0063] ;

[0064] ;

[0065] ;

[0066] where a(T r ) is a dimensionless factor related to temperature, used to correct the variation of gravitational parameter with temperature, T r is the reduced temperature, whose expression is: T r = T / T c , P is the system pressure, R is the gas constant, T is the absolute temperature, V is the molar volume, a and b are the parameters of the state equation, which take into account the correction of molecular volume and intermolecular forces respectively, and the parameter values are only related to the properties of the substance, which are usually determined by the critical temperature T c , critical pressure p c and molar volume V, etc.; w is the eccentric factor, which is used to calculate the state equation parameters of the mixture corresponding to the state equation parameters of the pure substance by using the empirical van der Waals mixing rule when dealing with multi-component mixed systems. The form is simple and very common, and is mainly aimed at the cubic state equation, whose formula is as follows:

[0067] ;

[0068] ;

[0069] ;

[0070] ;

[0071] where a m , b m are the cubic state equation parameters of the mixture system, x i , x j are the mole fractions of a component in the mixture system, b i , b j are the volume parameters of a component in the mixture system, a i , a j are the gravitational parameters of a component in the mixture system, and k ij is the binary interaction parameter, which is obtained by regression from phase equilibrium experimental data, but since its value has low dependence on temperature and pressure, and is considered to be independent of composition, it does not need too much experimental data for the correlation of k ij .

[0072] The single-balance-stage phase balance module can realize accurate calculation of vapor-liquid balance of a complex molecular system, the flash evaporation process is a single-stage balance separation process, and the p-T isothermal flash evaporation calculation is a vapor-liquid balance calculation performed under specified temperature and pressure conditions, such as Figure 4 The flash evaporation process diagram provided by the embodiment of the present application is shown, the feed amount is F, and the composition is z i The feed flow of a mixed system of c components enters a flash evaporation tank with fixed temperature T and pressure P to perform flash evaporation, the vapor phase and the liquid phase system reach balance, and finally form a vapor phase product with flow V and composition y i , and a liquid phase product with flow L and composition x i , T F is the feed temperature, P F is the feed pressure, H V is the molar enthalpy of the gas phase mixture, and H L is the molar enthalpy of the liquid phase mixture. The balance flash evaporation process needs to satisfy the material balance equation and the phase balance equation at the same time, for the flash evaporation system, the total material balance equation is as follows:

[0073] ;

[0074] The material balance of component i is as follows:

[0075] ;

[0076] The definition of the vaporization fraction e in the flash evaporation process is e=V / F, and the material balance equation can be obtained by substituting e=V / F into the above equation:

[0077] ;

[0078] The phase balance of component i is as follows:

[0079] ;

[0080] Wherein, K i is a phase balance constant, which is a function of temperature and pressure.

[0081] The phase balance equation is substituted into the component material balance equation, and the definition of the vaporization fraction e can be obtained:

[0082] ;

[0083] ;

[0084] According to the molar fraction addition equation ∑x i =1, the following can be obtained:

[0085] ;

[0086] The isothermal flash module is built based on the PR state equation, and the Newton-Raphson method is used to solve the vaporization fraction in the isothermal flash calculation process of the vapor-liquid equilibrium, and further root iteration is performed, so that the vaporization fraction under the phase equilibrium condition can be obtained.

[0087] Based on the above-mentioned thermodynamic calculation module and the unit operation calculation method of the single balance stage phase equilibrium module, a molecular level distillation column unit model is built to realize the rigorous calculation of the vapor-liquid equilibrium of the multi-balance stage distillation column of a complex molecular system. The steady-state model of the distillation column unit device in the computer simulation process is composed of the MESH equations (material balance equation, phase equilibrium equation, mole fraction addition equation and heat balance equation) on each tray. The establishment of the four equation groups is derived from the balance stage model of the Nth layer of the distillation column tray as shown in Figure 5 .

[0088] As shown in Figure 5 , the N-stage tray mathematical model of the distillation column provided by the embodiment of the present application is shown in the figure, which shows the conversion process of the distillation column in the atmospheric and vacuum distillation device from the actual equipment to the simulation model. The core is to disassemble the complex industrial tower equipment into a standardized "stage (tray / theoretical stage)" model, which is convenient for process simulation and calculation. The left side is a schematic diagram of the actual distillation column equipment, which presents the distillation column (such as an atmospheric column and a vacuum column) in the atmospheric and vacuum distillation device, and labels key process elements such as "tray number, pressure distribution, feed information, and liquid phase transport". It represents the equipment form (such as the separation of gasoline, diesel and other side-line products in crude oil fractionation by using the boiling point difference of different components in the tower) for realizing the separation of mixtures through multiple trays in the actual industry. The right side is a "cascade" disassembly schematic diagram of the simulation model, which abstracts the actual tower equipment into an "N-stage" cascaded model unit. Each stage (1st stage, 2nd stage…Nth stage) corresponds to a theoretical separation level (which can be understood as a simplified tray function) in the tower, and labels material input, F1 to F n , which correspond to the tray feed, W1 to W n , which correspond to the tray gas phase side-line production, O1 to O n , which refer to the heat load of the corresponding tray, U1 to U n , which correspond to the tray liquid phase side-line production, inter-stage transfer (V1 to V n , which are the gas phase internal flows, L1 to L n , which are the liquid phase internal flows). Through this standardized disassembly, the separation effect of each stage is calculated based on thermodynamic equations (such as phase equilibrium), and finally the fractionation process of the whole tower is reproduced.

[0089] For any theoretical plate of the distillation column, when the system reaches chemical equilibrium, it is generally believed that the mixed system of vapor-liquid two phases can be completely separated. Based on the calculation principle of multi-component and multi-stage separation of distillation process, a simulation model of atmospheric and vacuum distillation unit is built, including the primary distillation column, atmospheric column and vacuum column. The solving method of equation set is the bubble point method (BP) in the three-diagonal matrix method.

[0090] The optimization model of atmospheric and vacuum distillation unit is a process optimization tool based on the simulation model of atmospheric and vacuum distillation unit. Its core function is to find the optimal process operating parameters that meet the production targets through systematic scheme calculation and screening, and to provide decision support for efficient operation of the unit. The molecular level atmospheric and vacuum distillation separation model is corrected based on actual industrial data, and real-time data communication is established to realize the visualization of the actual atmospheric and vacuum distillation unit operating conditions.

[0091] It needs to be explained that the above-mentioned optimization model of atmospheric and vacuum distillation unit generates a scheme matrix by setting the upper and lower limits of the key operating variables of the atmospheric and vacuum distillation unit (within the safety constraints and process limit conditions) and the variable step size, imports all schemes into the strict mechanism model for parallel calculation, and aggregates all schemes and their calculation results into a scheme result pool. By setting the optimization target parameter conditions, the schemes that meet the optimization target are screened out, and the detailed parameters and optimization effects of the screened schemes are displayed to the user.

[0092] In one specific embodiment, the present application realizes the molecular level dynamic visualization and precise control of the device operation; relying on the strict mechanism model, the product yield and properties are accurately calculated and predicted in real time based on molecular composition; the model is continuously calibrated by actual operation data, which can output accurate results in real time, overcoming the defects of traditional simulation lag and large deviation; through the comparison of real-time data and simulation results, abnormal conditions can be quickly identified, and dynamic control can be guided combined with the change of molecular composition, so that the operator can intuitively understand the influence of molecular level on the product.

[0093] Specifically, the molecular level atmospheric and vacuum distillation separation model is corrected based on actual industrial data, and the specific analysis process is as follows: the actual industrial data includes the extraction temperature, extraction amount and macroscopic properties of the side stream of the actual atmospheric and vacuum distillation unit. First, the correspondence between the actual industrial data and the output of the molecular level atmospheric and vacuum distillation separation model is determined, and then the deviations between the calculated values of the molecular level atmospheric and vacuum distillation separation model and the actual industrial data are compared and iteratively adjusted for thermodynamic parameters, distillation column model parameters and molecular mapping parameters, so as to ensure that the calculation results of the molecular level atmospheric and vacuum distillation unit model are consistent with the actual industrial data.

[0094] It needs to be explained that the above side stream macroscopic properties include the density, distillation range, nitrogen content, etc. of each side line; the obtained data is pre-processed, abnormal values are removed by 3σ criterion, data of the same operation period is time-aligned, and the extraction amount is converted into yield, distillation range is standardized, to ensure the effectiveness and consistency of the data, and to provide a reliable basis for subsequent correction; the mapping relationship between the calculation results of the molecular level atmospheric and vacuum separation model and the actual industrial data is determined: the side line extraction plate temperature calculated by the model corresponds to the actual extraction temperature, the side stream mass yield corresponds to the actual extraction amount, and the macroscopic properties (such as distillation range) obtained based on molecular attributes are weighted to correspond to the actual laboratory analysis value, and the proportion of key heteroatomic molecules corresponds to the actual sulfur / nitrogen content. Through this one-to-one correspondence, the adjustment direction of the model parameters is directly related to the deviation of the actual data.

[0095] The above iterative adjustment refers to hierarchical optimization of parameters. For thermodynamic parameters, the binary interaction parameters of SRK / PR equation or the correction term of group contribution are fitted or adjusted to reduce the deviation of side line distillation range and density; for distillation column model parameters, the number of theoretical plates is increased or decreased, the plate efficiency is adjusted, the reflux ratio or extraction plate position is adjusted to make the side line yield and extraction temperature deviation converge; for molecular mapping parameters, the virtual component division is refined, the deviation factor weight is adjusted or the algorithm is replaced to control the key molecular proportion and sulfur / nitrogen content deviation within the threshold; finally, through cross-parameter collaborative iteration, high-impact parameters are optimized in priority according to the deviation contribution, forming a "adjustment-calculation-evaluation" closed loop, and finally ensuring that the global results of the model are consistent with the actual industrial data.

[0096] In one specific embodiment, the present application realizes efficient global optimization of device flow. Using the high-precision prediction ability of the strict mechanism model, multi-dimensional process scheme matrix can be quickly simulated and evaluated, and the product yield, molecular composition and macroscopic properties of each scheme can be accurately output; parallel computing technology greatly speeds up the scheme screening and evaluation process. This provides timely and comprehensive data support for process adjustment, and completely changes the traditional inefficient and high-cost trial-and-error mode.

[0097] Further, real-time data communication is established, and the specific analysis process is as follows: a communication connection is established between the simulation model of the atmospheric-vacuum distillation unit and the DCS system and the LIMS system of the actual atmospheric-vacuum distillation unit, so that the simulation model of the atmospheric-vacuum distillation unit can obtain target data, the simulation model of the atmospheric-vacuum distillation unit automatically pulls real-time operating data at the time point from the DCS system, pulls the corresponding stream laboratory analysis data at the time point from the LIMS system, performs real-time simulation prediction, and then realizes visualization of the reaction process and product distribution in the actual atmospheric-vacuum distillation unit under the current operating condition; it needs to be explained that the real-time operating data of the DCS system (distributed control system) is stored in the simulation database according to the time sequence; the stream laboratory analysis data of the LIMS system (laboratory information management system) is stored in the simulation database according to the sample information; the above target data includes the real-time operating data pulled from the DCS system and the stream laboratory analysis data pulled from the LIMS system.

[0098] The above real-time operating data refers to dynamic operating parameters in the running process of the actual atmospheric-vacuum distillation unit, for example, operating temperature, pressure of each key equipment (such as a heating furnace); flow rate, flow velocity of each stream (such as feed); heating furnace heat load, tower top condensation temperature, tower bottom liquid level and other control parameters; valve opening, pump power and other equipment running state parameters.

[0099] The above stream laboratory analysis data refers to the molecular composition or physical and chemical property data of each stream stored in the LIMS system, for example, physical properties such as density, viscosity, distillation range, flash point of feed crude oil and each side product (such as gasoline); key component content (such as hydrocarbon composition); parameters representing molecular level composition (such as carbon number distribution).

[0100] The above real-time simulation prediction refers to that the simulation model of the atmospheric-vacuum distillation unit, according to a preset interval, first pre-processes the real-time operating data obtained from the DCS system and the stream laboratory analysis data of the LIMS system, including checking integrity, aligning time axis and standardizing units; then takes the processed data as input, calls a molecular level atmospheric-vacuum separation model and related modules, calculates the molecular level composition of each side stream, predicts macroscopic properties, dynamically evaluates mapping accuracy and corrects deviation; finally, the results are output in a visual form to generate a deviation report, realizing dynamic presentation and decision support of the reaction process and product distribution of the unit under the current operating condition.

[0101] In one specific embodiment, the present application breaks through the limitations of traditional virtual component simulation by providing a process simulation optimization method for a molecular-level atmospheric and vacuum distillation unit, realizes fine modeling and dynamic regulation at the molecular level, accurately converts the composition of streams before and after process simulation using a bidirectional mapping mechanism of molecules and virtual components, and completely retains molecular details; combines a strict mechanism model calibrated by actual data to predict product yield and properties in real time, which provides immediate and accurate data support for process adjustment, and associates macroscopic process parameters with microscopic molecular characteristics, thereby opening up a new path for intelligent and fine operation of the atmospheric and vacuum distillation unit.

[0102] Referring to Figure 2 The system includes a molecular library construction and mapping module, a simulation prediction and dynamic regulation module, a multi-scheme optimization and screening module, and a simulation database.

[0103] The molecular library construction and mapping module is connected to the simulation prediction and dynamic regulation module, and the simulation prediction and dynamic regulation module is connected to the multi-scheme optimization and screening module. The molecular library construction and mapping module, the simulation prediction and dynamic regulation module, and the multi-scheme optimization and screening module are all connected to the simulation database. The simulation database is used to store various parameters involved in the process simulation optimization system for a molecular-level atmospheric and vacuum distillation unit.

[0104] The molecular library construction and mapping module is used to construct a molecular library containing molecules of all distillation fractions of a petroleum sample, and to predict and optimize the molecules in the molecular library, thereby establishing a molecular-level composition model. Based on the initialized mapping rules, virtual components and molecules in the molecular-level composition model are mapped and converted to each other, thereby deducing the molecular composition of the separated products in the crude oil separation process, monitoring and obtaining the parameters of the mapping and conversion process, and classifying the accuracy of the current mapping rules. The simulation prediction and dynamic regulation module is used to realize simulation of the mapping and conversion process through an atmospheric and vacuum distillation unit simulation model, thereby predicting the macroscopic properties and molecular-level composition of each side product of the actual atmospheric and vacuum distillation unit, and realizing working condition visualization and dynamic regulation in combination with real-time data of the actual atmospheric and vacuum distillation unit. The multi-scheme optimization and screening module is used to generate a multi-dimensional process scheme matrix through an atmospheric and vacuum distillation unit optimization model, and to screen out a target process scheme based on parallel computing of a strict mechanism model, thereby realizing process simulation optimization of the molecular-level atmospheric and vacuum distillation unit.

[0105] The above-described embodiments can be implemented in whole or in part by software, hardware (e.g., circuitry), firmware, or any combination thereof. When implemented in software, the above-described embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions according to the embodiments of the present application are entirely or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center through a wired (e.g., infrared, wireless, microwave, etc.) manner. The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server, data center, etc. containing one or more available medium collections. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. The semiconductor medium can be a solid-state disk.

[0106] It should be understood that the term "and / or" herein merely describes an association relationship of associated objects, which means that there can be three relationships, for example, A and / or B can represent three cases of A alone, A and B together, and B alone, where A and B can be singular or plural. In addition, the character " / " herein generally represents an "or" relationship between the associated objects before and after it, but it can also represent an "and / or" relationship, which can be understood in the context before and after it.

[0107] In the present application, "at least one" means one or more, and "multiple" means two or more. "At least one of the following" or similar expressions means any combination of these items, including any combination of single or multiple items. For example, at least one of a, b, or c can mean a, b, c, a-b, a-c, b-c, or a-b-c, where a, b, and c can be single or multiple.

[0108] It should be understood that in various embodiments of the present application, the size of the sequence number of the above-described processes does not mean the order of execution, and the execution order of the processes should be determined by their functions and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0109] Those skilled in the art can clearly understand that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software manner depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0110] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working processes of the devices, apparatuses and units described above can refer to the corresponding processes in the foregoing method embodiments, which will not be described here.

[0111] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A process simulation and optimization method for a molecular-level atmospheric and vacuum distillation unit, characterized in that, The method includes: Step 1: Construct a molecular library containing molecules from the entire fraction of a petroleum sample, and predict and optimize the molecules in the library to establish a molecular-level composition model. Based on the initialized mapping rules, map and transform virtual components with molecules in the molecular-level composition model to deduce the molecular composition of the separation products during crude oil separation. Monitor and acquire the mapping and transformation process parameters, and classify the accuracy of the current mapping rules. The specific classification process is as follows: First, by analyzing the mapping and transformation process parameters, obtain the mapping accuracy coefficient of the molecular-level composition model, and compare it with the preset first and second classification thresholds in the simulation database. When the mapping accuracy coefficient of the molecular-level composition model is not lower than the first classification threshold... When the accuracy of the current mapping rule is high, the core parameters of the current mapping rule are recorded as a benchmark template for similar crude oil samples. When the accuracy coefficient of the molecular composition model mapping is between the first and second classification thresholds, the accuracy of the current mapping rule is classified as medium accuracy and the current mapping rule is optimized. When the accuracy coefficient of the molecular composition model mapping is lower than the second classification threshold, the accuracy of the current mapping rule is classified as low accuracy and the current mapping rule is adjusted. The accuracy coefficient of the molecular composition model mapping is re-acquired and marked as the final accuracy value of the molecular composition model mapping, thereby determining whether to issue a warning for the current mapping rule. Step 2: Simulate the mapping transformation process using a simulation model of the atmospheric and vacuum distillation unit, thereby predicting the macroscopic properties and molecular composition of the side-line products of the actual atmospheric and vacuum distillation unit. At the same time, combine the real-time data of the actual atmospheric and vacuum distillation unit to realize the visualization and dynamic control of the operating conditions. Step 3: Generate a multi-dimensional process scheme matrix through the atmospheric and vacuum distillation unit optimization model, and simultaneously screen out the target process scheme based on the parallel calculation of the rigorous mechanism model, thereby realizing the process simulation optimization of the molecular-level atmospheric and vacuum distillation unit.

2. The process simulation and optimization method for a molecular-level atmospheric and vacuum distillation unit according to claim 1, characterized in that, The optimization process for the current mapping rules is as follows: The average of the first classification threshold and the second classification threshold is taken and marked as the median of mapping accuracy. The mapping accuracy coefficients of the molecular composition model are then compared with the median of mapping accuracy. When the mapping accuracy coefficient of the molecular composition model is not lower than the median of the mapping accuracy, the high-area deviation value is obtained based on the mapping accuracy coefficient of the molecular composition model and the first classification threshold. Single-level optimization is performed based on the high-area deviation value, that is, the density weight enhancement value of the virtual component is directly matched based on the high-area deviation value to improve the mapping accuracy of the virtual component density. When the mapping accuracy coefficient of the molecular composition model is lower than the median of the mapping accuracy, a dual optimization is performed. The specific process is as follows: The first optimization refers to obtaining the low-range deviation value based on the mapping accuracy coefficient of the molecular composition model and the second classification threshold, and matching the boiling range interval reduction coefficient based on the low-range deviation value, thereby reducing the boiling range interval of the virtual components. The second optimization refers to introducing the correction term of the group contribution method to refine the molecular density prediction model.

3. The process simulation and optimization method for a molecular-level atmospheric and vacuum distillation unit according to claim 1, characterized in that, The adjustment process for the current mapping rules is as follows: Based on the accuracy coefficient of the molecular composition model mapping and the second classification threshold, the lower limit deviation value is obtained and compared with the preset failure deviation value in the simulation database. When the lower limit deviation value is lower than the failure deviation value, the criteria for dividing virtual components are redefined; When the lower limit deviation value is not lower than the failure deviation value, the mapping algorithm of the current mapping rule is changed.

4. The process simulation and optimization method for a molecular-level atmospheric and vacuum distillation unit according to claim 1, characterized in that, The specific analysis process for monitoring and acquiring parameters of the mapping transformation process is as follows: The molecular attribute distribution deviation factor, yield conservation deviation factor, and key molecule type proportion deviation factor are extracted from the parameters of the mapping and transformation process as core evaluation parameters. By presetting the effect intensity coefficient of each factor in the simulation database, the weight contribution value of each factor to the mapping accuracy coefficient of the molecular composition model is quantified. Finally, a weighted average fusion algorithm is used to synthesize the mapping accuracy coefficient of the molecular composition model, where the mapping accuracy coefficient of the molecular composition model represents the degree of accuracy of mutual mapping and transformation between virtual components and molecules in the molecular composition model.

5. The process simulation and optimization method for a molecular-level atmospheric and vacuum distillation unit according to claim 1, characterized in that, The specific process for determining whether to issue a warning for the current mapping rule is as follows: The accurate final value of the molecular-level composition model mapping is compared with the first classification threshold; When the final value of the molecular composition model mapping is not lower than the first classification threshold, it is determined that no warning will be issued for the current mapping rule, and the core parameters of the current mapping rule will be recorded as a benchmark template for similar crude oil samples. When the final value of the molecular composition model mapping accuracy is lower than the first classification threshold, an early warning is issued for the current mapping rule. At the same time, the mismatch pattern between the key features of the current crude oil and the mapping rule is recorded and marked as an adaptation blind zone. Meanwhile, the adaptability of the objective function of the optimization model is verified.

6. The process simulation and optimization method for a molecular-level atmospheric and vacuum distillation unit according to claim 1, characterized in that, The process simulation and optimization for realizing the molecular-level atmospheric and vacuum distillation device is specifically analyzed as follows: The molecular-level atmospheric and vacuum distillation device includes a molecular-level composition model, an atmospheric and vacuum distillation device simulation model, and an atmospheric and vacuum distillation device optimization model. The atmospheric and vacuum distillation unit simulation model is based on a molecular-level atmospheric and vacuum separation model, supplemented by a thermodynamic calculation module, a single-equilibrium stage phase equilibrium module, and a distillation column unit model. Together, they form a complete separation process simulation system to realize the separation process of an actual atmospheric and vacuum distillation unit. The atmospheric and vacuum distillation unit optimization model is a process optimization tool built based on the atmospheric and vacuum distillation unit simulation model. Its core function is to find the optimal process operation parameters that meet the production goals through systematic scheme calculation and screening, and to provide decision support for the efficient operation of the unit. The molecular-level atmospheric and vacuum distillation separation model is calibrated based on actual industrial data, and real-time data communication is established to visualize the actual operating conditions of the atmospheric and vacuum distillation unit.

7. The process simulation and optimization method for a molecular-level atmospheric and vacuum distillation unit according to claim 6, characterized in that, The specific analysis process for correcting the molecular-level atmospheric and vacuum separation model based on actual industrial data is as follows: The actual industrial data includes the extraction temperature, extraction rate, and macroscopic properties of the side stream of the actual atmospheric and vacuum distillation unit. First, the correspondence between the actual industrial data and the output of the molecular-level atmospheric and vacuum distillation separation model is determined. Then, for thermodynamic parameters, distillation column model parameters, and molecular mapping parameters, iterative adjustments are made by comparing the deviations between the calculated values ​​of the molecular-level atmospheric and vacuum distillation separation model and the actual industrial data, thereby ensuring that the calculation results of the molecular-level atmospheric and vacuum distillation unit model are consistent with the actual industrial data.

8. The process simulation and optimization method for a molecular-level atmospheric and vacuum distillation unit according to claim 6, characterized in that, The specific analysis process for establishing real-time data communication is as follows: A communication connection is established between the atmospheric and vacuum distillation unit simulation model and the actual atmospheric and vacuum distillation unit's DCS and LIMS systems. This enables the simulation model to acquire target data. The simulation model automatically retrieves real-time operational data from the DCS system and corresponding flow laboratory analysis data from the LIMS system at preset simulation intervals to perform real-time simulation and prediction. This allows for visualization of the reaction process and product distribution in the actual atmospheric and vacuum distillation unit under the current operating conditions.

9. A process simulation and optimization system for a molecular-level atmospheric and vacuum distillation apparatus, employing the process simulation and optimization method for a molecular-level atmospheric and vacuum distillation apparatus as described in any one of claims 1 to 8, characterized in that: include: The molecular library construction and mapping module is used to construct a molecular library containing molecules from the entire distillation fraction of petroleum samples, predict and optimize the molecules in the library to establish a molecular-level composition model. Based on the initialized mapping rules, virtual components are mapped and transformed with molecules in the molecular-level composition model to deduce the molecular composition of the separation products during crude oil separation. The module monitors and acquires mapping and transformation process parameters and classifies the accuracy of the current mapping rules. The specific classification process is as follows: First, by analyzing the mapping and transformation process parameters, the mapping accuracy coefficient of the molecular-level composition model is obtained and compared with the preset first and second classification thresholds in the simulation database. When the mapping accuracy coefficient of the molecular-level composition model is not lower than... When the first classification threshold is reached, the accuracy of the current mapping rule is classified as high-precision mapping, and the core parameters of the current mapping rule are recorded as a benchmark template for similar crude oil samples. When the accuracy coefficient of the molecular composition model mapping is between the first and second classification thresholds, the accuracy of the current mapping rule is classified as medium-precision mapping, and the current mapping rule is optimized. When the accuracy coefficient of the molecular composition model mapping is lower than the second classification threshold, the accuracy of the current mapping rule is classified as low-precision mapping, and the current mapping rule is adjusted. The accuracy coefficient of the molecular composition model mapping is re-acquired and marked as the final accuracy value of the molecular composition model mapping, thereby determining whether to issue a warning for the current mapping rule. The simulation prediction and dynamic control module is used to simulate the mapping transformation process through the simulation model of the atmospheric and vacuum distillation unit, thereby predicting the macroscopic properties and molecular composition of each side-line product of the actual atmospheric and vacuum distillation unit. At the same time, it combines the real-time data of the actual atmospheric and vacuum distillation unit to realize the visualization of the operating conditions and dynamic control. The multi-scheme optimization and screening module is used to generate a multi-dimensional process scheme matrix through the atmospheric and vacuum distillation unit optimization model, and simultaneously screen out the target process scheme based on the parallel calculation of a rigorous mechanism model, thereby realizing the process simulation optimization of the molecular-level atmospheric and vacuum distillation unit.

Images