Simulation method and device for mass transfer of carbon dioxide bubbles in alcohol amine solution

Abstract

The invention provides a simulation method and device for mass transfer of carbon dioxide bubbles in an alcohol amine solution, and belongs to the technical field of chemical engineering. The method comprises the following steps: outputting first working condition data of mass transfer of carbon dioxide bubbles in an alcohol amine solution according to a computational fluid dynamics model; obtaining second working condition data through a preset data conversion network according to the first working condition data; wherein the grid precision of the second working condition data is greater than that of the first working condition data; according to the second working condition data, whether the second working condition data meet the precision requirement or not is judged through a preset calibration network, and a judgment result is obtained; and outputting related working condition data of mass transfer of the carbon dioxide bubbles in the alcohol amine solution according to the judgment result. According to the simulation method provided by the invention, the calculation efficiency is remarkably improved while the calculation precision is ensured.

Description

Technical Field

[0001] The present invention belongs to the field of chemical technology, and specifically relates to a simulation method and device for mass transfer of carbon dioxide bubbles in an alcohol amine solution, electronic equipment, a computer-readable storage medium, and a computer program product. Background Art

[0002] With the development of social economy and industrialization, CO2 emissions from human activities are increasing dramatically. Studies have shown that large amounts of carbon emissions are the main cause of global climate change and global warming. In order to meet this challenge, the world has reached a consensus on greenhouse gas emission reduction, which has greatly promoted the development of carbon capture technology. Chemical absorption is the most widely used carbon capture technology in industry, and its absorbent is usually various types of alcohol amine solutions.

[0003] Bubble column reactors have attracted widespread attention due to their simple structure, excellent mixing characteristics, high energy efficiency and easy control. Computational fluid dynamics (CFD), as a method with the advantages of low cost, high safety, easy parameterization and visualization analysis, has become an important tool for studying complex reaction processes inside bubble column reactors. However, traditional CFD calculations face several challenges in simulating the mass transfer process of CO2 bubbles in amine solutions: first, the computational cost of relying solely on CFD methods for this mass transfer process is high and the calculation speed is slow; second, as the complexity of fluid-related problems increases, the calculation time and resource consumption increase exponentially. This has a huge impact in some environments such as material screening and performance prediction that require a large number of repeated calculations.

[0004] At present, machine learning methods also face some limitations when solving complex fluid dynamics problems: they usually require a large amount of accurate data for training, especially when dealing with complex fluid dynamics problems, which may lead to increased data acquisition and labeling costs; compared with traditional CFD methods, machine learning models may have poor interpretability and difficulty in providing detailed physical insights and mechanism understanding; in addition, the generalization ability and robustness of machine learning models may also face challenges, especially when facing new fields or unknown situations. Therefore, a new computational method is urgently needed to balance computational efficiency and computational accuracy. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art and to provide a simulation method and device for mass transfer of carbon dioxide bubbles in an alcohol amine solution, a computer-readable storage medium and a computer program product that can balance calculation efficiency and calculation accuracy.

[0006] In order to achieve the above objectives, in a first aspect, an embodiment of the present disclosure provides a method for simulating mass transfer of carbon dioxide bubbles in an alcoholamine solution, comprising:

[0007] Outputting first operating condition data of mass transfer of carbon dioxide bubbles in an alcoholamine solution according to a computational fluid dynamics model;

[0008] According to the first operating condition data, second operating condition data is obtained through a preset data conversion network; wherein the grid accuracy of the second operating condition data is greater than the grid accuracy of the first operating condition data;

[0009] According to the second operating condition data, judging whether the second operating condition data meets the accuracy requirement through a preset calibration network, and obtaining a judgment result;

[0010] According to the judgment result, relevant operating data of the mass transfer of carbon dioxide bubbles in the alcoholamine solution are output.

[0011] In some embodiments, the first operating condition data includes at least one of the following data: concentration data of carbon dioxide bubbles, concentration data of alcohol amine solution, pressure data inside carbon dioxide bubbles, phase field variables, and flow field velocity data.

[0012] In some embodiments, judging whether the second operating condition data meets the accuracy requirement through a preset calibration network according to the second operating condition data to obtain a judgment result includes:

[0013] Calculating an actual error integral value of the second operating condition data;

[0014] Inputting the error integral value of the first operating condition data into the preset calibration network to obtain a standard error integral value of the first operating condition data;

[0015] The actual error integral value is compared with the standard error integral value, and whether the second operating condition data meets the accuracy requirement is determined according to the comparison result to obtain a determination result.

[0016] In some embodiments, the outputting of relevant operating data of mass transfer of carbon dioxide bubbles in an alcoholamine solution according to the judgment result includes:

[0017] If the judgment result is that the second operating condition data does not meet the accuracy requirement, the following steps are repeatedly performed:

[0018] Inputting the second operating condition data into the computational fluid dynamics model to obtain an output result, and using the output result as the first operating condition data of mass transfer of carbon dioxide bubbles in an alcoholamine solution;

[0019] According to the first operating condition data, obtaining second operating condition data through a preset data conversion network;

[0020] According to the second operating condition data, judging whether the second operating condition data meets the accuracy requirement through a preset calibration network, and obtaining a judgment result;

[0021] According to the judgment result, outputting relevant working condition data of mass transfer of carbon dioxide bubbles in alcoholamine solution;

[0022] Until the judgment result is that the second operating condition data meets the accuracy requirement, the relevant operating condition data of the mass transfer of carbon dioxide bubbles in the alcohol amine solution is output.

[0023] In some embodiments, the method further comprises:

[0024] Acquire working condition sample data with different grid precisions, wherein the working condition sample data includes first working condition sample data and second working condition sample data, and the grid precision of the second working condition sample data is greater than the grid precision of the first working condition sample data;

[0025] The recurrent neural network is trained according to the operating condition sample data to obtain the preset data conversion network.

[0026] In some embodiments, outputting first operating condition data of mass transfer of carbon dioxide bubbles in an alcohol amine solution according to a computational fluid dynamics model includes:

[0027] The Euler-Euler method is used to calculate the viscous stress of the gas and liquid phases based on mass conservation;

[0028] The surface tension of gas and liquid phases is calculated using the phase field model;

[0029] According to the viscous stress and the surface tension, first operating condition data of mass transfer of carbon dioxide bubbles in the alcoholamine solution is output through diffusion mass transfer, convection mass transfer and chemical reaction.

[0030] In some embodiments, the calculating the surface tension of the gas phase and the liquid phase using a phase field model comprises:

[0031] The square of the grid width at the maximum phase interface or twice the grid width at the maximum phase interface is used as the phase field interface thickness;

[0032] The surface tension of the gas phase and the liquid phase is calculated using a phase field model based on the phase field interface thickness.

[0033] In some embodiments, the method further comprises:

[0034] According to the population size and the fitness of dung beetles, the acceptance degree of the optimal solution of the young ball individuals and the stealing dung beetle individuals in the dung beetle optimization algorithm is adjusted adaptively according to the fitness, and the first dung beetle optimization algorithm is obtained;

[0035] The global search mechanism of the sparrow search algorithm is integrated with the first dung beetle optimization algorithm to obtain a second dung beetle optimization algorithm;

[0036] Based on the second dung beetle optimization algorithm, adaptive distributed perturbation is performed on the foraging of small dung beetles to obtain a third dung beetle optimization algorithm;

[0037] The solution space of the third dung beetle optimization algorithm is expanded by using Cauchy-Gauss mutation to obtain an improved dung beetle optimization algorithm;

[0038] An improved dung beetle optimization algorithm is used to optimize the hyperparameters of a deep neural network, wherein the deep neural network includes the computational fluid dynamics model, the preset data conversion network and / or the preset calibration network.

[0039] In some embodiments, when the concentration of the alcohol amine solution is between 1.5% and 12%, the mass transfer of carbon dioxide bubbles in the alcohol amine solution is simulated using interpolation conditions;

[0040] When the concentration of the alcohol amine solution is between 4% and 8%, the extrapolation conditions are used to simulate the mass transfer of carbon dioxide bubbles in the alcohol amine solution.

[0041] In a second aspect, the present disclosure provides a device for simulating mass transfer of carbon dioxide bubbles in an alcohol amine solution, the device comprising:

[0042] An output module, used for outputting first operating condition data of mass transfer of carbon dioxide bubbles in an alcohol amine solution according to a computational fluid dynamics model;

[0043] A conversion module, used to obtain second operating condition data through a preset data conversion network according to the first operating condition data; wherein the grid accuracy of the second operating condition data is greater than the grid accuracy of the first operating condition data;

[0044] A judgment module, used for judging whether the second operating condition data meets the accuracy requirement through a preset calibration network according to the second operating condition data, and obtaining a judgment result;

[0045] The simulation module is used to output relevant operating data of mass transfer of carbon dioxide bubbles in the alcohol amine solution according to the judgment result.

[0046] In a third aspect, the present disclosure further provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the computing method of the first aspect described above is implemented.

[0047] In a fourth aspect, the present disclosure further provides a computer-readable storage medium storing a computer program for executing the computing method of the first aspect.

[0048] In a fifth aspect, the present disclosure further provides a computer program product, including a computer program / instruction, which implements the steps of the computing method of the first aspect when the computer program / instruction is executed by a processor.

[0049] The simulation method of mass transfer of carbon dioxide bubbles in an alcohol amine solution provided by the embodiment of the present disclosure firstly converts the first operating condition data output by the computational fluid dynamics model into fine grid data through a preset data conversion network for error calibration; secondly, the second operating condition data obtained after error correction is passed through a preset calibration network to determine whether the second operating condition data meets the accuracy requirements, so as to ensure that the relevant operating condition data finally outputted meets the accuracy requirements. At the same time, the present disclosure uses a neural network to process the first operating condition data and the second operating condition data, which can significantly improve the calculation efficiency of the data. The method provided by the present disclosure significantly improves the calculation efficiency while ensuring the calculation accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0051] Figure 1 A flow chart of a method for simulating mass transfer of carbon dioxide bubbles in an alcohol amine solution provided in an embodiment of the present disclosure;

[0052] Figure 2 A schematic diagram of a method for simulating mass transfer of carbon dioxide bubbles in an alcohol amine solution provided in an embodiment of the present disclosure;

[0053] Figure 3 A structural block diagram of a simulation device for mass transfer of carbon dioxide bubbles in an alcohol amine solution provided in an embodiment of the present disclosure;

[0054] Figure 4 A schematic diagram of the physical structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0055] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention is further described in detail and completely in conjunction with the accompanying drawings and specific implementation methods. Obviously, the described embodiments are only part of the embodiments of the present disclosure, rather than all the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present disclosure.

[0056] In a first aspect, embodiments of the present disclosure provide a method for simulating the mass transfer of carbon dioxide bubbles in an alcohol amine solution.

[0057] Figure 1A flow chart of a method for simulating mass transfer of carbon dioxide bubbles in an alcohol amine solution provided in an embodiment of the present disclosure. Figure 2 The schematic diagram of a simulation method for mass transfer of carbon dioxide bubbles in an alcohol amine solution provided in an embodiment of the present disclosure. Figure 1-Figure 2 As shown, the method comprises the following steps:

[0058] S11. Outputting first operating condition data of mass transfer of carbon dioxide bubbles in an alcoholamine solution according to a computational fluid dynamics model.

[0059] Specifically, the computational fluid dynamics model, referred to as the CFD model, is a technology that uses computers to simulate fluid flow. The CFD model uses numerical calculation methods to solve mathematical equations that describe fluid motion and simulate various related physical phenomena when fluid flows. The CFD model is a coarse grid model, and the output first working condition data is also coarse grid data, that is, Figure 2 The first operating condition data refers to the data output by the CFD model related to the mass transfer of carbon dioxide (abbreviated as CO2) in the alcohol amine solution.

[0060] In some embodiments, the first operating condition data includes at least one of the following data: concentration data of carbon dioxide bubbles, concentration data of alcohol amine solution, pressure data inside carbon dioxide bubbles, phase field variables, and flow field velocity data.

[0061] S12. According to the first operating condition data, obtain second operating condition data through a preset data conversion network; wherein the grid accuracy of the second operating condition data is greater than the grid accuracy of the first operating condition data.

[0062] Specifically, the second operating condition data refers to fine grid data. The preset data conversion network can convert the coarse grid data into fine grid data, so as to achieve the purpose of performing error calibration on the coarse grid first operating condition data output by the CFD model to obtain fine grid data.

[0063] S13. According to the second operating condition data, determine whether the second operating condition data meets the accuracy requirement through a preset calibration network to obtain a determination result.

[0064] Specifically, considering that under certain working conditions, when the amount of first working condition data is small, large errors are likely to occur, and at this time, the simulation results have no data for comparison and verification. The present disclosure uses a preset calibration network to determine whether the second working condition data obtained after error calibration meets the accuracy requirements.

[0065] S14. Outputting relevant operating data of mass transfer of carbon dioxide bubbles in alcoholamine solution according to the judgment result.

[0066] Specifically, judging whether to output relevant operating condition data of mass transfer of carbon dioxide bubbles in an alcohol amine solution according to the judgment result can make the output result more accurate and better meet the accuracy requirement.

[0067] The simulation method of mass transfer of carbon dioxide bubbles in an alcohol amine solution provided by the embodiment of the present disclosure firstly converts the first operating condition data output by the computational fluid dynamics model into fine grid data through a preset data conversion network for error calibration; secondly, the second operating condition data obtained after error correction is passed through a preset calibration network to determine whether the second operating condition data meets the accuracy requirements, so as to ensure that the relevant operating condition data finally outputted meets the accuracy requirements. At the same time, the present disclosure uses a neural network to process the first operating condition data and the second operating condition data, which can significantly improve the calculation efficiency of the data. The method provided by the present disclosure significantly improves the calculation efficiency while ensuring the calculation accuracy.

[0068] In some embodiments, the method not only includes the above steps S11 to S14, but also includes a process of generating a preset data conversion network, specifically including the following steps:

[0069] S15. Acquire operating condition sample data with different grid precisions, wherein the operating condition sample data includes first operating condition sample data and second operating condition sample data, and the grid precision of the second operating condition sample data is greater than the grid precision of the first operating condition sample data.

[0070] Among them, the first working condition sample data is coarse grid data, and the second working condition sample data is fine grid data.

[0071] See also Figure 2 , the process of generating a preset data conversion network can be processed in the offline stage. In this way, the amount of calculation in the online stage can be reduced and the calculation efficiency can be improved. The present disclosure can use the CFD model to construct a data set for neural network training. Assume that the simulation environment for the mass transfer of carbon dioxide bubbles in the alcohol amine solution is: the operating environment temperature is constant at 30°C. The material property variables that mainly affect the absorption during the mass transfer of carbon dioxide bubbles in the alcohol amine solution are the viscosity μ of the alcohol amine solution, the surface tension σ of the alcohol amine solution and the Henry coefficient H E .

[0072] In order to construct a sufficient data set, 50 sets of working condition data were taken under different parameter conditions. Each set of working condition data contained the precise calculation results of the fine grid and the uncorrected calculation results of the coarse grid. The data were recorded every 4.4ms, and a total of 900 frames of data were taken. The coarse grid only increased the size of the grid.

[0073] In some embodiments, the number of first working condition sample data is 1 / 5 of the total number of samples, that is, the number of coarse grid samples is 1 / 5 of the total number of samples, and the ratio of the number of coarse grid sample data to the number of fine grid sample data is 1 / 4. Setting the ratio of coarse grid data to fine grid data in this way can achieve better training results.

[0074] S16. Training a recurrent neural network according to the operating condition sample data to obtain the preset data conversion network.

[0075] In some embodiments, the recurrent neural network adopts a BiLSTM neural network architecture.

[0076] The present disclosure introduces a recurrent neural network represented by BiLSTM for offline training and online retrieval to improve the response speed of the model. Compared with the traditional LSTM, BiLSTM can more comprehensively capture the time dependency of sequence data by adding a reverse LSTM layer on the basis of the forward LSTM layer, thereby more accurately predicting the bubble rising process. Of course, the recurrent neural network can also be a long short-term memory network, a gated recurrent unit, a bidirectional recurrent neural network, etc., and the present disclosure does not limit this.

[0077] In some embodiments, step S16 specifically includes the following steps:

[0078] S161. Preprocess the working condition sample data with different grid accuracies to construct a spatial data set.

[0079] S162, dividing the number of data with different grid precisions under the same working condition into training data and verification data in a ratio of 7:3.

[0080] S163. According to the training data, the recurrent neural network is trained with the grid finest working condition data as the objective function to obtain a trained neural network model, and the network parameters of the trained neural network model are adjusted according to the verification data to obtain a preset data conversion network.

[0081] In some embodiments, step S16 not only includes the above steps S161 to S163, but also specifically includes the following steps: S164, evaluating the performance of the preset data conversion network according to the test set.

[0082] The preset data conversion network obtained by the training in the offline stage through the above steps disclosed in the present invention can, on the one hand, reduce the amount of online calculation and improve the calculation efficiency; on the other hand, it can also establish a mapping relationship between the calculation results of coarse grid data (first working condition data) and fine grid data (second working condition data) through a neural network, perform error correction on the first working condition data output by the CFD model, and improve the accuracy of the final simulation results.

[0083] In some embodiments, step S13, judging whether the second operating condition data meets the accuracy requirement through a preset calibration network according to the second operating condition data, and obtaining a judgment result, comprises the following steps:

[0084] S131, calculating the actual error integral value of the second operating condition data, wherein the actual error represents the final error value corresponding to the first operating condition data output by the current CFD model after error calibration.

[0085] S132: Input the error integral value of the first operating condition data into the preset calibration network to obtain the standard error integral value of the first operating condition data.

[0086] Specifically, the error integral value of the first operating condition data is used to construct the time data of the error integral value of the first operating condition data through the preset calibration network, and based on the time data of the error integral value, the standard error integral value of the first operating condition data is finally obtained.

[0087] S133. Compare the actual error integral value with the standard error integral value, and determine whether the second operating condition data meets the accuracy requirement based on the comparison result to obtain a determination result.

[0088] Specifically, the standard error integral value represents the upper limit of the error corresponding to the first operating condition data. If the actual error integral value of the second operating condition data obtained after error calibration is within the upper limit of the error, it means that the accuracy of the data after error calibration through the preset data conversion network meets the accuracy requirements, and the second operating condition data can be directly output; if the actual error integral value of the second operating condition data obtained after error calibration still exceeds the upper limit of the error, it means that the accuracy of the data after error calibration through the preset data conversion network still does not meet the accuracy requirements, further indicating that the error of the first operating condition data output by the CFD model itself is very large, and still does not meet the accuracy requirements after being processed by the preset data conversion network. At this time, further processing is required to ensure the calculation accuracy of the simulation value finally output.

[0089] In some embodiments, step S14, outputting the relevant operating data of the mass transfer of carbon dioxide bubbles in the alcoholamine solution according to the judgment result, specifically comprises the following steps: if the judgment result is that the second operating data does not meet the accuracy requirement, repeating the following steps:

[0090] Step S141, inputting the second operating condition data into the computational fluid dynamics model to obtain an output result, and using the output result as the first operating condition data of the mass transfer of carbon dioxide bubbles in the alcohol amine solution.

[0091] Step S142: According to the first operating condition data, obtain second operating condition data through a preset data conversion network.

[0092] Step S143: According to the second operating condition data, determine whether the second operating condition data meets the accuracy requirement through a preset calibration network to obtain a determination result.

[0093] Step S144: outputting relevant operating data of mass transfer of carbon dioxide bubbles in the alcoholamine solution according to the judgment result.

[0094] Until the judgment result is that the second operating condition data meets the accuracy requirement, the relevant operating condition data of the mass transfer of carbon dioxide bubbles in the alcohol amine solution is output.

[0095] That is, when the judgment result is that the second operating condition data does not meet the accuracy requirement, the second operating condition data is input into the computational fluid dynamics model, and the above steps S11 to S14 are repeated; until the judgment result is that the second operating condition data meets the accuracy requirement, the relevant operating condition data of the mass transfer of carbon dioxide bubbles in the alcoholamine solution is output.

[0096] Specifically, if the actual error integral value of the second working condition data obtained after error calibration still exceeds the upper limit of the error, it means that the error of the first working condition data output by the CFD model itself is very large, and even after error calibration through the preset data conversion network, it does not meet the accuracy requirement. Figure 2 As shown, the second operating condition data needs to be refined again by the CFD model, and then the above steps S11 to S14 are repeated to ensure the calculation accuracy of the simulation value finally outputted.

[0097] In some embodiments, the method includes not only the above steps S11 to S14, but also includes the step of generating a preset calibration network, specifically including:

[0098] S21. Obtain error integral values ​​of operating condition sample data with different grid precisions, wherein the operating condition sample data includes first operating condition sample data and second operating condition sample data, and the grid precision of the second operating condition sample data is greater than the grid precision of the first operating condition sample data.

[0099] S22. Training the time series integral value network according to the integral value of the working condition sample data to obtain the preset calibration network.

[0100] Specifically, Figure 2As shown, the process of training to obtain a preset calibration network can also be processed in the offline stage. Among them, for the preset calibration network that provides calibration accuracy, the main construction process is similar to the aforementioned preset data conversion network. The only difference is that when the preset calibration network is trained, the sample data is the first working condition sample data (that is, the coarse grid sample data), the training parameter is the integral value of the second working condition data (that is, the integral value of the fine grid data), and the network used is a time series integral value network. The present disclosure uses a time series single-step solution network to provide the variable integral value of the precision grid under the prediction parameter for the online process. The other processes are the same as the process of training the preset data conversion network, and will not be repeated here.

[0101] It should be noted that the preset data conversion network, preset calibration network, CFD model, and recurrent neural network and time series integral value network mentioned above may all include a BiLSTM network layer and an Attention mechanism layer. The BiLSTM network layer is used to process time series data; the Attention mechanism layer is used to adaptively assign weights to key time steps in sequence data.

[0102] In some embodiments, step S11, outputting first operating condition data of mass transfer of carbon dioxide bubbles in an alcoholamine solution according to a computational fluid dynamics model, comprises the following steps:

[0103] S111. Based on the conservation of mass, the Euler-Euler method is used to calculate the viscous stress of the gas phase and the liquid phase.

[0104] Specifically, the embodiment of the present disclosure assumes that both the gas phase (carbon dioxide) and the liquid phase (alcoholamine solution) satisfy the continuity equation, and adopts the Euler-Euler method (also known as the Euler-Euler model), then the mass and momentum conservation equations of the two can be written as follows:

[0105]

[0106] Wherein, formula (1) is the mass conservation equation, which means that in any volume, the rate of change of mass is equal to the net flow rate of mass flowing in and out of the volume, ρ represents the density of the gas phase or liquid phase, and V is the velocity vector. Formula (2) is the momentum conservation equation, which means that the rate of change of momentum is equal to the sum of the external forces acting on the fluid. Wherein, p represents the pressure of the two phases, τ represents the viscous stress tensor of the two phases, g is the gravitational acceleration, and F σ Represents the force between phases, such as friction between phases.

[0107] Assuming that both phases are weakly compressible fluids, the viscous stress τ of the two phases can be expressed as:

[0108]

[0109] where μ is the dynamic viscosity of the fluid. is the velocity gradient tensor, which describes the rate of change of the fluid velocity in space. is the transpose of the velocity gradient tensor. is the divergence of the velocity, which represents the rate of change of the volume of the fluid. I is the unit tensor.

[0110] S112. Use the phase field model to calculate the surface tension of the gas phase and the liquid phase.

[0111] Specifically, for the surface tension of the gas-liquid two phases, the present disclosure adopts the Cahn-Hilliard equation of the phase field model to solve it, and the surface tension is equivalent to the product of the field variable gradient and the chemical potential, and is added to the formula (1) in the form of volume force.

[0112] Compared with the traditional Level-Set method, the phase field method used in the embodiment of the present disclosure can ensure that the total energy of the system is reasonably reduced, thereby reflecting more detailed information at the physical level.

[0113] In some embodiments, step S112, calculating the surface tension of the gas phase and the liquid phase using the phase field model, specifically includes:

[0114] S1121. Use the square of the grid width at the maximum phase interface or twice the grid width at the maximum phase interface as the phase field interface thickness.

[0115] S1122. Calculate the surface tension of the gas phase and the liquid phase using a phase field model according to the phase field interface thickness.

[0116] Specifically, the transition thickness between the gas phase and the liquid phase depends on the grid size at the phase interface, and is generally the square of the maximum grid width. If the difference between the maximum grid and the minimum grid in the grid is small, twice the maximum grid width can also be selected.

[0117] S113. Outputting first operating condition data of mass transfer of carbon dioxide bubbles in an alcoholamine solution according to the viscous stress and the surface tension through diffusion mass transfer, convection mass transfer and chemical reaction.

[0118] Specifically, during the rising process of carbon dioxide (CO2) bubbles, the components in the gas phase continuously dissolve into the liquid phase, resulting in a higher gas phase concentration near the gas-liquid interface than the liquid phase concentration far from the interface. This concentration gradient drives the gas phase substances to diffuse into the liquid phase. At the same time, the rising of the bubbles will cause the movement of the liquid phase, produce a convective mass transfer effect, and also promote the transfer of gas phase substances to the liquid phase. However, the CO2 in the liquid phase will react with the free amine solution (MEA) molecules and consume the CO2 in the liquid phase. In summary, in this process, diffusion mass transfer, convective mass transfer and chemical reactions interweave and affect the transfer process of gas phase substances to the liquid phase, which can be expressed as formula (8):

[0119]

[0120] Among them, c represents the concentration of gas phase CO2 component, R represents the reaction rate of CO2 in MEA solution, J represents the molar flux of CO2 component, and the expression of J is:

[0121]

[0122] In the formula, is the diffusion coefficient of CO2 in MEA solution, which can be expressed as equations (10)-(11):

[0123]

[0124] Where T represents temperature, Represents the dynamic viscosity of water, μ Solution Represents the dynamic viscosity of MEA aqueous solution. For the R term in equation (8), the reaction rate R in the process can be expressed as equation (12):

[0125]

[0126] Where [CO2] and [MEA] represent the CO2 concentration in the solution and the unreacted MEA concentration, respectively, and k2 is measured as:

[0127]

[0128] Based on the above process, the simulation of mass transfer of carbon dioxide bubbles in alcohol amine solution is as follows:

[0129] Assuming that the gas introduced is pure CO2, the mass transfer rate k during the bubble rise process is l It can be defined as:

[0130]

[0131] Where A represents the bubble area, which can be obtained by numerical integration of the phase field, n b Represents the molar amount of gas phase CO2 reduction, C l , g represents the CO2 concentration in the gas-liquid boundary layer, C l Represents the CO2 concentration inside the liquid. According to Kasturi’s research, CO2 is consumed by the reaction at the gas-liquid interface. l The concentration of dissolved CO2 in the gas-liquid boundary layer is C l ,g is assumed to be equal to the equilibrium concentration of CO2 in the liquid phase. According to Henry's law, it can be obtained that:

[0132]

[0133] Where pg represents the pressure inside the bubble, H E is the Henry coefficient, which can be calculated based on the bubble pressure Pg obtained by solving equations (1) and (2) mentioned above, thereby obtaining an accurate solution for the mass transfer rate of the complete process.

[0134] The CFD model provided by the embodiment of the present disclosure satisfies the mass conservation equation and the momentum conservation equation, and can solve the equation of fluid motion through numerical methods to simulate various related physical phenomena during fluid flow.

[0135] In some embodiments, when the concentration of the alcoholamine solution is between 1.5% and 12%, the interpolation condition is used to simulate the mass transfer of carbon dioxide bubbles in the alcoholamine solution. In some embodiments, when the concentration of the alcoholamine solution is between 4% and 8%, the extrapolation condition is used to simulate the mass transfer of carbon dioxide bubbles in the alcoholamine solution. This ensures that the method provided by the present disclosure can have good simulation performance and can still maintain a high calculation accuracy under the extrapolation condition.

[0136] In some embodiments, the method includes not only steps S11 to S14, but also the following steps:

[0137] S31. According to the population size and the fitness of dung beetles, the acceptance degree of the young ball individuals and the stealing dung beetle individuals in the dung beetle optimization algorithm to the optimal solution is adaptively adjusted according to the fitness, so as to obtain the first dung beetle optimization algorithm.

[0138] Specifically, the acceptance of the optimal solution by the chicks and thieving dung beetles is optimized to enhance the global search capability. The method is as follows:

[0139]

[0140] Where pop represents the population size; ranki(t) represents the fitness of the i-th dung beetle at the t-th time; is the position of the i-th egg ball in the t+1-th iteration; b1 and b2 are 1*D independent random vectors, and D is the dimension of the optimization problem; The position of the i-th stealing dung beetle at the t+1th iteration;

[0141] Indicates the local optimal position of the group in this iteration; represents the global optimal position; Lb * is the lower bound of the safe area, Ub * is the upper bound of the safe area; g is a random vector of size 1*D and obeys the normal distribution, and S represents a constant.

[0142] By modifying equations (24) and (25), the acceptance of the optimal solution by the Dung Ball and Thief Dung Beetle individuals can be adjusted adaptively according to their fitness. Individuals with lower fitness are given higher acceptance so that they can quickly approach the optimal solution, and individuals with higher fitness are given lower acceptance so that they can maintain good local search capabilities.

[0143] S32: integrating the global search mechanism of the sparrow search algorithm with the first dung beetle optimization algorithm to obtain a second dung beetle optimization algorithm.

[0144] Specifically, the first dung beetle optimization algorithm integrates the random tracking mechanism of the sparrow search algorithm (SSA) to divide individuals into discoverers, followers, and scouts. Among them, discoverers are individuals with higher fitness, accounting for 10%-20% of the group, and are responsible for providing followers with information and directions on the optimal solution to the problem; followers are responsible for tracking discoverers to find the optimal solution; scouts are individuals with lower fitness and need to independently search for the optimal solution elsewhere.

[0145] In some embodiments, the present disclosure optimizes the location update of the follower as follows:

[0146]

[0147] Among them, Q represents a random number that follows a normal distribution, X p t+1 represents the optimal position of the finder, A is a column vector of length d, where the elements are 1 or -1, A + =A T (AA T ) -1 , A T is the transposed matrix of A, and L represents a column vector whose elements are all 1. When i>pop / 2, it means that the follower has not obtained food, has a low fitness, and needs to search elsewhere. The significance of this optimization is to introduce randomness through matrix multiplication, give certain disturbances and independent search opportunities to followers with low fitness, prevent them from falling into local optimality, and enhance global search capabilities. At the same time, integrating the three-role mechanism of the SSA algorithm, such as the discoverer-follower-scout, into the DBO algorithm can also better improve the convergence and global search capabilities of the original algorithm.

[0148] S33. Based on the second dung beetle optimization algorithm, adaptive distributed perturbation is performed on the foraging of small dung beetles to obtain a third dung beetle optimization algorithm.

[0149] Specifically, an adaptive t-distribution perturbation is performed on the foraging of dung beetles to improve the early global search ability and the later local search ability. The method is as follows:

[0150]

[0151] In the formula, t(iter) represents the adaptive random interference term, which can gradually reduce the disturbance amplitude as the search time increases. This means that in the early stage of the search, the interference term is large, and the dung beetle has a larger search range, while in the later stage of the search, the interference term is small, and the dung beetle has a smaller search range, which makes it easier to converge and improve the local search ability.

[0152] S34, using Cauchy-Gauss mutation to expand the solution space of the third dung beetle optimization algorithm to obtain an improved dung beetle optimization algorithm.

[0153] Specifically, the Cauchy-Gauss mutation is used to expand the solution space. When the optimal solution of the group remains unchanged after 5 iterations, the algorithm is judged to be stagnant, and the Cauchy-Gauss mutation is used for the entire group to jump out of the local optimal solution:

[0154]

[0155] Where, β1 = 1-t 2 / T 2 , β2=t 2 / T 2 , T is the maximum number of iterations, Cauchy(0,1) represents random numbers that satisfy the Cauchy distribution, and Gauss(0,1) represents random numbers that obey the Gaussian distribution. The above optimization can make the optimization process temporarily jump out of the current optimal solution and avoid falling into the local optimum.

[0156] S35. Use an improved dung beetle optimization algorithm to optimize the hyperparameters of the deep neural network, wherein the deep neural network includes the computational fluid dynamics model, the preset data conversion network and / or the preset calibration network.

[0157] It should be noted that the improvement of the dung beetle optimization algorithm can also be improved only for any one or more of the above steps S31 to S34, and the improved dung beetle optimization algorithm can be applied to the optimization of the hyperparameters of any neural network involved in the present disclosure, and the present disclosure does not limit this.

[0158] The disclosed embodiment adopts an improved dung beetle optimization algorithm (IDBO) to optimize the hyperparameters of the deep neural network. The improvement of the improved dung beetle optimization algorithm over the traditional dung beetle optimization algorithm (DBO) is that the local search capability of the algorithm is enhanced by optimizing the acceptance of the optimal solution of the chicks and thieving dung beetles, the global search mechanism of the sparrow search algorithm is integrated with the dung beetle optimization algorithm to improve the global optimization effect, the Cauchy-Gaussian variation characteristics are introduced in the population update process to increase the population diversity, and the adaptive t-distribution perturbation is introduced to the foraging stage of the small dung beetles to balance the exploration and development capabilities of the algorithm.

[0159] The present invention integrates the sparrow search algorithm (SSA) on the basis of the original DBO algorithm, aiming to integrate the efficient global search capability and convergence performance of the SSA algorithm into the original DBO algorithm, so that it has both advantages. At the same time, in order to avoid the algorithm from falling into the local optimum during the calculation process, the present invention also adopts the Cauchy-Gaussian mutation method for optimization. In addition, the present invention also adjusts the position update mechanism of the chick ball and the stealing dung beetle, so that it can dynamically adjust the individual's acceptance of the optimal solution, thereby improving the convergence accuracy and convergence stability of the algorithm.

[0160] The present disclosure provides a method for simulating mass transfer of carbon dioxide bubbles in an alcohol amine solution, the method comprising an offline stage and an online stage: in the offline stage, operating data of different grid precisions are obtained, the data are preprocessed and a training data set is constructed, and the hyperparameters of a deep neural network (preset data conversion network) are optimized using an improved dung beetle optimization algorithm; in the online stage, a CFD coarse grid model is run for calculation and calibration. By combining the offline stage with the current stage, the amount of calculation in the online stage can be reduced and the calculation efficiency can be improved. The present disclosure adopts the Euler-Euler model to describe the gas-liquid two-phase flow, uses the phase field method to process the interfacial tension, and establishes a mapping relationship between the coarse and fine grid calculation results through a deep neural network (preset data conversion network) to perform error calibration on the first operating condition data. At the same time, the error integral value of the second operating condition data obtained after calibration is compared with the standard error value output by the preset calibration network, and it can be further determined whether the second operating condition data meets the accuracy requirements, and when the second operating condition data does not meet the accuracy requirements, the second operating condition data is further meshed. While ensuring the calculation efficiency, the method also improves the calculation accuracy. At the same time, the calculation acceleration ratio can reach 6 times in the high-precision mode, and the calculation accuracy R2 is greater than 0.95; the calculation acceleration ratio can reach 19.2 times in the high-speed mode, and the calculation accuracy R2 is greater than 0.85. The disclosure can be widely used in the simulation of the mass transfer process of CO2 bubbles and alcohol amine solution in a bubble column reactor, providing an effective tool for rapid optimization design in industrial applications.

[0161] On the second aspect, based on the same invention, the disclosed embodiment also provides a simulation device for mass transfer of carbon dioxide bubbles in an alcohol amine solution.

[0162] Figure 3 The structure block diagram of a simulation device 500 for mass transfer of carbon dioxide bubbles in an alcohol amine solution provided in an embodiment of the present disclosure. Figure 3 As shown, the device 500 includes an output module 501 , a conversion module 502 , a judgment module 503 and a simulation module 504 .

[0163] The output module 501 is used to output first operating condition data of mass transfer of carbon dioxide bubbles in an alcohol amine solution according to a computational fluid dynamics model.

[0164] The conversion module 502 is used to obtain second operating condition data through a preset data conversion network according to the first operating condition data; wherein the grid accuracy of the second operating condition data is greater than the grid accuracy of the first operating condition data.

[0165] The judgment module 503 is used to judge whether the second operating condition data meets the accuracy requirement through a preset calibration network according to the second operating condition data, and obtain a judgment result.

[0166] The simulation module 504 is used to output relevant operating data of the mass transfer of carbon dioxide bubbles in the alcoholamine solution according to the judgment result.

[0167] In some embodiments, the first operating condition data includes at least one of the following data: concentration data of carbon dioxide bubbles, concentration data of alcohol amine solution, pressure data inside carbon dioxide bubbles, phase field variables, and flow field velocity data.

[0168] In some embodiments, the determination module 503 includes a first calculation unit 513 , a generation unit 523 and a comparison unit 533 .

[0169] The first calculation unit 513 is used to calculate the actual error integral value of the second operating condition data.

[0170] The generating unit 523 is used to input the error integral value of the first operating condition data into the preset calibration network to obtain the standard error integral value of the first operating condition data.

[0171] The comparison unit 533 is used to compare the actual error integral value with the standard error integral value, and determine whether the second operating condition data meets the accuracy requirement according to the comparison result to obtain a determination result.

[0172] In some embodiments, the device 500 includes not only an output module 501, a conversion module 502, a judgment module 503 and a simulation module 504, but also a repetition module 505. The repetition module 505 is used to input the second operating condition data into the computational fluid dynamics model to obtain an output result if the judgment result is that the second operating condition data does not meet the accuracy requirement, and use the output result as the first operating condition data of the mass transfer of carbon dioxide bubbles in the alcohol amine solution. In this way, the conversion module 502 can obtain the second operating condition data through a preset data conversion network based on the first operating condition data; the judgment module 503 can judge whether the second operating condition data meets the accuracy requirement through a preset calibration network based on the second operating condition data to obtain a judgment result; until the judgment result is that the second operating condition data meets the accuracy requirement, the simulation module 504 outputs the relevant operating condition data of the mass transfer of carbon dioxide bubbles in the alcohol amine solution.

[0173] In some embodiments, the device 500 not only includes an output module 501, a conversion module 502, a judgment module 503 and a simulation module 504, but also includes a training module 506 for training the neural network to obtain a preset data conversion network and / or a preset calibration network.

[0174] In some embodiments, the training module 506 includes an acquisition unit 516 and a training unit 526, wherein the acquisition unit 516 is used to acquire working condition sample data with different grid precisions, wherein the working condition sample data includes first working condition sample data and second working condition sample data, and the grid precision of the second working condition sample data is greater than the grid precision of the first working condition sample data; the training unit 526 is used to train the recurrent neural network according to the working condition sample data to obtain the preset data conversion network. The process of the training module 506 generating the preset calibration network is similar to the process of obtaining the preset data conversion network, and will not be repeated here.

[0175] In some embodiments, the output module 501 includes a second computing unit 511 , a third computing unit 521 , and an output unit 531 .

[0176] The second calculation unit 511 is used to calculate the viscous stress of the gas phase and the liquid phase by using the Euler-Euler method based on the conservation of mass.

[0177] The third calculation unit 521 is used to calculate the surface tension of the gas phase and the liquid phase using a phase field model.

[0178] The output unit 531 is used to output first operating condition data of mass transfer of carbon dioxide bubbles in the alcohol amine solution through diffusion mass transfer, convection mass transfer and chemical reaction according to the viscous stress and the surface tension.

[0179] In some embodiments, the third calculation unit 521 is specifically used to use the square of the grid width at the maximum phase interface or twice the grid width at the maximum phase interface as the phase field interface thickness; and calculate the surface tension of the gas phase and the liquid phase using the phase field model based on the phase field interface thickness.

[0180] In some embodiments, the device 500 not only includes an output module 501, a conversion module 502, a judgment module 503 and a simulation module 504, but also includes an optimization module 508 for optimizing the hyperparameters of a deep neural network, wherein the deep neural network includes the computational fluid dynamics model, the preset data conversion network and / or the preset calibration network.

[0181] In some embodiments, the optimization module 508 includes a first optimization unit 518 , a second optimization unit 528 , a third optimization unit 538 , an expansion unit 548 , and a fourth optimization unit 558 .

[0182] The first optimization unit 518 is used to adjust the acceptance of the optimal solution of the chicks and thieving dung beetles in the dung beetle optimization algorithm according to the population size and the fitness of the dung beetles, so as to obtain the first dung beetle optimization algorithm.

[0183] The second optimization unit 528 is used to integrate the global search mechanism of the sparrow search algorithm with the first dung beetle optimization algorithm to obtain a second dung beetle optimization algorithm.

[0184] The third optimization unit 538 is used to perform adaptive distributed perturbation on the foraging of small dung beetles based on the second dung beetle optimization algorithm to obtain a third dung beetle optimization algorithm.

[0185] The expansion unit 548 is used to expand the solution space of the third dung beetle optimization algorithm by using Cauchy-Gaussian mutation to obtain an improved dung beetle optimization algorithm.

[0186] The fourth optimization unit 558 is used to optimize the hyperparameters of the deep neural network by using the improved dung beetle optimization algorithm.

[0187] It should be noted that, for the specific optimization process corresponding to each optimization unit, reference can be made to the detailed description in the method embodiment of the first aspect above, which will not be repeated here.

[0188] In some embodiments, when the concentration of the alcoholamine solution is between 1.5% and 12%, the device 500 uses interpolation conditions to simulate the mass transfer of carbon dioxide bubbles in the alcoholamine solution; when the concentration of the alcoholamine solution is between 4% and 8%, the device 500 uses extrapolation conditions to simulate the mass transfer of carbon dioxide bubbles in the alcoholamine solution.

[0189] It can be understood that the specific details and corresponding technical effects of the simulation device provided in the embodiment of the present disclosure can refer to the specific details and technical effects of the simulation method embodiment of the first aspect mentioned above, and will not be repeated here.

[0190] In a third aspect, an embodiment of the present disclosure further provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method described in any one of the first aspects is implemented.

[0191] Specifically, Figure 4 A schematic diagram of the physical structure of an electronic device provided in an embodiment of the present invention, such as Figure 4As shown, the electronic device 003 includes a processor 301, a memory 302 and a bus 303. The processor 301 and the memory 302 communicate with each other via the bus 303.

[0192] The processor 301 is used to call the program instructions in the memory 302 to execute the methods provided by the above method embodiments.

[0193] In a fourth aspect, an embodiment of the present disclosure further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program of the method described in any one of the first aspects above.

[0194] In a fifth aspect, an embodiment of the present disclosure further provides a computer program product, comprising a computer program / instructions, which, when executed by a processor, implement the steps of any one of the methods described in the first aspect above.

[0195] Those skilled in the art will appreciate that the embodiments of the present disclosure may be provided as methods, systems, or computer program products. Therefore, the present disclosure may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present disclosure may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0196] The present disclosure is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present disclosure. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0197] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0198] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0199] Specific embodiments are used in the present disclosure to illustrate the principles and implementation methods of the present disclosure. The description of the above embodiments is only used to help understand the method of the present disclosure and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present disclosure, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present disclosure.

Claims

1. A method for simulating mass transfer of carbon dioxide bubbles in an alcoholamine solution, characterized in that: The method comprises: Outputting first operating condition data of mass transfer of carbon dioxide bubbles in an alcoholamine solution according to a computational fluid dynamics model; According to the first operating condition data, second operating condition data is obtained through a preset data conversion network; wherein the grid accuracy of the second operating condition data is greater than the grid accuracy of the first operating condition data; According to the second operating condition data, judging whether the second operating condition data meets the accuracy requirement through a preset calibration network, and obtaining a judgment result; According to the judgment result, relevant operating data of the mass transfer of carbon dioxide bubbles in the alcoholamine solution are output.

2. The method according to claim 1, characterized in that The first operating condition data includes at least one of the following data: concentration data of carbon dioxide bubbles, concentration data of alcohol amine solution, pressure data inside carbon dioxide bubbles, phase field variables, and flow field velocity data.

3. The method according to claim 1, characterized in that The step of determining, based on the second operating condition data, whether the second operating condition data meets the accuracy requirement through a preset calibration network to obtain a determination result includes: Calculating an actual error integral value of the second operating condition data; Inputting the error integral value of the first operating condition data into the preset calibration network to obtain a standard error integral value of the first operating condition data; The actual error integral value is compared with the standard error integral value, and whether the second operating condition data meets the accuracy requirement is determined according to the comparison result to obtain a determination result.

4. The method according to claim 1, characterized in that: According to the judgment result, the relevant working condition data of the mass transfer of carbon dioxide bubbles in the alcohol amine solution is output, including: If the judgment result is that the second operating condition data does not meet the accuracy requirement, the following steps are repeatedly performed: Inputting the second operating condition data into the computational fluid dynamics model to obtain an output result, and using the output result as the first operating condition data of mass transfer of carbon dioxide bubbles in an alcoholamine solution; According to the first operating condition data, obtaining second operating condition data through a preset data conversion network; According to the second operating condition data, judging whether the second operating condition data meets the accuracy requirement through a preset calibration network, and obtaining a judgment result; According to the judgment result, outputting relevant working condition data of mass transfer of carbon dioxide bubbles in alcoholamine solution; Until the judgment result is that the second operating condition data meets the accuracy requirement, the relevant operating condition data of the mass transfer of carbon dioxide bubbles in the alcohol amine solution is output.

5. The method according to claim 1, characterized in that The method further comprises: Acquire operating condition sample data with different grid precisions, wherein the operating condition sample data includes first operating condition sample data and second operating condition sample data, and the grid precision of the second operating condition sample data is greater than the grid precision of the first operating condition sample data; The recurrent neural network is trained according to the operating condition sample data to obtain the preset data conversion network.

6. The method according to claim 1, characterized in that The first operating data of mass transfer of carbon dioxide bubbles in alcohol amine solution is output according to the computational fluid dynamics model, including: The Euler-Euler method is used to calculate the viscous stress of the gas and liquid phases based on mass conservation; The surface tension of gas and liquid phases is calculated using the phase field model; According to the viscous stress and the surface tension, first operating condition data of mass transfer of carbon dioxide bubbles in the alcoholamine solution is output through diffusion mass transfer, convection mass transfer and chemical reaction.

7. The method according to claim 6, characterized in that The method of using a phase field model to calculate the surface tension of the gas phase and the liquid phase includes: The square of the grid width at the maximum phase interface or twice the grid width at the maximum phase interface is used as the phase field interface thickness; The surface tension of the gas phase and the liquid phase is calculated using a phase field model based on the phase field interface thickness.

8. The method according to claim 1, characterized in that The method further comprises: According to the population size and the fitness of dung beetles, the acceptance degree of the optimal solution of the young ball individuals and the stealing dung beetle individuals in the dung beetle optimization algorithm is adjusted adaptively according to the fitness, and the first dung beetle optimization algorithm is obtained; The global search mechanism of the sparrow search algorithm is integrated with the first dung beetle optimization algorithm to obtain a second dung beetle optimization algorithm; Based on the second dung beetle optimization algorithm, adaptive distributed perturbation is performed on the foraging of small dung beetles to obtain a third dung beetle optimization algorithm; The solution space of the third dung beetle optimization algorithm is expanded by using Cauchy-Gauss mutation to obtain an improved dung beetle optimization algorithm; An improved dung beetle optimization algorithm is used to optimize the hyperparameters of a deep neural network, wherein the deep neural network includes the computational fluid dynamics model, the preset data conversion network and / or the preset calibration network.

9. The method according to claim 1, characterized in that: When the concentration of the alcoholamine solution is between 1.5% and 12%, the interpolation condition is used to simulate the mass transfer of carbon dioxide bubbles in the alcoholamine solution. When the concentration of the alcohol amine solution is between 4% and 8%, the extrapolation conditions are used to simulate the mass transfer of carbon dioxide bubbles in the alcohol amine solution.

10. A simulation device for mass transfer of carbon dioxide bubbles in an alcohol amine solution, characterized in that: The device comprises: An output module, used for outputting first operating condition data of mass transfer of carbon dioxide bubbles in an alcohol amine solution according to a computational fluid dynamics model; A conversion module, used to obtain second operating condition data through a preset data conversion network according to the first operating condition data; wherein the grid accuracy of the second operating condition data is greater than the grid accuracy of the first operating condition data; A judgment module, used for judging whether the second operating condition data meets the accuracy requirement through a preset calibration network according to the second operating condition data, and obtaining a judgment result; The simulation module is used to output relevant operating data of mass transfer of carbon dioxide bubbles in the alcohol amine solution according to the judgment result.

11. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 9 is implemented.

12. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program for executing the method according to any one of claims 1 to 9.

13. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 9 are implemented.

Images