Low-concentration flue gas CO2 absorbent screening method, system, equipment and medium

Abstract

According to the low-concentration flue gas CO2 absorbent screening method, system, equipment and medium disclosed by the embodiment of the invention, the RF for feature screening and the XGBoost for performance prediction are combined to construct a high-precision physical property prediction model, so that the dependence on experimental data is reduced; by introducing an NSGA-II algorithm of an adaptive constraint processing mechanism, weights of key features are dynamically adjusted, a global optimal solution is ensured, the accuracy of the model is effectively improved, and industrial requirements are met. Finally, the low-concentration flue gas CO2 absorbent formula which is excellent in performance and economical and feasible is screened out, an efficient absorbent design scheme is provided for industrial application, the efficiency and economical efficiency of the CO2 capturing process are improved, the industrialization process of the CO2 capturing technology is promoted, and powerful technical support is provided for achieving the purposes of carbon emission reduction and carbon neutralization.

Description

Technical Field

[0001] The embodiments of the present disclosure belong to the technical field, and specifically relate to a method, system, device and medium for screening a low-concentration flue gas CO2 absorbent. Background Art

[0002] Carbon dioxide (CO2), a major greenhouse gas, has become a key focus of global climate change research. Carbon capture and storage (CCS) technology is considered a key means of mitigating climate change and reducing CO2 emissions. However, for low-concentration CO2 flue gases such as those produced by gas-fired power plants and ship exhaust, CCS faces challenges such as low mass transfer efficiency and high desorption energy consumption.

[0003] In industrial flue gas, the capture of low-concentration CO2 (3-6 vol.%) faces two major difficulties: ① Low-concentration CO2 leads to insufficient driving force for gas-liquid mass transfer, and the mass transfer rate of traditional absorbents is difficult to meet the requirements; ② The heat-limited flue gas environment leads to a high desorption rate requirement for the absorbent at low temperatures (<100°C), and the regeneration energy consumption also needs to be controlled. The ideal absorbent not only needs to have a high CO2 absorption capacity, but also has low viscosity and a reasonable cost. In addition, the CO2 mass transfer rate and desorption rate of the absorbent are key performance indicators that determine the carbon capture system. The absorption and desorption of CO2 must be completed efficiently under low energy consumption conditions. Traditional amine absorbents have the problems of low efficiency and high energy consumption in the process of capturing low-concentration CO2 flue gas, and there is an urgent need to develop new absorbents.

[0004] Based on a large amount of experimental data and feature analysis, the machine learning model can screen out the absorbent formula that is most suitable for low-concentration CO2 flue gas, effectively reducing experimental costs and improving capture efficiency, providing a better solution for the screening and optimization of absorbents.

[0005] However, the existing technology of the following absorbent screening method still has many shortcomings:

[0006] ① Patent CN 116429981A uses a traditional trial-and-error method to screen compound solvents, which relies on experimental experience, has a long screening cycle (usually 6 to 12 months), and is difficult to balance multiple objective constraints (such as cost, viscosity, amine concentration, etc.);

[0007] ② Patent CN 116230115A uses a single machine learning model, which is only applicable to phase change absorbers and relies on quantum chemical calculations. It has high computing resource requirements and cannot quickly respond to the needs of low-concentration CO2 scenarios.

[0008] ③ Patent US20200155555A1 uses a genetic algorithm to optimize the absorbent formula, but its constraint processing is a static penalty function. The penalty function design is rigid, prone to falling into local optimality, and unable to dynamically adapt to complex physical property constraints. Summary of the Invention

[0009] The embodiments of the present disclosure aim to solve at least one of the technical problems existing in the prior art and provide a method, system, device and medium for screening a low-concentration flue gas CO2 absorbent.

[0010] One aspect of the present disclosure provides a method for screening a low-concentration flue gas CO2 absorbent, the method comprising:

[0011] Obtaining multivariate parameters of different absorbents and normalizing the multivariate parameters; wherein the multivariate parameters include material composition, physical properties, dynamics / thermodynamics, and cost parameters;

[0012] The importance score of each feature in the multivariate parameters is calculated using a random forest regression model, and the key features in the multivariate parameters are screened according to the importance score; wherein the key features include at least organic amine concentration, solution viscosity, and cost;

[0013] The key features are input into the pre-trained extreme gradient boosting regression model, which outputs the corresponding predicted mass transfer rate and desorption rate;

[0014] With the goal of maximizing mass transfer rate and desorption rate, a non-dominated sorting genetic algorithm II combined with an adaptive penalty function is used to generate a frontier solution set that meets the key characteristics.

[0015] According to the frontier solution set, the approximate ideal solution sorting method is used to output the optimal low-concentration flue gas CO2 absorbent formula.

[0016] Furthermore, the normalizing of the multivariate parameters includes:

[0017] One-hot encoding of categorical variables to generate binary feature vectors; and / or,

[0018] The continuous parameters were Z-score normalized; the Z-score normalization was performed by the following formula:

[0019]

[0020] Where x is the standard score of the data, a is the original data, σ is the mean of the data set, and σ is the standard deviation of the data set.

[0021] Furthermore, the non-dominated sorting genetic algorithm II adopts simulated binary crossover with a crossover probability of 0.9; and / or,

[0022] The non-dominated sorting genetic algorithm II adopts polynomial mutation with a mutation probability of 0.1.

[0023] Furthermore, the adaptive penalty function is expressed as follows:

[0024]

[0025] Where λ is the adaptive weight factor, t is the number of iterations, and β is the attenuation factor.

[0026] Furthermore, the extreme gradient boosting regression model is pre-trained by the following steps:

[0027] Obtain high concentration CO2 dataset and low concentration CO2 dataset;

[0028] Perform preliminary training of the extreme gradient boosting regression model using the high CO2 concentration dataset;

[0029] The parameters of the first three tree layers are frozen, and the model parameters of the extreme gradient boosting regression model are optimized using the low concentration CO2 dataset.

[0030] Another aspect of the present disclosure provides a low-concentration flue gas CO2 absorbent screening system, the system comprising:

[0031] An acquisition module, configured to acquire multivariate parameters of different absorbents and normalize the multivariate parameters; wherein the multivariate parameters include material composition, physical properties, dynamics / thermodynamics, and cost parameters;

[0032] A feature module is used to calculate the importance score of each feature in the multivariate parameter using a random forest regression model, and screen the key features in the multivariate parameter according to the importance score; wherein the key features include at least organic amine concentration, solution viscosity and cost;

[0033] The prediction module is used to input key features into a pre-trained extreme gradient boosting regression model and output the corresponding predicted mass transfer rate and desorption rate;

[0034] A sorting module is used to generate a frontier solution set that meets key characteristics by using a non-dominated sorting genetic algorithm II combined with an adaptive penalty function with the goal of maximizing the mass transfer rate and desorption rate;

[0035] The output module is used to output the optimal low-concentration flue gas CO2 absorbent formula based on the frontier solution set using the approximate ideal solution sorting method.

[0036] Furthermore, the acquisition module is specifically configured to:

[0037] One-hot encoding of categorical variables to generate binary feature vectors; and / or,

[0038] The continuous parameters were Z-score normalized; the Z-score normalization was performed by the following formula:

[0039]

[0040] Where x is the standard score of the data, a is the original data, σ is the mean of the data set, and σ is the standard deviation of the data set.

[0041] Furthermore, the system also includes a training module for:

[0042] Obtain high concentration CO2 dataset and low concentration CO2 dataset;

[0043] Perform preliminary training of the extreme gradient boosting regression model using the high CO2 concentration dataset;

[0044] The parameters of the first three tree layers are frozen, and the model parameters of the extreme gradient boosting regression model are optimized using the low concentration CO2 dataset.

[0045] Another aspect of the present disclosure provides an electronic device, comprising:

[0046] at least one processor; and,

[0047] The memory communicatively connected to the at least one processor is used to store one or more programs, which, when executed by the at least one processor, enable the at least one processor to implement the low-concentration flue gas CO2 absorbent screening method described above.

[0048] Another aspect of the present disclosure provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the method for screening a low-concentration flue gas CO2 absorbent described above is implemented.

[0049] The disclosed embodiments provide a method, system, device, and medium for screening low-concentration flue gas CO2 absorbents. By combining RF for feature screening and XGBoost for performance prediction, a high-precision physical property prediction model is constructed, reducing dependence on experimental data. By introducing the NSGA-II algorithm with an adaptive constraint processing mechanism, the weights of key features are dynamically adjusted to ensure a global optimal solution, effectively improving the accuracy of the model and meeting industrial needs. Ultimately, a formula for a low-concentration flue gas CO2 absorbent with superior performance and economic feasibility is screened out, providing an efficient absorbent design solution for industrial applications, helping to improve the efficiency and economy of the CO2 capture process and promote the industrialization of CO2 capture technology, providing strong technical support for achieving carbon emission reduction and carbon neutrality goals. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 This is a flow chart of a method for screening a low-concentration flue gas CO2 absorbent according to an embodiment of the present disclosure;

[0051] Figure 2 This is a structural schematic diagram of a low-concentration flue gas CO2 absorbent screening system according to another embodiment of the present disclosure;

[0052] Figure 3 This is a schematic structural diagram of an electronic device according to another embodiment of the present disclosure. DETAILED DESCRIPTION

[0053] The following will be combined with the accompanying drawings in the embodiments of the present disclosure to clearly and completely describe the technical solutions in the embodiments of the present disclosure. Obviously, the embodiments described are only part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments of the present disclosure, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of the present disclosure.

[0054] In addition, the described features, structures or characteristics may be combined in any suitable manner in one or more embodiments. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure can be practiced without one or more of the specific details, or other methods, components, devices, steps, etc. can be adopted. In other cases, well-known methods, devices, implementations or operations are not shown or described in detail to avoid blurring various aspects of the present disclosure.

[0055] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may be decomposed, while others may be combined or partially combined. Therefore, the actual execution order may vary depending on the actual situation.

[0056] It should be understood that although the terms first, second, third, etc. may be used in this disclosure to describe various components, these components should not be limited by these terms. These terms are used to distinguish one component from another. Thus, the first component discussed below can be referred to as the second component without departing from the teachings of the concepts of this disclosure. As used in this disclosure, the term "and / or" includes any one of the associated listed items and all combinations of one or more of them.

[0057] Those skilled in the art will understand that the drawings are merely schematic diagrams of example embodiments, and the modules or processes in the drawings are not necessarily necessary for implementing the present disclosure, and therefore cannot be used to limit the scope of protection of the present disclosure.

[0058] like Figure 1As shown, one aspect of the present disclosure provides a method for screening a low-concentration flue gas CO2 absorbent, the method comprising:

[0059] Step S1: Obtain multivariate parameters of different absorbents and normalize the multivariate parameters.

[0060] Specifically, the material composition, physical properties, dynamics / thermodynamics, cost parameters and other multivariate parameters of different organic amines (primary amines, secondary amines, tertiary amines, etc.), solvents (water, PEG, ionic liquids, etc.) and their composite systems at different temperatures (40°C to 100°C) and CO2 partial pressures (100kPa to 102kPa) can be obtained through a high-throughput automated experimental platform or literature research. The multivariate parameters include the water content of the absorbent Amine molar concentration c amine (mol / kg), solution density ρ(kg / m 3 ), constant pressure specific heat c p (J / (g·K)), solution dynamic viscosity μ (Pa·s), solution conductivity τ (S / m), absorbent cost C (yuan / ton), mass transfer rate k mass,transfer (mol / (m 2 ·s)), desorption rate k desorption (mol / (m 3 ·s)) etc. Among them, the mass transfer rate k mass,transfer and desorption rate k desorption This is the target performance of this embodiment.

[0061] Normalize the obtained multivariate parameters:

[0062] For continuous parameters (such as water content Amine molar concentration c amine etc.) to perform Z-score standardization:

[0063]

[0064] In the formula, x is the standard score of the data, a is the original data, is the mean of the data set, and σ is the standard deviation of the data set;

[0065] One-hot encoding was used for categorical variables (organic amine, solvent type, etc.) to generate binary feature vectors.

[0066] Step S2: Calculate the importance score of each feature in the multivariate parameter using the random forest regression model, and select the key features in the multivariate parameter based on the importance score.

[0067] Specifically, the initial feature set obtained in step S1 is input into the Random Forest (RF) regression model to calculate the importance score of each feature, and retain the key features with an importance score greater than 0.8, such as the molar concentration of organic amine c amine (importance score 0.92), solution dynamic viscosity μ (importance score 0.88), etc., and redundant features with importance score ≤ 0.8, such as constant pressure specific heat c p (importance score 0.75), etc., to form a key feature set. The key features of this embodiment retain organic amine concentration, solution viscosity and cost.

[0068] Step S3: input the key features into the pre-trained extreme gradient boosting regression model, and output the corresponding predicted mass transfer rate and desorption rate respectively.

[0069] Specifically, the Extreme Gradient Boosting (XGBoost) algorithm was used to build a multi-output regression model to predict the key performance indicator of the absorbent, namely the mass transfer rate k mass,transfer and desorption rate k desorption Assume that there are m key features with importance scores greater than 0.8, and denote the key feature set as X and the target performance set as Y:

[0070] X=[x1,x2,…,x m ]

[0071] Y=[k mass,transfer ,k desorption ]

[0072] For mass transfer rate k mass,transfer , a regression model of mass transfer rate is established through the relevant parameters in the key feature set X:

[0073] k mass,transfer =f1(x1,x2,…,x m )=ω1·x1+ω2·x2+···+ω m ·x m +b1

[0074] Among them, ω1, ω2,···,ω m is the mass transfer rate model parameter, and b1 is the mass transfer rate bias term.

[0075] For the desorption rate k desorption , a regression model of mass transfer rate is established through the relevant parameters in the feature set X:

[0076] k desorption =f2(x1,x2,…,x m )=α1·x1+α2·x2+···+αm ·x m +c1

[0077] Among them, α1, α2,···,α m is the desorption rate model parameter, and c1 is the desorption rate bias term.

[0078] Considering the scarcity of low-concentration CO2 experimental data, this example establishes a cross-concentration transfer learning method, which mainly includes pre-training and parameter fine-tuning:

[0079] 1. Obtain a high-concentration CO2 dataset (N1=500 groups) and a low-concentration CO2 dataset (N2=50 groups).

[0080] 2. Use high-concentration CO2 (10vol.%~15vol.%) experimental data to preliminarily train the XGBoost model to capture the general laws of absorbent properties.

[0081] 3. By freezing the first three layers of the model tree structure and fine-tuning the last two layers, the model parameters are optimized by combining a small amount of low-concentration CO2 experimental data to adapt it to the special low-concentration CO2 scenario.

[0082] The mass transfer rate k used in this example is thus obtained mass,transfer and desorption rate k desorption Extreme gradient boosting regression model.

[0083] The key feature set obtained in the previous step S2 is input into the extreme gradient boosting regression model, and the predicted mass transfer rate k is output respectively. mass,transfer and desorption rate k desorption .

[0084] Step S4: with the goal of maximizing the mass transfer rate and the desorption rate, a non-dominated sorting genetic algorithm II combined with an adaptive penalty function is used to generate a frontier solution set that meets the key characteristics.

[0085] Specifically, a multi-objective optimization is designed:

[0086] k mass,transfer ≥k mass,transfer,set

[0087] k desorption ≥k desorption,set

[0088] Among them, k mass,transfer,set and k desorption,set are the set mass transfer rate target and desorption rate target respectively.

[0089] With the goal of maximizing mass transfer rate and desorption rate, combined with hard constraints (m key features with importance scores > 0.8), the Pareto Front was generated based on the Non-dominated Sorting Genetic Algorithm II (NSGA-II) algorithm to obtain the optimal solution set that meets the requirements of mass transfer rate and desorption rate.

[0090] The optimization framework of NSGA-II in this embodiment includes: 1. Population initialization: Randomly generate 100 candidate particles (absorbent formulations), each of which includes m key features. 2. Crossover and mutation: Crossover probability is 0.9, using simulated binary crossover (SBX); mutation probability is 0.1, using polynomial mutation. 3. Adaptive penalty function design:

[0091] To handle constraint violations, an adaptive penalty function is designed. If a recipe violates a hard constraint, the penalty function takes into account the degree of constraint violation and adds a corresponding penalty to the fitness function. Specifically, when a recipe violates a constraint, the penalty function adjusts the recipe's fitness weight based on the degree of violation, reducing the recipe's fitness and forcing the optimization algorithm to find a new solution. The adaptive penalty function is expressed as follows:

[0092]

[0093] Where λ is the adaptive weight factor, t is the number of iterations, and β is the attenuation factor.

[0094] Step S5: Based on the frontier solution set, the optimal low-concentration flue gas CO2 absorbent formula is output using the approximate ideal solution sorting method.

[0095] Specifically, the Pareto frontier solution set generated by the NSGA-II algorithm in step S4 contains all solutions that strike a balance between multiple objectives. Based on this frontier solution set, the optimal solution is selected using the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS). The TOPSIS method calculates the distance between each solution and the ideal solution and the worst-case solution to select the absorbent formulation with the best overall performance, ultimately outputting the optimal low-concentration flue gas CO2 absorbent formulation.

[0096] The method of steps S1 to S5 above was applied to low-concentration CO2 flue gas from a gas-fired power plant after desulfurization, where the CO2 concentration was 4.2 vol.% and the flue gas temperature was 50°C. The solution viscosity was measured using a rotational viscometer as specified in GB / T 22235-2008 Determination of Viscosity of Liquids. The CO2 mass transfer rate was measured using a wetted-wall column (WWC) CO2 absorption reaction device, where the WWC was a stainless steel hollow tube with an outer diameter of 12 mm, a height of 100 mm, and a gas-liquid reaction contact area of ​​37.70 cm. 2 The flue gas after desulfurization of the simulated gas power plant was controlled at 2.0L / min by a mass flow meter, the temperature was maintained at 50 degrees, the absorbent temperature was also maintained at 50 degrees, the flow rate was controlled at 50mL / min, and the gas-liquid countercurrent contact was performed. The CO2 concentration at the inlet and outlet of the device was measured using a non-dispersive infrared (NDIR) multi-gas analyzer, and the CO2 mass transfer rate k was calculated based on the double membrane theory. mass,transfer The final optimization results are shown in Table 1 below.

[0097] Table 1

[0098]

[0099] Compared with a single machine learning model (such as SVM), the prediction accuracy is improved by 20%.

[0100] Taking the low-concentration CO2 exhaust gas of a ship engine as an example, the CO2 concentration is 5.5vol.%, and the flue gas temperature is 80℃. In addition, due to space limitations, the absorbent organic amine concentration is required to be ≥30wt.%, and the solution viscosity is required to be ≤3.5mPa·s. A high-pressure reactor is used as the experimental instrument, and a certain volume of absorbent is placed in the reactor. The initial temperature is 60℃, and the temperature controller is turned on to set the target temperature to 100℃. After heating begins, the absorbent rich liquid is desorbed by heat. The temperature of the regenerated gas drops after passing through the condenser, and the water vapor condenses into the conical flask. Then the CO2 gas passes through the CO2 thermal flowmeter to record the CO2 desorption amount, and the maximum desorption rate k is calculated through the desorption curve. desorption The final optimization results are shown in Table 2 below.

[0101] Table 2

[0102]

[0103] Compared with existing ship-specific alcohol amine absorbents, the mass transfer rate is increased by 18% and the desorption rate is increased by 15%.

[0104] A method for screening low-concentration flue gas CO2 absorbents in an embodiment of the present disclosure, by combining RF for feature screening and XGBoost for performance prediction, constructs a high-precision physical property prediction model, reducing dependence on experimental data; by introducing the NSGA-II algorithm with an adaptive constraint processing mechanism, the weights of key features are dynamically adjusted to ensure the global optimal solution, effectively improve the accuracy of the model, and meet industrial needs; through transfer learning technology, the model is pre-trained using high-concentration CO2 experimental data, and then the model parameters are fine-tuned in combination with a small amount of low-concentration CO2 experimental data, which significantly improves the model prediction accuracy and adaptability, and adapts to low-concentration CO2 scenarios. Ultimately, a low-concentration flue gas CO2 absorbent formula with superior performance and economic feasibility is screened out, providing an efficient absorbent design solution for industrial applications, helping to improve the efficiency and economy of the CO2 capture process and promote the industrialization of CO2 capture technology, and providing strong technical support for achieving carbon emission reduction and carbon neutrality goals.

[0105] like Figure 2 As shown, another embodiment of the present disclosure provides a low-concentration flue gas CO2 absorbent screening system, the system comprising:

[0106] An acquisition module 210 is used to obtain multivariate parameters of different absorbents and normalize the multivariate parameters; wherein the multivariate parameters include material composition, physical properties, dynamics / thermodynamics, and cost parameters;

[0107] A feature module 220 is configured to calculate the importance score of each feature in the multivariate parameter using a random forest regression model, and to select key features in the multivariate parameter based on the importance score; wherein the key features include at least organic amine concentration, solution viscosity, and cost;

[0108] Prediction module 230, for inputting key features into a pre-trained extreme gradient boosting regression model and outputting corresponding predicted mass transfer rate and desorption rate;

[0109] A sorting module 240 is configured to generate a frontier solution set that meets key characteristics by using a non-dominated sorting genetic algorithm II combined with an adaptive penalty function with the goal of maximizing the mass transfer rate and the desorption rate;

[0110] The output module 250 is used to output the optimal low-concentration flue gas CO2 absorbent formula based on the frontier solution set using the approximate ideal solution sorting method.

[0111] Exemplarily, the acquisition module 210 is specifically configured to:

[0112] One-hot encoding of categorical variables to generate binary feature vectors; and / or,

[0113] The continuous parameters were Z-score normalized; the Z-score normalization was performed by the following formula:

[0114]

[0115] Where x is the standard score of the data, a is the original data, σ is the mean of the data set, and σ is the standard deviation of the data set.

[0116] Exemplarily, the system further includes a training module 260 for:

[0117] Obtain high concentration CO2 dataset and low concentration CO2 dataset;

[0118] Perform preliminary training of the extreme gradient boosting regression model using the high CO2 concentration dataset;

[0119] The parameters of the first three tree layers are frozen, and the model parameters of the extreme gradient boosting regression model are optimized using the low concentration CO2 dataset.

[0120] Specifically, a low-concentration flue gas CO2 absorbent screening system in an embodiment of the present disclosure is used to implement the low-concentration flue gas CO2 absorbent screening method described in the above embodiments. The specific implementation process has been described in detail in the above embodiments and will not be repeated here.

[0121] A low-concentration flue gas CO2 absorbent screening system of the disclosed embodiment, by combining RF for feature screening and XGBoost for performance prediction, constructs a high-precision physical property prediction model, reducing dependence on experimental data; by introducing the NSGA-II algorithm with an adaptive constraint processing mechanism, the weights of key features are dynamically adjusted to ensure the global optimal solution, effectively improve the accuracy of the model, and meet industrial needs; through transfer learning technology, the model is pre-trained using high-concentration CO2 experimental data, and then the model parameters are fine-tuned in combination with a small amount of low-concentration CO2 experimental data, which significantly improves the model prediction accuracy and adaptability, and adapts to low-concentration CO2 scenarios. Ultimately, a low-concentration flue gas CO2 absorbent formula with superior performance and economic feasibility is screened out, providing an efficient absorbent design solution for industrial applications, helping to improve the efficiency and economy of the CO2 capture process and promote the industrialization of CO2 capture technology, providing strong technical support for achieving carbon emission reduction and carbon neutrality goals.

[0122] like Figure 3 As shown, another embodiment of the present disclosure provides an electronic device, including:

[0123] At least one processor 301; and a memory 302 in communication with the at least one processor 301, for storing one or more programs, which, when executed by the at least one processor 301, enable the at least one processor 301 to implement the low-concentration flue gas CO2 absorbent screening method described above.

[0124] The memory 302 and processor 301 are connected using a bus. The bus can include any number of interconnected buses and bridges, connecting various circuits of one or more processors 301 and memory 302. The bus can also connect various other circuits such as peripheral devices, voltage regulators, and power management circuits. These are all well known in the art and are therefore not described further herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be a single component or multiple components, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by the processor 301 is transmitted over a wireless medium via an antenna. Furthermore, the antenna receives data and transmits it to the processor 301.

[0125] The processor 301 is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. The memory 302 can be used to store data used by the processor 301 when performing operations.

[0126] Yet another embodiment of the present disclosure provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the method for screening a low-concentration flue gas CO2 absorbent described above is implemented.

[0127] The computer-readable storage medium may be included in the system or electronic device of the present disclosure, or may exist independently.

[0128] Computer-readable storage media may be any tangible medium that contains or stores a program, which may be an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device. More specific examples include, but are not limited to, an electrical connection with one or more wires, a portable computer disk, a hard disk, optical fiber, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0129] The computer-readable storage medium may also include a data signal propagated in baseband or as part of a carrier wave, which carries the computer-readable program code. Specific examples include but are not limited to electromagnetic signals, optical signals, or any suitable combination thereof.

[0130] It is understood that the above embodiments are merely exemplary embodiments for illustrating the principles of the present disclosure, and the present disclosure is not limited thereto. Those skilled in the art may make various modifications and improvements without departing from the spirit and substance of the present disclosure, and such modifications and improvements are also considered to be within the scope of protection of the present disclosure.

Claims

1. A method for screening low-concentration flue gas CO2 absorbent, characterized in that: The method comprises: Obtaining multivariate parameters of different absorbents and normalizing the multivariate parameters; wherein the multivariate parameters include material composition, physical properties, dynamics / thermodynamics, and cost parameters; The importance score of each feature in the multivariate parameters is calculated using a random forest regression model, and the key features in the multivariate parameters are screened according to the importance score; wherein the key features include at least organic amine concentration, solution viscosity, and cost; The key features are input into the pre-trained extreme gradient boosting regression model, which outputs the corresponding predicted mass transfer rate and desorption rate; With the goal of maximizing mass transfer rate and desorption rate, a non-dominated sorting genetic algorithm II combined with an adaptive penalty function is used to generate a frontier solution set that meets the key characteristics. According to the frontier solution set, the approximate ideal solution sorting method is used to output the optimal low-concentration flue gas CO2 absorbent formula.

2. The method according to claim 1, characterized in that The normalizing of the multivariate parameters comprises: One-hot encoding of categorical variables to generate binary feature vectors; and / or, The continuous parameters were Z-score normalized; the Z-score normalization was performed by the following formula: Where x is the standard score of the data, a is the original data, σ is the mean of the data set, and σ is the standard deviation of the data set.

3. The method according to claim 1, characterized in that The non-dominated sorting genetic algorithm II uses simulated binary crossover with a crossover probability of 0.9; and / or, The non-dominated sorting genetic algorithm II adopts polynomial mutation with a mutation probability of 0.

1.

4. The method according to claim 1, wherein The adaptive penalty function is expressed as follows: Where λ is the adaptive weight factor, t is the number of iterations, and β is the attenuation factor.

5. The method according to any one of claims 1 to 4, characterized in that The extreme gradient boosting regression model is pre-trained by the following steps: Obtain high concentration CO2 dataset and low concentration CO2 dataset; Perform preliminary training of the extreme gradient boosting regression model using the high CO2 concentration dataset; The parameters of the first three tree layers are frozen, and the model parameters of the extreme gradient boosting regression model are optimized using the low concentration CO2 dataset.

6. A low-concentration flue gas CO2 absorbent screening system, characterized in that: The system comprises: An acquisition module, configured to acquire multivariate parameters of different absorbents and normalize the multivariate parameters; wherein the multivariate parameters include material composition, physical properties, dynamics / thermodynamics, and cost parameters; A feature module is used to calculate the importance score of each feature in the multivariate parameter using a random forest regression model, and screen the key features in the multivariate parameter according to the importance score; wherein the key features include at least organic amine concentration, solution viscosity and cost; The prediction module is used to input key features into a pre-trained extreme gradient boosting regression model and output the corresponding predicted mass transfer rate and desorption rate; A sorting module is used to generate a frontier solution set that meets key characteristics by using a non-dominated sorting genetic algorithm II combined with an adaptive penalty function with the goal of maximizing the mass transfer rate and desorption rate; The output module is used to output the optimal low-concentration flue gas CO2 absorbent formula based on the frontier solution set using the approximate ideal solution sorting method.

7. The system according to claim 6, characterized in that The acquisition module is specifically used for: One-hot encoding of categorical variables to generate binary feature vectors; and / or, The continuous parameters were Z-score normalized; the Z-score normalization was performed by the following formula: Where x is the standard score of the data, a is the original data, σ is the mean of the data set, and σ is the standard deviation of the data set.

8. The system according to claim 6 or 7, characterized in that The system further comprises a training module for: Obtain high concentration CO2 dataset and low concentration CO2 dataset; Perform preliminary training of the extreme gradient boosting regression model using the high CO2 concentration dataset; The parameters of the first three tree layers are frozen, and the model parameters of the extreme gradient boosting regression model are optimized using the low concentration CO2 dataset.

9. An electronic device, characterized in that: include: at least one processor; as well as, A memory communicatively connected to the at least one processor, for storing one or more programs, which, when executed by the at least one processor, enables the at least one processor to implement the low-concentration flue gas CO2 absorbent screening method according to any one of claims 1 to 5.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for screening low-concentration flue gas CO2 absorbent according to any one of claims 1 to 5 is implemented.

Images