Robust optimization method and system for p-xylene adsorption separation device

Through the robust optimization method, mathematical model and multi-objective optimization algorithm are used to optimize the operating variables of the xylene adsorption separation device, which solves the problem of poor robustness of the operating variables in simulated mobile bed technology, improves the separation efficiency and system stability, and reduces the Energy consumption enhances the competitiveness of industrial production.

CN119028467AActive Publication Date: 2024-11-26EAST CHINA UNIV OF SCI & TECH
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202411208045.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-30
Publication Date
2024-11-26
Estimated Expiration
2044-08-30

AI Technical Summary

Technical Problem

The simulated mobile bed technology has the problem of poor operating variable robustness during xylene adsorption separation, which leads to a sharp decline in adsorption separation effect under non-design conditions, and the challenges brought by feed volatility and system complexity are difficult to solve.

Method used

Using a robust optimization method, the operation variables are optimized to improve the robustness and efficiency of the xylene adsorption separation device by establishing mathematical models, uncertainty quantization, global sensitivity analysis and multi-objective optimization algorithm, including sparse grid polynomial chaotic expansion, Sobol index and NSGA-II algorithm.

Benefits of technology

It realizes the system efficiency and stability under different operating conditions, improves the separation efficiency of xylene, reduces energy consumption, and enhances the economic benefits and competitiveness of industrial production.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119028467A_ABST
    Figure CN119028467A_ABST
Patent Text Reader

Abstract

The invention discloses a robust optimization method and system for a p-xylene adsorption separation device, and solves the problem of poor operating variable robustness of a deterministic optimization method in a PX adsorption separation process. According to the technical scheme, a paraxylene adsorption separation process mathematical model is established by referring to paraxylene adsorption separation device operation variables and inlet and outlet material composition; performing uncertainty quantization on input variables of the mathematical model through a polynomial chaos expansion method based on sparse grids by utilizing the mathematical model; carrying out global sensitivity analysis by utilizing a Sobol index, evaluating the influence of the feeding flow and the component concentration of the p-xylene adsorption separation device on the p-xylene adsorption separation device, and screening key mathematical model input variables; the mathematical model is called through an NSGA-II multi-objective optimization algorithm, the mean value and the standard deviation obtained after uncertainty quantification of the performance indexes of the p-xylene adsorption separation device serve as optimization objectives, the operation variables are optimized, and the optimal operation variables are obtained.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of xylene separation, and in particular to an optimization method and system for separating p-xylene. Background Art

[0002] C8 aromatics mainly include paraxylene (PX), metaxylene (MX), orthoxylene (OX) and ethylbenzene (EB). PX is the main product of aromatic production and an important raw material for the production of purified terephthalic acid, which plays a key role in the production of polyester fibers, resins and films. In addition, PX can also be used as a solvent and raw material in the production of chemical products such as medicines and fragrances. Since these compounds have similar boiling points, it is not practical to separate them by distillation. Early industrial technology used deep cold crystallization to separate PX and other C8 isomers by using their melting point differences, but this method has the problems of low PX recovery rate and high energy consumption. In the early 1960s, simulated moving bed adsorption separation technology appeared and gradually replaced the crystallization process. Simulated moving bed technology uses the different affinities of C8 aromatics to adsorbents to improve separation efficiency and significantly reduce energy consumption. Therefore, simulated moving bed technology has become an important research direction and a widely used mainstream technology in the industry.

[0003] With the increasing emphasis on product quality and cost control in industrial production, various studies on reducing costs and improving production efficiency have been widely explored in recent years. In the process of simulated moving bed optimization, the key issue is to determine the optimal value of the operating variables, that is, to optimize under the assumption that the chromatographic column configuration, filler parameters and adsorption equilibrium characteristics of each component have been determined. However, there are inevitably uncertainties in various factors in engineering problems. These uncertainties will cause the optimal performance obtained by deterministic optimization to drop sharply under non-design conditions. Therefore, robust optimization research on the PX adsorption separation process in a simulated moving bed has the potential for practical application.

[0004] In summary, although simulated moving bed technology has achieved remarkable results in the adsorption separation of C8 aromatics, in actual industrial applications, uncertainties such as the complexity of system operation and feed volatility still pose challenges to the adsorption separation effect and production efficiency. Therefore, in view of these uncertainties, it is particularly important to study and develop a robust optimization method that can adapt to feed uncertainty. Summary of the invention

[0005] A brief summary of one or more aspects is given below to provide a basic understanding of these aspects. This summary is not an exhaustive overview of all conceived aspects, and is neither intended to identify the key or critical elements of all aspects nor to define the scope of any or all aspects. Its only purpose is to give some concepts of one or more aspects in a simplified form as a prelude to a more detailed description that will be given later.

[0006] The purpose of the present invention is to solve the above-mentioned problems and to provide a robust optimization method and system for a xylene adsorption separation device to solve the problem of poor robustness of operating variables obtained by the deterministic optimization method in the PX adsorption separation process. By maintaining the high efficiency and stability of the system under different operating conditions, the separation efficiency of PX is further improved, energy consumption is reduced, and the economic benefits and competitiveness of industrial production are enhanced.

[0007] The technical solution of the present invention is as follows: The present invention discloses a robust optimization method for a para-xylene adsorption separation device, the method comprising:

[0008] Step S1, establishing a mathematical model of the p-xylene adsorption separation process with reference to the operating variables of the p-xylene adsorption separation device and the composition of the inlet and outlet materials of the p-xylene adsorption separation device;

[0009] Step S2, using the mathematical model established in step S1, quantifying the uncertainty of the input variables of the mathematical model through a polynomial chaos expansion method based on a sparse grid, that is, propagating the uncertainty of the input variables;

[0010] Step S3, using the Sobol index to perform global sensitivity analysis, evaluate the impact of the feed flow rate and component concentration of the para-xylene adsorption separation device on the para-xylene adsorption separation device, and screen key mathematical model input variables to improve robust optimization efficiency;

[0011] Step S4, calling the mathematical model established in step S1 through the NSGA-II multi-objective optimization algorithm, taking the mean and standard deviation obtained after quantifying the uncertainty of the performance indicators of the xylene adsorption separation device as the optimization target, optimizing the operating variables, and obtaining the optimal operating variables, thereby realizing robust optimization of the xylene adsorption separation device.

[0012] According to an embodiment of the robust optimization method of a para-xylene adsorption separation device of the present invention, the main operating variables of the para-xylene adsorption separation device include: temperature, pressure, feed flow, desorbent flow, extract flow, residual liquid flow, circulation flow, switching time, xylene concentration in the feed, para-xylene concentration in the feed, o-xylene concentration in the feed, ethylbenzene concentration in the feed, and adsorbent specific capacity; the composition of the inlet and outlet materials of the para-xylene adsorption separation device includes: para-xylene PX, ethylbenzene EB, meta-xylene MX, o-xylene OX, para-diethylbenzene PDEB, carbon nine aromatics C9A, carbon ten aromatics C10A, other non-aromatic substances, etc.

[0013] According to an embodiment of the robust optimization method of the para-xylene adsorption separation device of the present invention, step S1 further comprises:

[0014] Step S11, based on the structural parameters of the simulated moving bed and the operating parameters of the para-xylene adsorption separation process, construct a mathematical model of the simulated moving bed, including but not limited to a mechanism model, a polynomial empirical model and a data-driven model of the para-xylene simulated moving bed, the mathematical model of the simulated moving bed is used to describe key data of the adsorption separation process including the component concentration of each bed layer of the simulated moving bed, the composition of the input and output materials of the simulated moving bed, the para-xylene purity of the simulated moving bed, and the para-xylene yield of the simulated moving bed;

[0015] Step S12, comparing the mathematical model simulation experiment with the axial concentration distribution data collected under actual working conditions, verifying the mathematical model parameters, and ensuring that the calculation error of the mathematical model for xylene purity and yield is within the industrial allowable range.

[0016] According to an embodiment of the robust optimization method of the para-xylene adsorption separation device of the present invention, the mathematical model in step S11 further includes a polynomial empirical model:

[0017]

[0018] Among them, purity PX and yield PX represent the purity and yield of p-xylene respectively, R1, R2, R3 and R4 represent the reflux ratios of the four regions of the xylene adsorption separation device respectively, a i , b j , c k and d l R1, R2, R3 and R4 are respectively for purity PX The coefficients of i, j, k and l are the purity of R1, R2, R3 and R4 respectively. PX The different powers in n, m, p and q are the purity of R1, R2, R3 and R4 respectively. PX The highest power in It is purity PX The constant term, e r , f s , g t and h x R1, R2, R3 and R4 are respectively PX The coefficients, r, s, t and x are the yields of R1, R2, R3 and R4 respectively. PX In the different powers, u, v, w and y are the yields of R1, R2, R3 and R4 respectively. PX The highest power in is yield PX The constant term of the polynomial is calculated by least squares, gradient descent, polynomial kernel method, genetic algorithm, Newton's method and other methods to help optimize the model parameters in the polynomial empirical model to obtain the best fit and prediction ability.

[0019] According to an embodiment of the robust optimization method of the para-xylene adsorption separation device of the present invention, the mathematical model in step S11 includes a neural network data-driven model. Based on the operation data of the para-xylene adsorption separation device, a regression model is established using a feedforward neural network model, a feedback neural network model and a graph neural network, thereby realizing a data-driven model:

[0020] Feedforward neural network model, the network structure includes an input layer, two hidden layers and an output layer. The input layer is the operating variable of the p-xylene adsorption separation device, and the output layer is the purity and yield of p-xylene or the concentration of the components in the extract and the residual liquid;

[0021] Feedback neural network model, the network structure includes an input layer, two recurrent neural network layers and an output layer, wherein the input layer contains operating variables and historical component concentrations, the two recurrent neural network layers are long short-term memory neural networks or gated recurrent units, and the output layer is a fully connected layer, which is used to predict the purity and yield of paraxylene or the concentration of components in the extract and the raffinate;

[0022] Graph neural network is a deep learning model that learns the feature representation of nodes through graph convolution operations and captures the topological structure and relationship between nodes in the graph. The graph data construction module converts the operation variables into the adjacency matrix and feature matrix required by the graph convolution layer module, and predicts the purity and yield of paraxylene or the concentration of components in the extract and raffinate through the fully connected layer.

[0023] According to an embodiment of the robust optimization method of the para-xylene adsorption separation device of the present invention, step S2 further includes the following steps:

[0024] Step S21, Latin hypercube sampling is performed on the operating variables of the simulated moving bed, each variable is uniformly sampled within a larger value range, and a sample set is formed after combination;

[0025] Step S22, performing simulation calculation on each sample, recording the axial concentration distribution data of p-xylene, ethylbenzene, m-xylene, o-xylene and p-xylene under each set of operating variables, and establishing a sample set;

[0026] Step S23, using non-embedded polynomial chaos expansion to quantify uncertainty of operating variables with uncertainty, and constructing a polynomial chaos expansion agent model to realize the propagation of uncertainty;

[0027] Step S24, using the Galerkin projection method to solve the chaotic polynomial coefficients;

[0028] Step S25: Use sparse grid integration instead of full tensor integration to solve the chaotic polynomial coefficients.

[0029] According to an embodiment of the robust optimization method of the para-xylene adsorption separation device of the present invention, step S3 further includes the following steps:

[0030] Step S31, randomly dividing the sample set established in step 2 into ten subsets of equal size, and using ten-fold cross validation to evaluate the prediction performance of the mathematical model;

[0031] Step S32, quantifying the uncertainty of random model input variables based on the mathematical model to study and characterize the uncertainty of the model input;

[0032] Step S33, performing a global sensitivity analysis using the Sobol index to evaluate the effects of feed flow rate and its component concentrations on PX purity and yield;

[0033] Step S34: determine key random model input variables to improve robust optimization efficiency.

[0034] According to an embodiment of the robust optimization method of the para-xylene adsorption separation device of the present invention, step S4 further includes the following steps:

[0035] Step S41, encoding and initializing the population required for generating the NSGA-Ⅱ algorithm by Latin hypercube sampling to ensure that the population has diversity;

[0036] Step S42, calling the mathematical model obtained in step S3 to predict the axial concentration distribution of each individual, calculate the corresponding PX purity and yield, and provide basic data for subsequent uncertainty quantification;

[0037] Step S43, applying the SGPCE method to quantify uncertainty and determine the mean and standard deviation of the PX purity yield;

[0038] Step S44: Calculate the fitness of each individual according to the constraints and objective function set in the NSGA-II algorithm, and determine whether it meets the design requirements;

[0039] Step S45: If the design requirements are not met, the crossover, mutation and selection process is performed and the process returns to step S2 until the set maximum number of iterations is reached and the iteration ends.

[0040] The present invention also discloses a robust optimization system for a para-xylene adsorption separation device, the system comprising:

[0041] The model building module refers to the operating variables of the para-xylene adsorption separation unit and the composition of the inlet and outlet materials of the para-xylene adsorption separation unit to establish a mathematical model of the para-xylene adsorption separation process;

[0042] The uncertainty quantification module uses the mathematical model established by the model building module to quantify the uncertainty of the input variables of the mathematical model through the polynomial chaos expansion method based on sparse grids, that is, to propagate the uncertainty of the input variables;

[0043] The variable screening module uses the Sobol index to perform global sensitivity analysis, evaluate the impact of the feed flow rate and component concentration of the para-xylene adsorption separation unit on the para-xylene adsorption separation unit, and screen key mathematical model input variables to improve the efficiency of robust optimization;

[0044] The variable optimization module calls the mathematical model established by the model construction module through the NSGA-II multi-objective optimization algorithm, takes the mean and standard deviation obtained after quantifying the uncertainty of the performance indicators of the xylene adsorption separation unit as the optimization target, optimizes the operating variables, and obtains the optimal operating variables, thereby realizing the robust optimization of the xylene adsorption separation unit.

[0045] The present invention also discloses a computer-readable storage medium storing program instructions executable by a processor to implement the steps of the robust optimization method for the para-xylene adsorption separation device as described above.

[0046] The present invention also discloses a computer device, which includes a memory, a processor and program instructions stored in the memory for the processor to run, wherein the processor executes the program instructions to implement the steps of the robust optimization method of the para-xylene adsorption separation device as described above.

[0047] Compared with the prior art, the present invention has the following beneficial effects: the present invention establishes a sample set of axial concentration distributions of five components under different operating variables through mathematical model simulation, and uses mathematical models, uncertainty quantification methods and multi-objective optimization algorithms to achieve robust optimization of operating variables, thereby solving the complex mechanism and parameter change problems of the simulated moving bed adsorption separation device and improving operational stability and model calculation efficiency.

[0048] In detail, the present invention improves the efficiency and robustness of the system and can be applied to the xylene adsorption separation process in the petrochemical industry.

[0049] The present invention introduces uncertainty quantification of input variables, which is conducive to evaluating the robustness of the system to parameter changes, improving the reliability and predictive ability of the model, and thus better guiding process optimization and decision making. Using the Sobol index for global sensitivity analysis is conducive to identifying and quantifying the degree of influence of input variables on output results, helping to determine key influencing factors, and guiding further parameter adjustment and optimization, thereby improving the efficiency, reliability and robustness of the system.

[0050] Therefore, through the NSGA-II multi-objective optimization algorithm, the mean and standard deviation obtained after quantifying the uncertainty of the performance indicators of the xylene adsorption separation unit are used as the optimization objectives to optimize the operating variables, which is conducive to achieving robust optimization of the xylene adsorption separation unit and further improving the efficiency and robustness of the system, thereby optimizing the xylene adsorption separation process in the petrochemical industry and achieving more efficient production and resource utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The above features and advantages of the present invention can be better understood after reading the detailed description of the embodiments of the present disclosure in conjunction with the following drawings. In the drawings, the components are not necessarily drawn to scale, and components with similar related properties or features may have the same or similar reference numerals.

[0052] Figure 1 A flow chart showing an embodiment of a robust optimization method for a para-xylene adsorption separation device of the present invention;

[0053] Figure 2 Shows Figure 1 A diagram of a neural network structure of the illustrated embodiment;

[0054] Figure 3 Shows Figure 1 NSGA-II algorithm optimization logic diagram of the embodiment shown;

[0055] Figure 4 Shows Figure 1 The comparison chart of PX axial concentration distribution optimization results is shown.

[0056] Figure 5 A schematic diagram showing an embodiment of a robust optimization system for a para-xylene adsorption separation device of the present invention. DETAILED DESCRIPTION

[0057] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. Note that the aspects described below in conjunction with the accompanying drawings and specific embodiments are only exemplary and should not be construed as limiting the scope of protection of the present invention in any way.

[0058] Figure 1 The process of one embodiment of the robust optimization method of the para-xylene adsorption separation device of the present invention is shown. Figure 1 The implementation process of the method of this embodiment is described in detail as follows.

[0059] Step S1, referring to the operating variables of the p-xylene adsorption separation device and the composition of the inlet and outlet materials of the p-xylene adsorption separation device, a mathematical model of the p-xylene adsorption separation process is established.

[0060] Among them, the mathematical model of the p-xylene adsorption separation process includes but is not limited to a mechanism model, a polynomial model and a data-driven model.

[0061] Among them, the main operating variables of the xylene adsorption separation unit are as follows: temperature, pressure, feed flow, desorbent flow, extract flow, residual liquid flow, circulation flow, switching time, MX concentration in feed, PX concentration in feed, OX concentration in feed, EB concentration in feed, adsorbent specific capacity, etc.

[0062] Among them, the materials entering and leaving the paraxylene adsorption separation device mainly include: paraxylene, ethylbenzene, metaxylene, o-xylene, para-diethylbenzene, carbon nine aromatics, carbon ten aromatics, and other non-aromatic substances.

[0063] The step S1 of establishing a mathematical model for the para-xylene adsorption separation process further includes:

[0064] Step S11, based on the structural parameters of the simulated moving bed and the operating parameters of the para-xylene adsorption separation process, construct a mathematical model of the simulated moving bed. In this embodiment, the mathematical model is an adsorption separation simulated moving bed mechanism model, a polynomial empirical model and a data-driven model based on a graph convolutional neural network. The simulated moving bed mathematical model is used to describe the key data of the adsorption separation process, including the component concentration of each bed layer of the simulated moving bed, the composition of the input and output materials of the simulated moving bed, the para-xylene purity of the simulated moving bed, and the para-xylene yield of the simulated moving bed.

[0065] The mathematical model in step S11 further includes a polynomial empirical model:

[0066]

[0067] Among them, purity PX and yield PX represent the purity and yield of p-xylene respectively, R1, R2, R3 and R4 represent the reflux ratios of the four regions of the xylene adsorption separation device respectively, a i , b j , c k and d l R1, R2, R3 and R4 are respectively for purity PX The coefficients of n, m, p and q are respectively the purity of R1, R2, R3 and R4. PX The highest power in It is purity PX The constant term, e r , f s , g t and h x R1, R2, R3 and R4 are respectively PX The coefficients of u, v, w and y are the yield of R1, R2, R3 and R4 respectively. PX The highest power in is yield PX The polynomial coefficients, highest power and constant terms are calculated by least squares, gradient descent, polynomial kernel method, genetic algorithm, Newton's method and other methods. These methods each play an important role in the polynomial empirical model and help optimize the model parameters to obtain the best fit and prediction ability.

[0068] The mathematical model in step S11 includes an artificial neural network data-driven model. Based on the operating data of the para-xylene adsorption separation device, a feedforward neural network model, a feedback neural network model and a graph neural network can be used to establish a regression model, thereby realizing a data-driven model:

[0069] The feedforward neural network model has a network structure including an input layer, two hidden layers and an output layer. The input layer is the operating variable of the paraxylene adsorption separation device, and the output layer can be the purity and yield of paraxylene or the concentration of components in the extract and the residual liquid.

[0070] Feedback neural network model, the network structure includes an input layer, two RNN (recurrent neural network) layers and an output layer. Among them, the input layer contains the operation variables and historical component concentrations, the two RNN layers can be LSTM (long short-term memory neural network) or GRU (gated recurrent unit), and the output layer is a fully connected layer, which is used to predict the purity and yield of paraxylene or the concentration of components in the extract and the residual liquid.

[0071] Graph Neural Network (GNN) is a deep learning model that can learn the feature representation of nodes and capture the topological structure and relationship between nodes in the graph through graph convolution operations. The graph data construction module converts the operation variables into the adjacency matrix and feature matrix required by the graph convolution layer module, and predicts the purity and yield of paraxylene or the concentration of components in the extract and raffinate through the fully connected layer.

[0072] Step S12, comparing the mathematical model simulation experiment with the axial concentration distribution data collected under actual working conditions, verifying the mathematical model parameters, and ensuring that the calculation error of the mathematical model for xylene purity and yield is within the industrial allowable range.

[0073] Among them, the remaining process parameters of the mechanism model are: adsorption isotherm parameters, mass transfer coefficient, axial diffusion coefficient, chromatographic column geometry parameters, adsorbent characteristic parameters, etc. These parameters remain unchanged in subsequent optimization. In particular, the sample set of the mechanism model is randomly divided into ten subsets of equal size, and ten-fold cross validation is used to evaluate the prediction performance of the mechanism model. The evaluation indicators are mean absolute error (MAE), root mean square error (RMSE) and determination coefficient (R2).

[0074] Figure 2 Shows Figure 1 The architecture diagram of the graph convolutional neural network used in the robust optimization method of the para-xylene adsorption separation device shown in FIG. E , Q R , Q F , Q D , Q ref They represent the extraction flow rate, residual liquid flow rate, feed flow rate, analytical agent flow rate and circulation flow rate respectively, T represents the bed switching time, C PX , C MX , C OX and C EB Respectively represent the concentrations of PX, MX, OX and EB in the extract, such as Figure 2 As shown in the figure, the architecture consists of four modules: a graph data construction module for operational variables, a graph convolution layer module, a flattening layer module, and a fully connected layer module. The graph data construction module converts the operational variables into the adjacency matrix and feature matrix required by the graph convolution layer module. The graph convolution layer module includes two graph convolution layers. The purpose of the flattening layer module is to flatten the multi-dimensional input data into a one-dimensional vector to adapt to the subsequent fully connected layer. The input of the fully connected layer contains the values ​​of ten operational variables; the output of the fully connected layer contains the concentration values ​​of five components in 24 chromatographic columns (axial concentration distribution of five components). Each neuron in the hidden layer or output layer is determined by the weighted sum of all neurons in the previous layer.

[0075] Step S2: using the mathematical model established in step S1, the uncertainty of the input variables of the mathematical model is quantified by a polynomial chaos expansion method based on a sparse grid, that is, uncertainty propagation is performed on the input variables.

[0076] Step S2 further comprises the following steps:

[0077] Step S21, Latin hypercube sampling is performed on the operating variables of the simulated moving bed, each variable is uniformly sampled within a larger value range, and a sample set is formed after combination;

[0078] Step S22, performing simulation calculations on each sample, recording the axial concentration distribution data of PDEB, EB, MX, OX and PX under each set of operating variables, and establishing a sample set;

[0079] Step S23, using non-embedded polynomial chaos expansion to quantify uncertainty of uncertain operating variables, and constructing a polynomial chaos expansion agent model to realize uncertainty propagation;

[0080] Step S24, using the Galerkin projection method to solve the chaotic polynomial coefficients;

[0081] Step S25: Using sparse grid integration instead of full tensor integration to solve the chaotic polynomial coefficients.

[0082] In this embodiment, Latin hypercube sampling is used to define a four-dimensional sample space of the input variables.

[0083] In this embodiment, uncertainty quantification is divided into the following four important steps: identification of uncertainty sources, characterization of uncertainty sources, uncertainty propagation, and uncertainty analysis.

[0084] The specific steps are:

[0085] Identification of sources of uncertainty: Identify all input variables that may introduce uncertainty;

[0086] Characterization of sources of uncertainty: Choose an appropriate statistical representation for each uncertain input variable, characterize each random variable as a probability distribution, and determine the parameters of these distributions based on experimental data or literature data;

[0087] Uncertainty propagation: According to the determined probability distribution, appropriate orthogonal polynomials are selected to construct the polynomial chaotic expansion expression of the model output. The sparse grid method can calculate the coefficients in the polynomial chaotic expansion by selecting appropriate weights and node combinations;

[0088] Analyze uncertainty: Calculate the mean and standard deviation of the output based on the expansion coefficient.

[0089] Uncertain variables include the feed flow rate of the para-xylene adsorption separation unit, the para-xylene concentration in the feed, the intermediate xylene concentration in the feed, the o-xylene concentration in the feed, the ethylbenzene concentration in the feed, the temperature and pressure, etc.

[0090] Polynomial chaos expansion can be divided into two types: embedded and non-embedded:

[0091] Intrusive Polynomial Chaos Expansion (IPCE) deals with uncertainty propagation in dynamical systems by embedding the polynomial chaos expansion model into the system differential equations, but this introduces numerical errors and limits its application.

[0092] Non-intrusive Polynomial Chaos Expansion (NIPCE) adopts a more flexible approach, treating the system response function as a black box, and realizing the propagation of uncertainty by constructing a polynomial chaos expansion proxy model, thus avoiding the modification of the system equations. It is suitable for complex engineering analysis involving black box response functions and reduces the introduction of numerical errors.

[0093] Step S3, using the Sobol index to perform global sensitivity analysis, evaluate the impact of the feed flow rate and component concentration of the para-xylene adsorption separation device on the para-xylene adsorption separation device, screen key mathematical model input variables, and improve robust optimization efficiency.

[0094] The Sobol index is a statistical tool for sensitivity analysis based on full factorial ANOVA and is mainly used to determine which parts of the model output are most sensitive to the input variables.

[0095] The Sobol index can be directly calculated by the polynomial chaos expansion coefficient based on a sparse grid. Specifically, for each random variable, the first-order Sobol index is the ratio of the sum of the variances of all polynomial expansion terms related to the variable to the total variance, and the second-order Sobol index is the ratio of the sum of the variances of the polynomial expansion terms related to the interaction of the variable with other variables to the total variance. Through the above steps, the uncertainty contribution of random variables and their interactions to the model output can be quantified, so as to have a deeper understanding of the impact of each variable and its interaction in the system.

[0096] Step S3 further comprises the following steps:

[0097] Step S31, the sample set established in step 2 is randomly divided into ten subsets of equal size, and ten-fold cross validation is used to evaluate the predictive performance of the mathematical model; the data set contains 6000 samples, each sample includes 10 operating variables and their corresponding axial concentration distribution.

[0098] Step S32, based on the mathematical model, the uncertainty of the random model input variables is quantified to study and characterize the uncertainty of the model input. All uncertainty variables obey Gaussian distribution, and the range of variation is roughly set at ±0.9% of the initial value. When all uncertainty variables obey Gaussian distribution, the overall value of the second-order Sobol index of PX yield and purity is very small and can be ignored.

[0099] Step S33, perform a global sensitivity analysis using the Sobol index to evaluate the effects of feed flow rate and its component concentrations on PX purity and yield; the Sobol index includes first-order and second-order Sobol indices for PX purity and yield.

[0100] Among them, the SGPCE method (Sparse Grid based Polynomial Chaos Expansion), i.e., polynomial chaos expansion based on sparse grid, is used to perform Sobol index analysis on the finite element model;

[0101] The total variance of the target includes the variance of each parameter and its interaction, that is, the variance of each parameter and the interaction between them are taken into account. The specific calculation is as follows:

[0102]

[0103] Among them, Var(Y) represents the total variance of the target, D i (Y) represents the first-order variance of each input parameter, D ij (Y) and D 12…n (Y) represent the second-order variance and higher-order variance, respectively.

[0104] The Sobol index is evaluated based on the percentage contribution to the total variance:

[0105] First-order Sobol index:

[0106]

[0107] Second Order Sobol Index:

[0108] Calculation of the first and second order Sobol indices for PX purity and yield based on the SGPCE method.

[0109] Step S34: determine key random model input variables to improve robust optimization efficiency.

[0110] Step S4, calling the mathematical model established in step S1 through the NSGA-II multi-objective optimization algorithm, taking the mean and standard deviation obtained after quantifying the uncertainty of the performance indicators of the xylene adsorption separation device as the optimization target, optimizing the operating variables, and obtaining the optimal operating variables, thereby realizing robust optimization of the xylene adsorption separation device.

[0111] Step S4 further comprises the following steps:

[0112] Step S41, encoding and initializing the population required for generating the NSGA-Ⅱ algorithm by Latin hypercube sampling to ensure that the population has diversity;

[0113] The NSGA-II algorithm is used to solve optimization problems with multiple objective functions. The algorithm is based on a genetic algorithm and aims to find the Pareto optimal solution set of the problem. NSGA-II maintains the diversity of the population through non-dominated sorting and crowding sorting in order to find a set of approximately optimal solutions in the solution space, so that these solutions have no obvious room for improvement between the various objectives.

[0114] Step S42, calling the mathematical model obtained in step S2 to predict the axial concentration distribution of each individual, calculate the corresponding PX purity and yield, and provide basic data for subsequent uncertainty quantification;

[0115] Step S43, through step S23, applying the SGPCE method to quantify uncertainty and determine the mean and standard deviation of the PX purity yield;

[0116] Step S44: Calculate the fitness of each individual according to the constraints and objective function set in the NSGA-II algorithm, and determine whether it meets the design requirements;

[0117] Step S45: If the design requirements are not met, the crossover, mutation and selection process is performed and the process returns to step S2 until the set maximum number of iterations is reached and the iteration ends.

[0118] The following example illustrates the robust optimization method for the xylene adsorption separation unit.

[0119] In this example, a domestic petrochemical adsorption separation unit is used as the test object, and the robust optimization method of the para-xylene adsorption separation unit is specifically and detailedly explained.

[0120] Step S1, obtaining the process parameters and operation data of a petrochemical adsorption separation device, and establishing a mechanism model, the specific process is as follows:

[0121] Obtain physical parameters of the device described in step S1, such as device length, width, and diameter;

[0122] Calculate and obtain fixed parameters such as adsorption isotherm parameters, mass transfer coefficient, axial diffusion coefficient, chromatographic column geometric parameters, and adsorbent characteristic parameters of the device described in step S1;

[0123] Obtaining the production and operation parameters of the device in step S1 within one year, including but not limited to the inlet and outlet flow rates and compositions, the flow rates and compositions of each region, the reflux ratios of each region and other production parameters, and using them to build a mechanism model of the adsorption separation device;

[0124] Furthermore, Latin hypercube sampling was used to uniformly sample the operating variables and combine them to form a sample set. The simulation situation under each set of operating variables was obtained through simulation calculation using the mechanism model, and the axial concentration distribution data of the five components (PDEB, EB, MX, OX, and PX) in the 24 beds were recorded, which included a total of 6000 samples;

[0125] For the axial concentration distribution prediction task, a data-driven model based on graph convolutional neural network was established using this sample set, and the performance of four other classic machine learning models was compared. The main considerations were four indicators, namely, mean absolute error, root mean square error, determination coefficient, and single sample prediction time of each model. The specific results are shown in Table 1.

[0126] Table 1 Prediction results of data-driven model

[0127]

[0128] Compared with the mechanism model, the data-driven model shows unique computational advantages. The graph convolutional neural network (OV-GCN) shows the best prediction performance. Its mean absolute error and root mean square error are the lowest among all models, and the determination coefficient reaches 0.9984.

[0129] The results show that the graph convolutional neural network has the best prediction performance for the axial concentration distribution of the five components.

[0130] Step S2: using the mathematical model established in step S1, the uncertainty of the input variables of the mathematical model is quantified by a polynomial chaos expansion method based on a sparse grid, that is, uncertainty propagation is performed on the input variables.

[0131] In this embodiment, the specific process is as follows:

[0132] The uncertainty of the uncertain operating variables is quantified by NIPCE. X(θ) is a function representing the random process of random events. According to the probability distribution of the input parameters, the corresponding optimal orthogonal basis can be selected. In order to reduce the amount of calculation, the truncation strategy is often considered to remove some high-order cross polynomial terms, which can be expressed as follows:

[0133]

[0134] Among them, ψ j (ξ) has a one-to-one correspondence with the orthogonal basis of polynomial chaos expansion under multidimensional random parameters, ξ is a multidimensional random variable, k j is the corresponding coefficient, M represents the order of the highest-order term retained, n is a positive integer, and d is a positive integer less than n;

[0135] The method for solving the PCE coefficients of the present invention is the Galerkin projection method, which utilizes the orthogonal characteristics of the basis functions and solves the coefficients through inner product transformation decoupling;

[0136] The main idea of ​​the sparse grid based polynomial chaos expansion (SGPCE) method is to reduce the computational cost and solve the PCE coefficients by replacing the full tensor (FTP) integration with the sparse grid (SG) integration.

[0137] In particular, the calculation formulas for PX purity and yield are as follows:

[0138] Purity of PX in the extracted liquid:

[0139]

[0140] PX yield in the extracted liquid:

[0141]

[0142] in, and They represent the concentrations of PX, MX, OX and EB in the extract (i.e. the sixth bed), Q E , Q F , represent the extraction liquid flow rate and feed flow rate respectively, both of which can be calculated by the graph convolutional neural network model.

[0143] Step S3, using the Sobol index to perform a global sensitivity analysis, evaluate the impact of the feed flow rate and component concentrations of the p-xylene adsorption separation device on the p-xylene adsorption separation device, and screen key mathematical model input variables.

[0144] The present invention performs Sobol index global sensitivity analysis based on a data-driven model. The Sobol index analysis through SGPCE not only improves the efficiency of sensitivity analysis, but also takes into account the impact of the uncertainty of input random variables on the output. The specific calculation process is as follows:

[0145] The total variance of the target Var(Y) includes the variance of each parameter and its interaction, that is, the variance of the single parameter and the interaction between them are taken into account, which can be calculated by the following formula:

[0146]

[0147] Among them, D i (Y) represents each input parameter X i The first-order variance, D ij (Y) and D 12…n (Y) represents the second-order variance and higher-order variance respectively. The Sobol index is evaluated based on the percentage contribution to the total variance:

[0148] First-order Sobol index:

[0149]

[0150] Second Order Sobol Index:

[0151]

[0152] In this embodiment, all uncertainty variables obey Gaussian distribution, and the variation range is set at ±0.9% of the initial value. The results show that the PX concentration in the feed, the MX concentration in the feed, and the feed flow rate are the key uncertainty variables to be considered in the robust optimization, thereby improving the optimization efficiency to avoid falling into the dimensionality disaster.

[0153] Step S4, calling the mathematical model established in step S1 through the NSGA-II multi-objective optimization algorithm, taking the mean and standard deviation obtained after quantifying the uncertainty of the performance indicators of the xylene adsorption separation device as the optimization target, optimizing the operating variables, and obtaining the optimal operating variables, thereby realizing robust optimization of the xylene adsorption separation device.

[0154] Figure 3 Shows Figure 1 The axial concentration distribution robust optimization process of the robust optimization method embodiment of the para-xylene adsorption separation device shown in FIG. Figure 3 As shown, the axial concentration distribution optimization process of the present invention comprises the following steps:

[0155] Latin hypercube sampling is used to generate the population required by the NSGA-Ⅱ algorithm, and it is encoded and initialized to ensure that the population is diverse;

[0156] The data-driven model is used to predict the axial concentration distribution of each individual and calculate the corresponding PX purity and yield, providing basic data for subsequent uncertainty quantification;

[0157] The SGPCE method was applied to quantify uncertainty and determine the mean and standard deviation of PX yield;

[0158] According to the constraints and objective function set in the NSGA-II algorithm, the fitness of each individual is calculated and whether it meets the design requirements is determined;

[0159] If the design requirements are not met, the crossover, mutation and selection process is performed and the process returns to step S2 until the set maximum number of iterations is reached and the iteration ends.

[0160] According to the requirements of this implementation case, the optimization goal is determined as: robust optimization of the axial concentration distribution under certain working conditions.

[0161] This problem is a multi-objective optimization problem with constraints, and its mathematical description is as follows:

[0162] F(x)=[maxμ(yield PX ), minσ(yield PX )]

[0163] st99.70%≤purity PX ≤99.90%

[0164] 96.00%≤yield PX ≤99.00%

[0165] 310.000m 3 ·h -1 ≤Q E ≤410.000m 3 ·h -1

[0166] 1410.000m 3 ·h -1 ≤Q R ≤1470.000m 3 ·h -1

[0167] 2165.000m 3 ·h -1 ≤Q ref ≤2195.000m 3 ·h -1

[0168] Q D =Q E +Q R -Q F

[0169] 58.00s≤T≤65.00s

[0170] σ(yield PX )≤0.03

[0171] Among them, purity PX With yield PXRepresents the purity and yield of PX, μ(yield PX ) and σ(yield PX ) represent the mean and standard deviation of PX yield, Q E , Q R , Q F , Q D , Q ref They represent the extract flow rate, residual liquid flow rate, feed flow rate, analytical agent flow rate and circulation flow rate respectively, and T represents the bed switching time.

[0172] The mean of PX yield increases, and the standard deviation of PX yield increases, indicating that maximizing μ(yield PX ) and minimize σ(yield PX ) is a multi-objective optimization problem. The results of the deterministic optimization experiment are as follows Figure 4 The comparison results of the optimized operating variables are shown in Table 2

[0173] Table 2 Comparison of operating variables after optimization

[0174]

[0175]

[0176] The simulation results show that, compared with the deterministic optimization results, the robust optimization framework exhibits the characteristics of making the optimized axial concentration distribution robust, making the system or optimization scheme robust and reliable to external changes or internal disturbances; the robust optimization framework considers the uncertainty factors or the range of variation of parameters, and searches for solutions that not only perform well under deterministic conditions but also maintain good performance under uncertain conditions during the optimization process; it increases the stability of the system and makes the final optimization results more reliable.

[0177] In the optimization of axial concentration distribution, the improvement of robustness means that even in the face of external environmental changes, operating conditions fluctuations or model parameter uncertainties, the optimized axial concentration distribution can still remain relatively stable and reliable. This robustness feature is very important for practical engineering applications, because real systems are often affected by various uncertain factors, and robust optimization can better adapt to these changes and provide reliable solutions.

[0178] Figure 5 The principle of an embodiment of the robust optimization system of the para-xylene adsorption separation device of the present invention is shown. Figure 5 ,The system of this embodiment includes: a model building module, an uncertainty quantification module, a variable screening module, and a variable optimization module.

[0179] Model building module, referring to the operating variables of the para-xylene adsorption separation unit and the composition of the inlet and outlet materials of the para-xylene adsorption separation unit, establishes a mathematical model of the para-xylene adsorption separation process. Figure 1 Step S1 of the embodiment shown.

[0180] The uncertainty quantification module uses the mathematical model established by the model building module to quantify the uncertainty of the input variables of the mathematical model through the polynomial chaos expansion method based on sparse grids, that is, to propagate the uncertainty of the input variables. The specific processing of this module refers to Figure 1 Step S2 of the embodiment shown.

[0181] The variable screening module uses the Sobol index to perform global sensitivity analysis, evaluates the impact of the feed flow rate and component concentration of the para-xylene adsorption separation unit on the para-xylene adsorption separation unit, and screens key mathematical model input variables to improve the efficiency of robust optimization. The specific processing of this module is referenced in Figure 1 Step S3 of the embodiment shown.

[0182] The variable optimization module uses the mathematical model established by the uncertainty quantification module through the NSGA-II multi-objective optimization algorithm, takes the mean and standard deviation of the performance index uncertainty quantified by the para-xylene adsorption separation unit as the optimization target, optimizes the operating variables, and obtains the optimal operating variables, thereby realizing the robust optimization of the para-xylene adsorption separation unit. The specific processing of this module refers to Figure 1 Step S4 of the embodiment shown.

[0183] The present invention also discloses a computer-readable storage medium storing program instructions executable by a processor to implement the following Figure 1 The steps of an embodiment of a robust optimization method for a para-xylene adsorption separation device are shown.

[0184] The present invention also discloses a computer device, which comprises a memory, a processor and program instructions stored in the memory and executable by the processor, wherein the processor executes the program instructions to implement the following Figure 1 The steps of an embodiment of a robust optimization method for a para-xylene adsorption separation device are shown.

[0185] Although the above methods are illustrated and described as a series of actions for simplicity of explanation, it should be understood and appreciated that these methods are not limited by the order of the actions, because according to one or more embodiments, some actions may occur in a different order and / or concurrently with other actions from those illustrated and described herein or not illustrated and described herein but understandable to those skilled in the art.

[0186] Those skilled in the art will further appreciate that the various illustrative logic blocks, modules, circuits, and algorithm steps described in conjunction with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or a combination of the two. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps are generally described above in terms of their functionality. Whether such functionality is implemented as hardware or software depends on the specific application and the design constraints imposed on the overall system. The technician may implement the described functionality in different ways for each specific application, but such implementation decisions should not be interpreted as resulting in a departure from the scope of the present invention.

[0187] The various illustrative logic blocks, modules, and circuits described in conjunction with the embodiments disclosed herein may be implemented or performed with a general purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general purpose processor may be a microprocessor, but in the alternative, the processor may be any conventional processor, controller, microcontroller, or state machine. The processor may also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in cooperation with a DSP core, or any other such configuration.

[0188] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. The software module may reside in a RAM memory, a flash memory, a ROM memory, an EPROM memory, an EEPROM memory, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to a processor so that the processor can read and write information from / to the storage medium. In an alternative, a storage medium may be integrated into a processor. The processor and the storage medium may reside in an ASIC. The ASIC may reside in a user terminal. In an alternative, the processor and the storage medium may reside in a user terminal as discrete components.

[0189] In one or more exemplary embodiments, the functions described may be implemented in hardware, software, firmware, or any combination thereof. If implemented as a computer program product in software, each function may be stored on or transmitted by a computer-readable medium as one or more instructions or codes. Computer-readable media include both computer storage media and communication media, including any medium that facilitates the transfer of a computer program from one place to another. Storage media may be any available medium that can be accessed by a computer. As an example and not limitation, such a computer-readable medium may include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, disk storage or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of an instruction or data structure and can be accessed by a computer. Any connection is also properly referred to as a computer-readable medium. For example, if the software is transmitted from a website, a server, or other remote source using a coaxial cable, a fiber optic cable, a twisted pair, a digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwaves, the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwaves are included in the definition of the medium. Disk and disc as used herein include compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), floppy disk and Blu-ray disc, wherein disk often reproduces data magnetically, while disc reproduces data optically with lasers. Combinations of the above should also be included within the scope of computer-readable media.

[0190] The previous description of the disclosure is provided to enable any person skilled in the art to make or use the disclosure. Various modifications to the disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations without departing from the spirit or scope of the disclosure. Thus, the disclosure is not intended to be limited to the examples and designs described herein, but should be granted the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A robust optimization method for a para-xylene adsorption separation device, characterized in that: Methods include: Step S1, establishing a mathematical model of the p-xylene adsorption separation process with reference to the operating variables of the p-xylene adsorption separation device and the composition of the inlet and outlet materials of the p-xylene adsorption separation device; Step S2, using the mathematical model established in step S1, quantifying the uncertainty of the input variables of the mathematical model through a polynomial chaos expansion method based on a sparse grid, that is, propagating the uncertainty of the input variables; Step S3, using the Sobol index to perform global sensitivity analysis, evaluate the impact of the feed flow rate and component concentration of the para-xylene adsorption separation device on the para-xylene adsorption separation device, and screen key mathematical model input variables to improve robust optimization efficiency; Step S4, calling the mathematical model established in step S1 through the NSGA-II multi-objective optimization algorithm, taking the mean and standard deviation obtained after quantifying the uncertainty of the performance indicators of the xylene adsorption separation device as the optimization target, optimizing the operating variables, and obtaining the optimal operating variables, thereby realizing robust optimization of the xylene adsorption separation device.

2. The robust optimization method for a para-xylene adsorption separation device according to claim 1, characterized in that: The main operating variables of the para-xylene adsorption separation unit include: temperature, pressure, feed flow, desorbent flow, extract flow, residual liquid flow, circulation flow, switching time, xylene concentration in the feed, para-xylene concentration in the feed, o-xylene concentration in the feed, ethylbenzene concentration in the feed, and adsorbent specific capacity; the composition of the inlet and outlet materials of the para-xylene adsorption separation unit includes: para-xylene PX, ethylbenzene EB, m-xylene MX, o-xylene OX, para-diethylbenzene PDEB, carbon nine aromatics C9A, carbon ten aromatics C10A, and other non-aromatic substances.

3. The robust optimization method for a para-xylene adsorption separation device according to claim 1, characterized in that: Step S1 further comprises: Step S11, based on the structural parameters of the simulated moving bed and the operating parameters of the para-xylene adsorption separation process, construct a mathematical model of the simulated moving bed, including but not limited to a mechanism model, a polynomial empirical model and a data-driven model of the para-xylene simulated moving bed, the mathematical model of the simulated moving bed is used to describe key data of the adsorption separation process including the component concentration of each bed layer of the simulated moving bed, the composition of the input and output materials of the simulated moving bed, the para-xylene purity of the simulated moving bed, and the para-xylene yield of the simulated moving bed; Step S12, comparing the mathematical model simulation experiment with the axial concentration distribution data collected under actual working conditions, verifying the mathematical model parameters, and ensuring that the calculation error of the mathematical model for xylene purity and yield is within the industrial allowable range.

4. The robust optimization method for a para-xylene adsorption separation device according to claim 1, characterized in that: The mathematical model in step S11 further includes a polynomial empirical model: Among them, purity PX and yield PX represent the purity and yield of p-xylene respectively, R1, R2, R3 and R4 represent the reflux ratios of the four regions of the xylene adsorption separation device respectively, a i , b j , c k and d l R1, R2, R3 and R4 are respectively for purity PX The coefficients of i, j, k and l are the purity of R1, R2, R3 and R4 respectively. PX The different powers in n, m, p and q are the purity of R1, R2, R3 and R4 respectively. PX The highest power in C purityPX It is purity PX The constant term, e r , f s , g t and h x R1, R2, R3 and R4 are respectively PX The coefficients, r, s, t and x are the yields of R1, R2, R3 and R4 respectively. PX In the different powers, u, v, w and y are the yields of R1, R2, R3 and R4 respectively. PX The highest power in C yieldPX is yield PX The constant term of the polynomial is calculated by least squares, gradient descent, polynomial kernel method, genetic algorithm, Newton's method and other methods to help optimize the model parameters in the polynomial empirical model to obtain the best fit and prediction ability.

5. The robust optimization method for a para-xylene adsorption separation device according to claim 1, characterized in that: The mathematical model in step S11 includes a neural network data-driven model. Based on the operating data of the para-xylene adsorption separation device, a regression model is established using a feedforward neural network model, a feedback neural network model, and a graph neural network, thereby realizing a data-driven model: Feedforward neural network model, the network structure includes an input layer, two hidden layers and an output layer. The input layer is the operating variable of the p-xylene adsorption separation device, and the output layer is the purity and yield of p-xylene or the concentration of the components in the extract and the residual liquid; Feedback neural network model, the network structure includes an input layer, two recurrent neural network layers and an output layer, wherein the input layer contains operating variables and historical component concentrations, the two recurrent neural network layers are long short-term memory neural networks or gated recurrent units, and the output layer is a fully connected layer, which is used to predict the purity and yield of paraxylene or the concentration of components in the extract and the raffinate; Graph neural network is a deep learning model that learns the feature representation of nodes through graph convolution operations and captures the topological structure and relationship between nodes in the graph. The graph data construction module converts the operation variables into the adjacency matrix and feature matrix required by the graph convolution layer module, and predicts the purity and yield of paraxylene or the concentration of components in the extract and raffinate through the fully connected layer.

6. The robust optimization method for a para-xylene adsorption separation device according to claim 1, characterized in that: Step S2 further comprises the following steps: Step S21, Latin hypercube sampling is performed on the operating variables of the simulated moving bed, each variable is uniformly sampled within a larger value range, and a sample set is formed after combination; Step S22, performing simulation calculation on each sample, recording the axial concentration distribution data of p-xylene, ethylbenzene, m-xylene, o-xylene and p-xylene under each set of operating variables, and establishing a sample set; Step S23, using non-embedded polynomial chaos expansion to quantify uncertainty of uncertain operating variables, and constructing a polynomial chaos expansion agent model to realize uncertainty propagation; Step S24, using the Galerkin projection method to solve the chaotic polynomial coefficients; Step S25: Using sparse grid integration instead of full tensor integration to solve the chaotic polynomial coefficients.

7. The robust optimization method for a para-xylene adsorption separation device according to claim 1, characterized in that: Step S3 further comprises the following steps: Step S31, randomly dividing the sample set established in step 2 into ten subsets of equal size, and using ten-fold cross validation to evaluate the prediction performance of the mathematical model; Step S32, quantifying the uncertainty of random model input variables based on the mathematical model to study and characterize the uncertainty of the model input; Step S33, performing a global sensitivity analysis using the Sobol index to evaluate the effects of feed flow rate and its component concentrations on PX purity and yield; Step S34: determine key random model input variables to improve robust optimization efficiency.

8. The robust optimization method for a para-xylene adsorption separation device according to claim 1, characterized in that: Step S4 further comprises the following steps: Step S41, encoding and initializing the population required for generating the NSGA-Ⅱ algorithm by Latin hypercube sampling to ensure that the population has diversity; Step S42, calling the mathematical model obtained in step S3 to predict the axial concentration distribution of each individual, calculate the corresponding PX purity and yield, and provide basic data for subsequent uncertainty quantification; Step S43, applying the SGPCE method to quantify uncertainty and determine the mean and standard deviation of the PX purity yield; Step S44: Calculate the fitness of each individual according to the constraints and objective function set in the NSGA-II algorithm, and determine whether it meets the design requirements; Step S45: If the design requirements are not met, the crossover, mutation and selection process is performed and the process returns to step S2 until the set maximum number of iterations is reached and the iteration ends.

9. A robust optimization system for a para-xylene adsorption separation device, characterized in that: The system includes: The model building module refers to the operating variables of the para-xylene adsorption separation unit and the composition of the inlet and outlet materials of the para-xylene adsorption separation unit to establish a mathematical model of the para-xylene adsorption separation process; The uncertainty quantification module uses the mathematical model established by the model building module to quantify the uncertainty of the input variables of the mathematical model through the polynomial chaos expansion method based on sparse grids, that is, to propagate the uncertainty of the input variables; The variable screening module uses the Sobol index to perform global sensitivity analysis, evaluate the impact of the feed flow rate and component concentration of the para-xylene adsorption separation unit on the para-xylene adsorption separation unit, and screen key mathematical model input variables to improve the efficiency of robust optimization; The variable optimization module calls the mathematical model established by the model construction module through the NSGA-II multi-objective optimization algorithm, takes the mean and standard deviation obtained after quantifying the uncertainty of the performance indicators of the xylene adsorption separation unit as the optimization target, optimizes the operating variables, and obtains the optimal operating variables, thereby realizing the robust optimization of the xylene adsorption separation unit.

10. A computer-readable storage medium storing program instructions executable by a processor to implement the steps of the robust optimization method for a para-xylene adsorption separation device according to any one of claims 1 to 8.

11. A computer device comprising a memory, a processor and program instructions stored in the memory and executable by the processor, wherein the processor executes the program instructions to implement the steps of the robust optimization method for a para-xylene adsorption separation device according to any one of claims 1 to 8.

Citation Information

Patent Citations

  • Method for modeling and optimizing separation process of C8-aromatics simulated moving bed

    CN104331636A

  • Robust topological optimization method of thin-shell structure considering thickness uncertainty

    CN110287534A

  • Robust decoupling control system and control method for high-purity rectification process

    CN114995155A

  • Optimization method and system for separating p-xylene

    CN116072236A

  • Robust optimization method and robust optimization device for pollutant emission

    CN116341247A