CN115081293A - Multi-objective optimization method of bubble column carbon capture process based on proxy model - Google Patents

Multi-objective optimization method of bubble column carbon capture process based on proxy model Download PDF

Info

Publication number
CN115081293A
CN115081293A CN202210798202.XA CN202210798202A CN115081293A CN 115081293 A CN115081293 A CN 115081293A CN 202210798202 A CN202210798202 A CN 202210798202A CN 115081293 A CN115081293 A CN 115081293A
Authority
CN
China
Prior art keywords
model
bubble column
objective optimization
optimization method
proxy
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202210798202.XA
Other languages
Chinese (zh)
Inventor
邢磊
江海
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jiangsu University
Original Assignee
Jiangsu University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Jiangsu University filed Critical Jiangsu University
Priority to CN202210798202.XA priority Critical patent/CN115081293A/en
Publication of CN115081293A publication Critical patent/CN115081293A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/23—Design optimisation, verification or simulation using finite element methods [FEM] or finite difference methods [FDM]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/25—Design optimisation, verification or simulation using particle-based methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/12—Computing arrangements based on biological models using genetic models
    • G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2113/00—Details relating to the application field
    • G06F2113/08—Fluids

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Software Systems (AREA)
  • Biophysics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Health & Medical Sciences (AREA)
  • Mathematical Physics (AREA)
  • Geometry (AREA)
  • Computing Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Computer Hardware Design (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Computational Linguistics (AREA)
  • Medical Informatics (AREA)
  • Molecular Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Physiology (AREA)
  • Genetics & Genomics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention provides a multi-objective optimization method of a bubble column carbon capture process based on a proxy model, which comprises the following steps: constructing a physical simulation model of the packed bubble column reactor through COMSOL Multiphysics software; according to preselectionThe optimization target and the design variable are sampled in the design domain by the optimal Latin super-legislation; constructing an adaptive hybrid agent model, performing model training by using a simulation result, and using R 2 RMSE and the like to evaluate the accuracy and stability of the model and to carry out Sobol global sensitivity analysis on different design variables; performing multi-objective optimization solution on the selected agent model by adopting a non-dominant genetic algorithm; drawing a Pareto frontier chart according to a solution set obtained by NSGA-II to obtain a multi-objective optimization optimal solution; and outputting a multi-objective optimization scheme meeting the pre-specified process requirement to obtain the optimal balanced solution of the energy consumption and the carbon dioxide capture rate required by the process and the corresponding optimal operation condition, wherein the data have guiding significance for the subsequent industrial application of the process.

Description

Multi-objective optimization method of bubble column carbon capture process based on proxy model
Technical Field
The invention belongs to the technical field of optimization of a carbon dioxide capture process, and particularly relates to a multi-objective optimization method of a bubble column carbon capture process based on a proxy model.
Background
The need for a "net zero" target has prompted the development of Carbon capture, utilization, and storage (CCUS) technologies, where atmospheric Carbon dioxide, in the billions of grades, may be absorbed annually by the Enhanced Weathering (EW) action of mineral particles and sequestered in the ocean in the form of bicarbonate and carbonate salts. However, weathering in natural conditions is very slow and requires the use of chemical reactors to accelerate the process under controlled conditions. Therefore, finding a suitable chemical reactor and suitable operating conditions for the reactor are key to whether the process can be carried out industrially on a large scale. Recent studies have shown that Packed Bubble Columns (PBC) can be used as reactors for capturing carbon dioxide based on the action of mineral EW. Before the bubble column is filled for capturing carbon dioxide, performance indexes such as carbon dioxide capturing rate, energy consumption and the like need to be optimally designed so as to obtain optimal design variables meeting different targets.
Generally speaking, the optimization of an industrial process with long experiment time and high cost is usually to construct a physical simulation model of the process for calculation and research, but for multivariate and multi-objective optimization, thousands or even millions of simulation tasks are often required, and a method for directly solving the physical simulation model in the whole design domain to obtain an optimal solution is not feasible.
Disclosure of Invention
In view of the above technical problems, an object of one embodiment of the present invention is to provide a multi-objective optimization method for a bubble column carbon capture process based on a proxy model, so that the method can be used for solving the optimization problem of a carbon dioxide capture process of a packed bubble column and obtaining a required multi-objective optimization scheme.
The invention relates to a multi-objective optimization method for a process of capturing carbon dioxide by mineral particle enhanced weathering of a packed bubble column based on a proxy model, wherein a hybrid proxy model is used for simplifying a physical model to obtain a multi-objective optimization scheme. The proxy model is a simplified model which is based on statistical basis, has small calculated amount and quick solution, but has a calculated result close to that of a physical model and higher precision. The invention applies the hybrid agent model to the optimization of the carbon dioxide trapping process, greatly reduces the calculation amount of multivariate optimization, and simultaneously improves the accuracy of the simulation optimization result.
The invention provides a multi-objective optimization method for a process for capturing carbon dioxide by a packed bubble column based on a proxy model, which comprises the following steps: s1, constructing a multi-physical-field simulation model: constructing a physical simulation model of the packed bubble column reactor by COMSOL Multiphysics software, wherein the physical simulation model comprises reaction kinetics of carbon dioxide capture and a mass transfer process of carbon dioxide at a gas-liquid interface in the reactor; s2, sampling of the proxy model data set: sampling by an Optimal Latin Hypercube Sampling (OLHS) within the design domain according to a preselected optimization objective and design variables; s3, constructing and evaluating a proxy model: constructing an adaptive hybrid agent model, performing model training by using a simulation result, and using R 2 And RMSE and the like to evaluate the accuracy and stability of the model; s4, global sensitivity analysis: performing Sobol global sensitivity analysis on different design variables; s5, proxy model optimization solution: performing multi-objective optimization solution on the selected agent model by adopting a non-dominant genetic algorithm II (NSGA-II); s6, drawing a multi-target trade-off diagram: drawing a Pareto frontier chart according to a solution set obtained by NSGA-II to obtain a multi-objective optimization optimal solution; s7, outputting an optimization scheme: and outputting a multi-objective optimization scheme meeting the pre-specified process requirements. The multi-objective optimization method for the process of capturing carbon dioxide by the packed bubble tower based on the proxy model can obtain the optimal balanced solution of the energy consumption and the carbon dioxide capturing rate required by the process and the corresponding optimal operating conditions (design variables), and the data pairsThe subsequent industrial application of the process has guiding significance.
The present invention achieves the above-described object by the following technical means.
A multi-objective optimization method of a bubble column carbon capture process based on a proxy model comprises the following steps:
step S1, constructing a multi-physical-field simulation model: construction of mineral particle-based enhanced weathering for CO capture 2 The physical simulation model of the series-connected packed bubble column reactor comprises reaction kinetics of carbon dioxide capture and a mass transfer process of carbon dioxide at a gas-liquid interface in the reactor;
step S2, sampling of the proxy model data set: according to the capture of CO 2 The optimized objective function and the design variable preselected by the serial-connection packed bubble column reactor process are sampled by an optimal Latin hypercube sampling method in the design domain;
step S3, constructing and evaluating a proxy model: constructing an adaptive hybrid agent model according to the simulation result of the sampling point in the step S2, performing model training by using the simulation result, and using R 2 And RMSE estimates the accuracy and stability of the model;
step S4, global sensitivity analysis: after the precision of the proxy model reaches the standard in the step S3, carrying out Sobol global sensitivity analysis on different design variables;
step S5, proxy model optimization solution: after the sensitivity analysis of the step S4, performing multi-objective optimization solution on the selected agent model by adopting a non-dominant genetic algorithm;
step S6, drawing a multi-target trade-off diagram: drawing a Pareto frontier chart according to a solution set obtained by the non-dominant genetic algorithm in the step S5 to obtain an optimal solution;
step S7, outputting an optimization scheme: finding the CO with the maximum corresponding objective function value of the optimal design variable X corresponding to the optimal solution in the non-dominant genetic algorithm by comparing the optimal solution obtained in the step S6 2 Capturing the speed and the minimum process energy consumption, and outputting a multi-objective optimization scheme meeting the pre-specified process requirements.
In the above scheme, the stepsStep S1 physical simulation model of filling bubble column reactor is based on enhanced weathering of mineral particles for CO capture 2 The physical simulation model of the series packed bubble column reactor of (1) comprises at least two packed bubble column reactors, the first reactor is a packed bubble column based on seawater, and the second reactor is a packed bubble column based on fresh water.
Further, the step S1 is to construct a multi-physics simulation model based on a finite element method in COMSOL Multiphysics software.
In the above solution, the objective function in step S2:
f(X)＝[f 1 (X),f 2 (X)] T
wherein f is 1 (X) represents maximum CO 2 Capture rate, f 2 (X) represents the minimum process energy consumption;
the design variable X:
X＝[x 1 ,x 2 ,x 3 ,x 4 ,x 5 ,x 6 ,x 7 ,x 8 ] T ,X∈K
wherein x is 1 ,x 2 ,x 3 And x 4 Respectively the mineral particle size, bed height, gas flow rate and liquid flow rate in the seawater-filled bubble column reactor; x is a radical of a fluorine atom 5 ,x 6 ,x 7 And x 8 Corresponding to the mineral particle size, bed height, gas flow rate and liquid flow rate in the fresh water-filled bubble column reactor, K is the design domain of the process to be optimized and is expressed as:
K＝[[1,10],[1,10],[0.0001,0.001],[0.0001,0.001],[1,10],[1,10],[0.0001,0.001],[0.001,0.01]] T 。
in the above solution, the hybrid proxy model of step S3 includes at least two single proxy models.
Further, the step S3 hybrid proxy model includes a polynomial response surface, kriging, radial basis functions, and a support vector machine.
In the above scheme, R in the step S3 2 And RMSE is obtained from the following equation:
Figure BDA0003736489120000031
Figure BDA0003736489120000032
wherein N is the number of the test sets,
Figure BDA0003736489120000033
for the prediction value of the hybrid agent model, y i For the test set obtained by the physical model simulation,
Figure BDA0003736489120000034
the mean value of the test set obtained by physical model simulation.
In the above scheme, the process of the non-dominant genetic algorithm in the proxy model optimization solution in step S5 is as follows:
step S1), initializing the population;
step S2), obtaining an initial population P through rapid non-dominated sorting, selecting, crossing and mutation operations;
step S3), merging the parent population and the child population into 2P, and then calculating to obtain next generation population individuals through rapid non-dominated sorting and crowding degree;
step S4), continuing to generate the next generation according to genetic operation until reaching the maximum algebra Gen of evolution max Stopping;
step S5), the optimal solution set is output.
Further, the crossover distribution index and the mutation distribution index of the crossover and mutation operations in step S2) are both 20.
In the above scheme, the multi-objective optimization scheme is an optimization scheme that achieves the maximum capture rate with the minimum energy consumption.
Compared with the prior art, the invention has the beneficial effects that:
the invention simplifies the physical model by constructing the hybrid agent model so as to obtain a multi-objective optimization scheme, fully considers the coupling relation between design variables and the correlation characteristic between objective functions, adopts a rapid non-dominated multi-objective optimization algorithm (NSGA-II) with an elite retention strategy, realizes the optimization of multiple variables and multiple objectives, finally obtains the optimization scheme which simultaneously meets the minimum energy consumption and the maximum capture rate, and provides more accurate support data for the use of the subsequent process.
Drawings
FIG. 1 is a flow chart of a multi-objective optimization method for a packed bubble column carbon dioxide capture process based on a surrogate model according to an embodiment of the invention;
FIG. 2 is a flow chart of the NSGA-II optimization algorithm in step (S5) of FIG. 1;
FIG. 3 is a schematic diagram of a series packed bubble column reactor in an example of an embodiment of the invention;
FIG. 4 is a diagram illustrating the R of the hybrid proxy model for two performance indicators according to an embodiment of the present invention 2 And RMSE analysis plots;
FIG. 5 is a different variable global Sobol sensitivity analysis chart according to an embodiment of the present invention;
fig. 6 is a dual target optimized Pareto frontier obtained in an example of an embodiment of the present invention.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the drawings are illustrative and intended to be illustrative of the invention and are not to be construed as limiting the invention.
In the description of the present invention, it is to be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "front", "rear", "left", "right", "upper", "lower", "axial", "radial", "vertical", "horizontal", "inner", "outer", etc. indicate orientations and positional relationships based on those shown in the drawings, and are used merely for convenience in describing the present invention and for simplicity in description, and do not indicate or imply that the device or element so referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus should not be considered as limiting the present invention. Furthermore, the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the present invention, "a plurality" means two or more unless specifically defined otherwise.
In the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," "secured," and the like are to be construed broadly and can, for example, be fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood by those skilled in the art according to specific situations.
Fig. 1 shows a preferred embodiment of the multi-objective optimization method for the bubble column carbon capture process based on the proxy model, which comprises the following steps:
step S1, constructing a multi-physical-field simulation model: construction of mineral particle-based enhanced weathering for CO capture 2 The physical simulation model of the series-connected packed bubble column reactor comprises reaction kinetics of carbon dioxide capture and a mass transfer process of carbon dioxide at a gas-liquid interface in the reactor;
step S2, proxy model data set sampling: according to the capture of CO 2 The optimized objective function and the design variable preselected by the series-connection packed bubble column reactor process are sampled in a design domain by an optimal Latin hypercube sampling method;
step S3, constructing and evaluating a proxy model:constructing an adaptive hybrid agent model according to the simulation result of the sampling point in the step S2, performing model training by using the simulation result, and using R 2 And RMSE estimates the accuracy and stability of the model;
step S4, global sensitivity analysis: after the precision of the proxy model reaches the standard in the step S3, carrying out Sobol global sensitivity analysis on different design variables;
step S5, proxy model optimization solution: after the sensitivity analysis of the step S4, performing multi-objective optimization solution on the selected agent model by adopting a non-dominant genetic algorithm;
step S6, drawing a multi-target trade-off solution graph: drawing a Pareto frontier chart according to a solution set obtained by the non-dominant genetic algorithm in the step S5 to obtain an optimal solution;
step S7, outputting an optimization scheme: finding CO with the maximum corresponding objective function value of the optimal design variable X corresponding to the optimal solution in the non-dominant genetic algorithm by comparing the optimal solution obtained in the step S6 2 Capturing the speed and the minimum process energy consumption, and outputting a multi-objective optimization scheme meeting the pre-specified process requirements.
According to this embodiment, preferably, the building of the multi-physics simulation model in step S1: FIG. 3 illustrates Enhanced Weathering (EW) CO capture based on mineral particles 2 A schematic diagram of a series packed bubble column reactor process according to (1), wherein the first reactor is a seawater-based packed bubble column and the second reactor is a fresh water-based packed bubble column, the mineral particles used in both are calcite, and the inlet gas is assumed to contain 15% CO 2 The power plant exhaust gas of (1).
The waste gas of the power plant to be treated is firstly introduced into a first reactor, a bubble tower is filled with seawater for primary capture, and CO is generated along with the reaction 2 The content of (A) is gradually reduced because only CO can be captured by the enhanced weathering in the seawater-filled bubble column 2 High content of mixed gases, non-captured CO 2 It is discharged together with the exhaust gas. The exhaust gas is then introduced into a seawater-based packed bubble column of a fresh water reactor of a second reactor for secondary capture, and the main chemical reaction equations involved in the whole processComprises the following steps:
Figure BDA0003736489120000061
Figure BDA0003736489120000062
Figure BDA0003736489120000063
Figure BDA0003736489120000064
Figure BDA0003736489120000065
Figure BDA0003736489120000066
Figure BDA0003736489120000067
based on the process flow, a corresponding physical simulation model is constructed by a Finite Element Method (FEM) in COMSOL Multiphysics software. The model comprises the reaction kinetics of carbon dioxide capture and the mass transfer process of carbon dioxide at a gas-liquid interface in a reactor, and one-dimensional geometric space and time related variables are obtained by solving a control equation consisting of a plurality of Partial Differential Equations (PDEs). Where the computational domain is treated as a homogeneous porous medium, where the actual geometry of the individual particles is not taken into account. And (3) adopting an MUMPS time correlation solver, setting default parameters, setting calculation tolerance as physical control, and setting relative tolerance as 0.001.
The step S2 represents the sampling of the model data set, which preferably begins with the determination of the optimization objectives and design variables for the process, which in this embodiment, has a total of two objective functions:
f(X)＝[f 1 (X),f 2 (X)] T
wherein f is 1 (X) represents maximum CO 2 Capture Rate (CO captured hourly 2 Mass), f 2 (X) represents the minimum process energy consumption (energy required per unit mass of carbon dioxide captured).
The design variables X studied were eight in total:
X＝[x 1 ,x 2 ,x 3 ,x 4 ,x 5 ,x 6 ,x 7 ,x 8 ] T ,X∈K
wherein x 1 ,x 2 ,x 3 And x 4 The particle size of mineral, the height of bed layer, the gas flow rate and the liquid flow rate in the seawater-filled bubble column reactor are respectively; x is the number of 5 ,x 6 ,x 7 And x 8 Corresponding to the mineral particle size, bed height, gas flow rate and liquid flow rate in the fresh water reactor. K is the design domain of the flow to be optimized, and is expressed as:
K＝[[1,10],[1,10],[0.0001,0.001],[0.0001,0.001],[1,10],[1,10],[0.0001,0.001],[0.001,0.01]] T because the difference between the values of different variables is large, all the variables are normalized to be in the range of 0,1]. After the design variables and the research targets are determined, sampling is carried out in a design domain through an optimal Latin hypercube sampling method (OLHS), and the proportion of a training set (M) and a testing set (N) in an obtained data set is 4: 1.
In the step S3, preferably, an Extended adaptive hybrid agent model (E-AHF) is constructed according to the simulation result of the OLHS sampling point, and the hybrid agent model in this example includes four single agent models (polynomial response surface, kriging, radial basis function, and support vector machine). After model training, calculate R 2 And RMSE evaluation of model accuracy and stabilityEstimate, calculate the formula as follows:
Figure BDA0003736489120000071
Figure BDA0003736489120000072
wherein N is the number of test sets;
Figure BDA0003736489120000073
is a predicted value of the hybrid agent model;
Figure BDA0003736489120000074
average value, R, of test set obtained for physical model simulation 2 The closer to 1, the smaller the RMSE, the higher the prediction accuracy and the better the stability of the corresponding model. If the obtained model has poor performance, the accuracy of the model needs to be further improved by adding a data set.
In this embodiment, a total of 400 sets of data are ultimately used to build the desired proxy model. R of model under different sample points 2 And the values for RMSE are shown in the following table:
TABLE 1R of PBC tandem reactor E-AHF model based on different data sets 2 And RMSE
Figure BDA0003736489120000075
CR–CO 2 Capture Rate, EC-energy consumption
FIG. 4 shows more intuitively the R of the model after adding sample points 2 And changes in the value of RMSE. From the analysis of table 1 and fig. 4, the following conclusions can be drawn:
1) r of the model as the number of sample points increases 2 Will gradually increase, the RMSE will gradually decrease;
2) when the number of sample points is small, the change of the number of sample points is relative to R 2 And RMSE effects are more pronounced;
3) after the number of samples is more than 300, the number of samples is increased to CO 2 R of capture rate 2 And RMSE is less affected;
4) when the number of data sets reaches 400, CO 2 Capture rate and energy consumption of 2 All exceed 0.95, and the surface model has better prediction accuracy. CO in the finally obtained E-AHF proxy model 2 R of capture rate 2 And RMSE 0.9686 and 0.0061, R energy consumption 2 And RMSE 0.9720 and 1.2955, respectively. The results show that both models have better prediction accuracy and stability, and the accuracy of the obtained hybrid agent model can meet the requirement of realizing multi-objective optimization.
The step S4 global sensitivity analysis, preferably, after the proxy model accuracy reaches the standard, Sobol global sensitivity analysis is performed on different design variables to obtain a global sensitivity index S T As shown in table 2 below:
TABLE 2 Sobol Global sensitivity index values for different design variables
Figure BDA0003736489120000081
From the results in table 2, fig. 5 is plotted. By analyzing table 2 and fig. 5, the effect of different design variables on different objective functions is different. For CO 2 The design variable with the greatest effect on capture rate is the gas flow rate (x) of the seawater-filled packed column 3 ) It is to CO 2 The Sobol global sensitivity index for the capture rate is 0.55710; secondly, the liquid flow rate (x) of the fresh water filled bubble column 8 ) It is to CO 2 The Sobol global sensitivity index for the capture rate is 0.34641. And for energy consumption, the bed height (x) of the bubble column is filled with fresh water 6 ) And the bed height (x) of the seawater-filled packed tower 2 ) Are far more significant than other design variables, their Sobol global sensitivity indexes to energy consumption are 0.50203 and 0.25873, respectively. A larger global sensitivity index indicates that the result of the objective function will change significantly after these design variables have changed.
In the step S5, the agent model is optimized and solved, preferably, after the sensitivity analysis, a non-dominant genetic algorithm (NSGA-II) is used to perform multi-objective optimization and solution on the selected agent model.
FIG. 2 shows the main processes of NSGA-II:
(1) initializing a population;
(2) obtaining an initial population P through rapid non-dominated sorting, selecting, crossing and mutation operations;
(3) merging the parent population and the offspring population into 2P, and then calculating to obtain next generation population individuals through rapid non-dominated sorting and crowding degree;
(4) continuing to generate the next generation based on genetic manipulation until reaching the maximum generation of evolution (Gen) max ) Stopping;
(5) outputting an optimal solution set;
in this example, it is preferable that the cross distribution index and the variation distribution index are both 20, the population number P is 35, and the maximum evolution generation Gen is max Is 20 generations. The two objective functions used in the non-dominated sorting in this example are respectively f determined in step (S2) 1 (X)，f 2 (X), i.e. maximum CO 2 Capture rate and minimum energy consumption.
The step S6 is performed by using a multi-objective trade-off map, preferably, a Pareto solution set distribution map is performed according to the solution set obtained by NSGA-II, the red part in fig. 6 is the optimal solution set obtained by NSGA-II, and the black point is the data set used for the hybrid agent model before. Ten sets of Pareto optimal solutions from the NSGA-II algorithm are shown in table 3.
TABLE 3 partial Pareto solution set from NSGA-II
Figure BDA0003736489120000091
It can be seen from fig. 6 that the resulting Pareto front is superior to the simulation data used previously.
The step S7 is used for outputting an optimization scheme, preferably, multi-target is obtained through a Pareto frontier chartAnd (4) carrying out Pareto optimal solution, and recording a corresponding objective function value. And then, corresponding data of the optimal design variable X corresponding to the optimal solution in the NSGA-II is found out by contrasting the given optimal solution. In this case, energy consumption and CO are found 2 Optimal trade-off for capture rate, i.e. capturing more CO with as little energy as possible 2 The corresponding design variable value in the case of (2).
In this example, the optimal solution selected after comprehensively analyzing the Pareto distribution diagram is the green dot in the diagram, i.e., the 4 th group of optimal solutions in table 3. The corresponding optimum design variable values are:
X＝[x 1 ,x 2 ,x 3 ,x 4 ,x 5 ,x 6 ,x 7 ,x 8 ] T ＝[2.2315,4.8145,0.001,0.0007,8.3635,1,0.00047,0.00835] T wherein x 1 ,x 2 ,x 3 And x 4 Respectively the mineral particle size, bed height, gas flow rate and liquid flow rate in the seawater-filled bubble column reactor; x is the number of 5 ,x 6 ,x 7 And x 8 Corresponding to the mineral particle size, bed height, gas and liquid flow rates in the fresh water reactor, corresponding to the energy consumption and CO 2 The capture rates were 5.58933MJ kg -1 CO 2 And 0.11756kg h -1 . And finally, outputting the scheme into a final multi-objective optimization scheme.
According to the invention, the energy consumption and capture rate performance indexes of the process for capturing carbon dioxide under the ore weathering action of the packed bubble tower are associated with design variables such as flow rate and height of a bed by constructing a hybrid agent model, the coupling relation among the design variables and the association characteristics among the performance indexes are fully considered, and a fast non-dominated multi-objective optimization algorithm (NSGA-II) with an elite retention strategy is adopted, so that the optimization of multiple variables and multiple targets is realized, the optimization scheme for achieving the maximum capture rate at the minimum energy consumption can be finally obtained, and more accurate support data is provided for the subsequent process.
It should be understood that although the present description has been described in terms of various embodiments, not every embodiment includes only a single embodiment, and such description is for clarity purposes only, and those skilled in the art will recognize that the embodiments described herein may be combined as suitable to form other embodiments, as will be appreciated by those skilled in the art.
The above-listed detailed description is only a specific description of a possible embodiment of the present invention, and they are not intended to limit the scope of the present invention, and equivalent embodiments or modifications made without departing from the technical spirit of the present invention should be included in the scope of the present invention.

Claims (10)

1. A multi-objective optimization method of a bubble column carbon capture process based on a proxy model is characterized by comprising the following steps:
step S1, constructing a multi-physical-field simulation model: construction of mineral particle-based enhanced weathering for CO capture 2 The physical simulation model of the series packed bubble column reactor;
step S2, proxy model data set sampling: according to the capture of CO 2 The optimized objective function and the design variable preselected by the series-connected packed bubble column reactor process are sampled in the design domain by an optimal Latin hypercube sampling method, and the sampling points are input into the physical simulation model constructed in the step S1 to obtain the simulation result of the corresponding sampling points;
step S3, constructing and evaluating a proxy model: constructing an adaptive hybrid agent model according to the simulation result of the sampling point in the step S2, training the hybrid agent model by using the simulation result, and using R 2 And the RMSE evaluates the accuracy and stability of the hybrid agent model;
step S4, global sensitivity analysis: after the accuracy and the stability of the hybrid agent model reach the standard in the step S3, carrying out Sobol global sensitivity analysis on different design variables;
step S5, proxy model optimization solution: after the sensitivity analysis of the step S4, performing multi-objective optimization solution on the selected hybrid agent model by adopting a non-dominant genetic algorithm;
step S6, drawing a multi-target trade-off diagram: drawing a Pareto frontier chart according to a solution set obtained by the non-dominant genetic algorithm in the step S5 to obtain an optimal solution;
step S7, outputting an optimization scheme: finding CO with the maximum corresponding objective function value of the optimal design variable X corresponding to the optimal solution in the non-dominant genetic algorithm by comparing the optimal solution obtained in the step S6 2 Capturing the speed and the minimum process energy consumption, and outputting a multi-objective optimization scheme meeting the pre-specified process requirements.
2. The multi-objective optimization method for bubble column carbon capture process based on proxy model as claimed in claim 1, wherein the physical simulation model of step S1 filling the bubble column reactor is based on enhanced weathering of mineral particles for capturing CO 2 The physical simulation model of a series packed bubble column reactor of (1), the series packed bubble column reactor comprising at least two packed bubble column reactors, the packed bubble column reactor comprising a seawater-based packed bubble column and a fresh water-based packed bubble column.
3. The multi-objective optimization method for bubble column carbon capture process based on surrogate model as claimed in claim 2, wherein the step S1 multi-physical field simulation model is constructed based on finite element method in COMSOL Multiphysics software; the physical simulation model comprises the reaction kinetics of carbon dioxide capture and the mass transfer process of carbon dioxide at a gas-liquid interface in the reactor.
4. The multi-objective optimization method for bubble column carbon capture process based on surrogate model as claimed in claim 2, wherein the objective function in step S2 is:
f(X)＝[f 1 (X),f 2 (X)] T
wherein f is 1 (X) represents maximum CO 2 Capture rate, f 2 (X) represents the minimum process energy consumption;
the design variable X:
X＝[x 1 ,x 2 ,x 3 ,x 4 ,x 5 ,x 6 ,x 7 ,x 8 ] T ,X∈K
wherein x is 1 ,x 2 ,x 3 And x 4 Respectively the mineral particle size, bed height, gas flow rate and liquid flow rate in the seawater-filled bubble column reactor; x is the number of 5 ,x 6 ,x 7 And x 8 Corresponding to the mineral particle size, bed height, gas flow rate and liquid flow rate in the freshwater-packed bubble column reactor, K is the design field of the process to be optimized and is expressed as:
K＝[[1,10],[1,10],[0.0001,0.001],[0.0001,0.001],[1,10],[1,10],[0.0001,0.001],[0.001,0.01]] T 。
5. the multi-objective optimization method for a bubble column carbon capture process based on a proxy model of claim 2, wherein the step S3 hybrid proxy model comprises at least two single proxy models.
6. The multi-objective optimization method for bubble column carbon capture process based on proxy model of claim 5, wherein the step S3 hybrid proxy model comprises polynomial response surface, kriging, radial basis function and support vector machine.
7. The multi-objective optimization method for bubble column carbon capture process based on surrogate model as claimed in claim 1, wherein R in the step S3 2 And RMSE is obtained from the following equation:
Figure FDA0003736489110000021
Figure FDA0003736489110000022
wherein N is the number of the test sets,
Figure FDA0003736489110000023
for the prediction value of the hybrid agent model, y i For the test set obtained by the physical model simulation,
Figure FDA0003736489110000024
the mean value of the test set obtained by physical model simulation.
8. The multi-objective optimization method for bubble column carbon capture process based on surrogate model according to claim 1, wherein the process of the non-dominant genetic algorithm in the surrogate model optimization solution in step S5 is as follows:
step S1), initializing the population;
step S2), obtaining an initial population P through rapid non-dominated sorting, selecting, crossing and mutation operations;
step S3), merging the parent population and the child population into 2P, and then calculating to obtain next generation population individuals through rapid non-dominated sorting and crowding degree;
step S4), the next generation is continuously generated according to the genetic operation until the maximum generation Gen of evolution is reached max Stopping;
step S5), the optimal solution set is output.
9. The multi-objective optimization method for bubble column carbon capture process based on surrogate model as claimed in claim 8, wherein the crossover distribution index and the variation distribution index of the crossover and variation operations in step S2) are both 20.
10. The multi-objective optimization method for a bubble column carbon capture process based on a surrogate model of claim 1, wherein the multi-objective optimization scheme comprises achievable minimum energy consumption and maximum capture rate and corresponding values of optimal design variable X.
CN202210798202.XA 2022-07-08 2022-07-08 Multi-objective optimization method of bubble column carbon capture process based on proxy model Pending CN115081293A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210798202.XA CN115081293A (en) 2022-07-08 2022-07-08 Multi-objective optimization method of bubble column carbon capture process based on proxy model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210798202.XA CN115081293A (en) 2022-07-08 2022-07-08 Multi-objective optimization method of bubble column carbon capture process based on proxy model

Publications (1)

Publication Number Publication Date
CN115081293A true CN115081293A (en) 2022-09-20

Family

ID=83257990

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210798202.XA Pending CN115081293A (en) 2022-07-08 2022-07-08 Multi-objective optimization method of bubble column carbon capture process based on proxy model

Country Status (1)

Country Link
CN (1) CN115081293A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115795814A (en) * 2022-11-10 2023-03-14 宝鸡文理学院 Sensitivity parameter calibration method and system of dynamic vegetation model

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115795814A (en) * 2022-11-10 2023-03-14 宝鸡文理学院 Sensitivity parameter calibration method and system of dynamic vegetation model

Similar Documents

Publication Publication Date Title
Reagen et al. A case for efficient accelerator design space exploration via bayesian optimization
Yan et al. Modeling of solid oxide fuel cell (SOFC) electrodes from fabrication to operation: Microstructure optimization via artificial neural networks and multi-objective genetic algorithms
Iooss et al. Numerical studies of the metamodel fitting and validation processes
Wan et al. Machine learning-assisted design of flow fields for redox flow batteries
Chen et al. Geometry optimization and pressure analysis of a proton exchange membrane fuel cell stack
CN114447378B (en) Parameter optimization method for proton exchange membrane fuel cell
CN112070356B (en) Method for predicting carbonization resistance of concrete based on RF-LSSVM model
CN104933245A (en) Ship reactor shielding design optimization method based on neural network and genetic algorithm
CN115081293A (en) Multi-objective optimization method of bubble column carbon capture process based on proxy model
CN107992645A (en) Sewage disposal process soft-measuring modeling method based on chaos-fireworks hybrid algorithm
CN112927769B (en) MOFs material defect structure prediction method based on pore size distribution curve
Kumar et al. Adaptation and application of the in situ adaptive tabulation (ISAT) procedure to reacting flow calculations with complex surface chemistry
Wang et al. Modified algorithms for fast construction of optimal Latin-hypercube design
CN115860173A (en) Construction and prediction method and medium of carbon emission prediction model based on Stacking algorithm
CN112765902A (en) RBF neural network soft measurement modeling method based on TentFWA-GD and application thereof
Li et al. Multi/many-objective evolutionary algorithm assisted by radial basis function models for expensive optimization
CN103984996A (en) Lake-reservoir algal bloom generating mechanism time varying model optimization and prediction method based on taboo searching algorithm and genetic algorithm
CN105389442A (en) Reverse design method for coupling genetic algorithm, neural network and numerical simulation
CN103177291A (en) Variable-search-space ribonucleic acid (RNA) genetic algorithm modeling method for continuous stirred tank reactor
CN105426959A (en) Aluminium electrolysis energy conservation and emission reduction method based on BP neural network and self-adaptive MBFO algorithm
Ghasabehi et al. Multi-objective optimization of two-phase flow in the proton exchange membrane fuel cells based on a data driven surrogate model
CN114743617A (en) Efficient MOFs material searching and generating method
Garg et al. Framework of model selection criteria approximated genetic programming for optimization function for renewable energy systems
Bourisli et al. General correlation of building energy use via hybrid genetic programming/genetic algorithm
CN111859807A (en) Initial pressure optimizing method, device, equipment and storage medium for steam turbine

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination