Multi-objective optimization method and system for low-nitrogen combustion of a cracking furnace, and storage medium

Abstract

The application provides a kind of pyrolysis furnace low nitrogen combustion multi-objective optimization method, optimization system and storage medium.The multi-objective optimization method includes the following steps: establishing the turbulent combustion coupling model describing the flow and fuel combustion condition of the target pyrolysis furnace flue gas;Simplify the methane combustion mechanism of the turbulent combustion coupling model to obtain a simplified model describing the simplified mechanism of combustion reaction kinetics;Determine the input variables and output variables of the proxy model for predicting NOx emissions and flue heat efficiency;Obtain experimental sample points about the input variables and the output variables, and obtain the corresponding simulation results;Based on the experimental sample points and the simulation results, the proxy model is established via radial basis neural network;And the NSGA-II algorithm is used for multi-objective optimization of pyrolysis furnace low nitrogen combustion to determine the optimized input variable parameters balancing nitrogen emissions and thermal efficiency.

Description

Technical Field

[0001] This invention relates to the field of combustion numerical simulation, and in particular to a multi-objective optimization method for low-NOx combustion in a pyrolysis furnace, a multi-objective optimization system for low-NOx combustion in a pyrolysis furnace, and a computer-readable storage medium. Background Technology

[0002] Environment and development are major issues of universal concern in the international community today, and protecting the environment is a common task for all mankind. Ethylene production is an important standard for measuring a country's chemical industry level, but traditional ethylene cracking furnaces generate large amounts of NOx during production and operation, causing serious harm to the atmospheric environment and people's production and lives. Therefore, there is an urgent need in this field for a new technology and method for low-NOx combustion.

[0003] Studies of pyrolysis furnace linings have revealed that while traditional experimental methods are relatively accurate, they are prohibitively time-consuming and expensive, and make it difficult to observe NOx and component concentrations. Furthermore, combustion is a complex process involving intense chemical reactions and strong flow. While detailed reaction kinetics simulations yield accurate results, they require significant computational resources, are time-consuming, and struggle to balance low NOx emissions with high thermal efficiency. Therefore, selecting appropriate turbulence and combustion models to describe the pyrolysis furnace lining's operating mechanism is crucial for accurately predicting emissions, and simplifying this detailed mechanism using appropriate methods is also an important means of improving optimization efficiency.

[0004] In order to overcome the above-mentioned defects in the existing technology, there is an urgent need in the field for a multi-objective optimization technology for low-NOx combustion in pyrolysis furnaces. This technology simplifies the detailed mechanism of reaction kinetics by using reasonable simplification methods, and efficiently determines the optimal process conditions that balance low NOx emissions and high thermal efficiency. Summary of the Invention

[0005] The following provides a brief overview of one or more aspects to offer a basic understanding of them. This overview is not an exhaustive summary of all conceived aspects, nor is it intended to identify key or decisive elements of all aspects, nor to define the scope of any or all aspects. Its sole purpose is to present some concepts of one or more aspects in a simplified form as a prelude to the more detailed descriptions that follow.

[0006] To overcome the aforementioned deficiencies in the prior art, this invention provides a multi-objective optimization method for low-NOx combustion in a pyrolysis furnace, a multi-objective optimization system for low-NOx combustion in a pyrolysis furnace, and a computer-readable storage medium. These methods can simplify the detailed mechanism of reaction kinetics using reasonable simplification methods, thereby efficiently determining the optimal process conditions that balance low NOx emissions and high thermal efficiency.

[0007] Specifically, a multi-objective optimization method for low-NOx combustion in a pyrolysis furnace according to a first aspect of the present invention includes the following steps: establishing a turbulent combustion coupling model describing the flue gas flow and fuel combustion in the furnace of the target pyrolysis furnace; simplifying the methane combustion mechanism of the turbulent combustion coupling model using a directed graph method and a computational singular perturbation method to obtain a simplified model describing the simplified mechanism of combustion reaction kinetics; determining the input and output variables of a surrogate model for predicting NOx emissions and furnace thermal efficiency; obtaining experimental sample points for the input and output variables using a Latin hypercube sampling method, and performing computational fluid dynamics calculations based on the simplified model to obtain corresponding simulation results; establishing the surrogate model using a radial basis function neural network based on the experimental sample points and the simulation results; and performing multi-objective optimization of low-NOx combustion in the pyrolysis furnace using the NSGA-II algorithm based on the surrogate model, obtaining the Pareto solution set of NOx value and thermal efficiency within the permissible range of the input variables, to determine the optimal input variable parameters for balancing nitrogen emissions and thermal efficiency.

[0008] Furthermore, in some embodiments of the present invention, the step of establishing a turbulent combustion coupling model describing the flue gas flow and fuel combustion of the target pyrolysis furnace includes: constructing a geometric model and meshing the target pyrolysis furnace to determine the mesh model of the target pyrolysis furnace; and performing computational fluid dynamics simulation on the mesh model to determine the turbulent combustion coupling model.

[0009] Furthermore, in some embodiments of the present invention, the turbulent combustion coupling model is determined based on a turbulence model, a radiative heat transfer model, a flue gas radiation characteristic model, a combustion model, and / or a combustion reaction kinetic model. The turbulence model is the standard k-ε model. The radiative heat transfer model is a discrete coordinate model. The flue gas radiation characteristic model is a multi-ash gas weighted model. The combustion model is a vortex dissipation concept model. The combustion reaction kinetic model is determined based on the GRI3.0 methane combustion mechanism.

[0010] Furthermore, in some embodiments of the present invention, the step of simplifying the methane combustion mechanism of the turbulent combustion coupling model via the directed relation graph method and the computational singular perturbation method to obtain a simplified model describing the simplified mechanism of combustion reaction kinetics includes: performing a preliminary simplification of the GRI3.0 methane combustion mechanism via the directed relation graph method, spatially eliminating insensitive reaction components in its reaction mechanism to obtain a first-order simplified mechanism; performing a second simplification of the first-order simplified mechanism via the computational singular perturbation method to achieve decoupling between fast and slow reactions, and eliminating reactions with excessively fast and slow reaction rates in the first-order simplified mechanism to eliminate rigidity problems in the first-order simplified mechanism; and performing accuracy analysis and verification on the second-order simplified mechanism after the second simplification, and determining the simplified model based on the second-order simplified mechanism that meets the accuracy requirements.

[0011] Furthermore, in some embodiments of the present invention, the step of determining the input and output variables of the surrogate model for predicting NOx emissions and furnace thermal efficiency includes: based on prior knowledge and actual engineering operating parameters, determining the input variables of the surrogate model as fuel gas flow rate, excess air coefficient, and air preheating temperature, and determining the output variables of the simplified model as NOx emissions and furnace thermal efficiency. The furnace thermal efficiency is used to represent the combustion efficiency of the target pyrolysis furnace and is expressed as:

[0012]

[0013] In the formula, Q out Q is the outlet heat of the furnace. in Q represents the inlet heat of the furnace. combustion This represents the total heat released by the fuel.

[0014] Furthermore, in some embodiments of the present invention, the step of establishing a surrogate model for predicting NOx emissions and furnace thermal efficiency via a radial basis function neural network based on the experimental sample points and the simulation results includes: locally optimizing the experimental sample space obtained by the Latin hypercube sampling, removing at least one experimental sample point that contributes little to the fitting; establishing the surrogate model with the fuel gas flow rate, the excess air coefficient, and the air preheating temperature as input variables, and the NOx emissions and thermal efficiency as output variables, the regression expression of which is:

[0015]

[0016] In the formula, G i (x) is the neuron transfer function, expressed as:

[0017]

[0018] Among them, Cj σ represents the j-th center vector of the hidden layer within the proxy model. j F is the base width parameter of node j. j (X) represents the linear output value of the proxy model, ω j0 ω ji G(‖Xt) represents the connection weights of the output unit. i || ci ) represents the i-th output unit of the j-th node in the hidden layer; and based on the experimental sample space that has been locally optimized, the established surrogate model is trained using a radial basis neural network to determine a surrogate model that accurately predicts NOx emissions and furnace thermal efficiency.

[0019] Furthermore, in some embodiments of the present invention, the multi-objective optimization of low-NOx combustion in the pyrolysis furnace based on the surrogate model and the NSGA-II algorithm is used to obtain NO within the permissible range of the input variables. x The steps for determining the optimal input variable parameters for balancing nitrogen emissions and thermal efficiency, based on the Pareto solution set of the values ​​and thermal efficiency, include: determining the constraints of the multi-objective decision model based on the NSGA-II algorithm as follows:

[0020] q min ≤q≤q max

[0021] a min ≤a≤a max

[0022] temp min ≤temp≤temp max

[0023] Where q is the fuel gas flow rate, q min q max Let a be the minimum and maximum values ​​of the fuel gas flow rate, and a be the excess air coefficient. min a max The minimum and maximum values ​​of the excess air coefficient are given, where temp is the air preheating temperature. min temp max These are the minimum and maximum values ​​of the air preheating temperature;

[0024] Based on the aforementioned constraints, the multi-objective decision-making model is determined as follows:

[0025]

[0026]

[0027] as well as

[0028] Based on the surrogate model, the NSGA-II algorithm is used to determine the Pareto solution set of the multi-objective optimization model within the allowable range of the constraints, so as to determine the optimal input variable parameters for balancing nitrogen emissions and thermal efficiency.

[0029] Furthermore, in some embodiments of the present invention, the step of determining the Pareto solution set of the multi-objective optimization model within the permissible range of the constraints using the NSGA-II algorithm based on the surrogate model, in order to determine the optimization input variable parameters for balancing nitrogen emissions and thermal efficiency, includes: setting the initial population size N, maximum number of iterations, crossover probability, mutation probability, crossover distribution index, and mutation distribution index of the NSGA-II evolutionary algorithm to obtain the parent population P. t Where t = 1; for the parent population P t Multiple individuals in the population are sorted in a non-dominated manner to obtain multiple first non-dominated sets. The crowding degree of each individual in each first non-dominated set is calculated. Then, the offspring population Q is obtained through selection, crossover, and mutation. t The offspring population Q t With the parent population P t Merge to form a population R of size 2N. t According to the elite strategy, the population R t A non-dominated sort is performed to generate multiple second non-dominated sets, and the crowding degree of each individual in each second non-dominated set is calculated to produce the offspring population Q. t+1 ; Determine whether the current iteration count has reached the preset maximum iteration count; In response to the determination that the current iteration count has not reached the maximum iteration count, perform the next round of iteration calculation; and In response to the determination that the current iteration count has reached the maximum iteration count, output the optimized solution of the input variable in the current round as the optimized input variable parameter.

[0030] Furthermore, a multi-objective optimization system for low-NOx combustion in a pyrolysis furnace according to a second aspect of the present invention includes a memory and a processor. The memory stores computer instructions. The processor is connected to the memory and configured to execute the computer instructions stored in the memory to implement the multi-objective optimization method for low-NOx combustion in a pyrolysis furnace as described in any one of the first aspects of the present invention.

[0031] Furthermore, according to a third aspect of the present invention, a computer-readable storage medium is provided thereon storing computer instructions. When the computer instructions are executed by a processor, they implement a multi-objective optimization method for low-NOx combustion in a pyrolysis furnace as described in any one of the first aspects of the invention. Attached Figure Description

[0032] The above-described features and advantages of the present invention will be better understood after reading the following detailed description of embodiments of the present disclosure in conjunction with the accompanying drawings. In the drawings, components are not necessarily drawn to scale, and components having similar related characteristics or features may have the same or similar reference numerals.

[0033] Figure 1 A flowchart illustrating a multi-objective optimization method for low-NOx combustion in a pyrolysis furnace, provided according to some embodiments of the present invention, is shown.

[0034] Figure 2A and Figure 2B A schematic diagram of the geometric model and mesh model of an actual industrial pyrolysis furnace provided according to some embodiments of the present invention is shown.

[0035] Figure 3 A schematic diagram of a simplified process of reaction kinetics provided according to some embodiments of the present invention is shown.

[0036] Figure 4 A flowchart illustrating a fast non-dominated sorting genetic algorithm with an elitist strategy, provided according to some embodiments of the present invention, is shown.

[0037] Figure 5 A schematic diagram of an approximate Pareto front provided according to some embodiments of the present invention is shown. Detailed Implementation

[0038] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Although the description of the present invention is presented in conjunction with preferred embodiments, this does not mean that the features of the invention are limited to these embodiments. On the contrary, the purpose of describing the invention in conjunction with embodiments is to cover other options or modifications that may be derived based on the claims of the present invention. To provide a thorough understanding of the invention, many specific details will be included in the following description. The invention may also be implemented without using these details. Furthermore, to avoid confusion or obscuring the focus of the invention, some specific details will be omitted in the description.

[0039] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0040] Furthermore, the terms "upper," "lower," "left," "right," "top," "bottom," "horizontal," and "vertical" used in the following description should be understood as the orientations shown in the relevant paragraphs and accompanying drawings. These relative terms are for illustrative purposes only and do not imply that the described apparatus must be manufactured or operated in a specific orientation, and therefore should not be construed as limiting the invention.

[0041] It is understood that although terms such as "first," "second," and "third" may be used herein to describe various components, regions, layers, and / or parts, these components, regions, layers, and / or parts should not be limited by these terms, and these terms are only used to distinguish different components, regions, layers, and / or parts. Therefore, the first components, regions, layers, and / or parts discussed below may be referred to as second components, regions, layers, and / or parts without departing from some embodiments of the present invention.

[0042] As mentioned above, studies on pyrolysis furnace linings have revealed that while traditional experimental methods are relatively accurate, they are prohibitively expensive and time-consuming, and make it difficult to observe NOx and component concentrations. Furthermore, combustion is a complex process involving intense chemical reactions and strong flow. While detailed reaction kinetics simulations yield accurate results, they require significant computational resources, are time-consuming, and struggle to balance low NOx emissions with high thermal efficiency. Therefore, selecting appropriate turbulence and combustion models to describe the operating mechanism of the pyrolysis furnace lining is crucial for accurately predicting emissions, and simplifying this detailed mechanism using appropriate methods is an important means of improving optimization efficiency.

[0043] With the rapid advancement of computer technology, the optimal solution to related problems has gradually shifted from traditional experimental methods to the emerging field of Computational Fluid Dynamics (CFD). In particular, the development of Artificial Intelligence (AI) technology has provided new approaches to optimization problems through data-driven modeling using intelligent algorithms. To overcome the aforementioned shortcomings of existing technologies, this invention provides a multi-objective optimization method for low-NOx combustion in pyrolysis furnaces, a multi-objective optimization system for low-NOx combustion in pyrolysis furnaces, and a computer-readable storage medium. This method can simplify the detailed mechanism of reaction kinetics using reasonable simplification methods, thereby efficiently determining the optimal process conditions that balance low NOx emissions and high thermal efficiency.

[0044] In some non-limiting embodiments, the multi-objective optimization method for low-NOx combustion in a pyrolysis furnace provided in the first aspect of the present invention can be implemented based on the multi-objective optimization system for low-NOx combustion in a pyrolysis furnace provided in the second aspect of the present invention. Specifically, the multi-objective optimization system is equipped with a memory and a processor. The memory includes, but is not limited to, the computer-readable storage medium provided in the third aspect of the present invention, on which computer instructions are stored. The processor is connected to the memory and is configured to execute the computer instructions stored in the memory to implement the multi-objective optimization method for low-NOx combustion in a pyrolysis furnace provided in the first aspect of the present invention.

[0045] The working principle of the above-described multi-objective optimization system will be described below with reference to some embodiments of multi-objective optimization methods. Those skilled in the art will understand that these embodiments of multi-objective optimization methods are merely non-limiting implementations provided by the present invention, intended to clearly demonstrate the main concepts of the invention and provide specific solutions convenient for public implementation, rather than limiting all functions or all working methods of the multi-objective optimization system. Similarly, the multi-objective optimization system is also merely a few limiting implementations provided by the present invention, and does not constitute a limitation on the executing entity or execution order of the steps in these multi-objective optimization methods.

[0046] Please refer to the following first. Figure 1 , Figure 1 A flowchart illustrating a multi-objective optimization method for low-NOx combustion in a pyrolysis furnace, provided according to some embodiments of the present invention, is shown.

[0047] like Figure 1 As shown, in the process of performing a multi-objective optimization method for low-NOx combustion in a pyrolysis furnace, the multi-objective optimization system can first execute step S1: establish a turbulent combustion coupling model that describes the flue gas flow and fuel combustion in the furnace of the target pyrolysis furnace.

[0048] Please refer to the details. Figure 2A and Figure 2B , Figure 2A and Figure 2B A schematic diagram of the geometric model and mesh model of an actual industrial pyrolysis furnace provided according to some embodiments of the present invention is shown.

[0049] exist Figure 2A and Figure 2BIn the illustrated embodiment, the present invention uses an ethylene cracking furnace equipped with multi-stage fuel burners as the research object. The overall dimensions of this ethylene cracking furnace are 24.89m (length) × 2.964m (width) × 11.609m (height). The multi-stage bottom burners are installed near the side walls. Each multi-stage burner has one preheating air inlet, three primary fuel gas inlets for main combustion, and three secondary fuel gas inlets for auxiliary combustion. The flue gas outlet is located at the top of the furnace. Due to structural symmetry and calculation simplification, a 1 / 32 furnace structure is used for the simulation.

[0050] In establishing a coupled turbulent combustion model, for multi-objective systems, the Gambit software can first be used to perform geometric modeling of the target pyrolysis furnace to obtain, for example... Figure 2A The geometric model shown is then meshed and boundary layers are set to construct a structure as follows: Figure 2B The mesh model shown serves as the preliminary preparation for CFD calculations. It should be noted that the specific schemes for constructing the geometric model and mesh generation model are existing technologies in the field and do not involve any technical improvements to this invention; therefore, they will not be elaborated upon here.

[0051] Furthermore, after obtaining the mesh model of the target pyrolysis furnace, the present invention can perform CFD simulation on the mesh model to determine the turbulent combustion coupling model that describes the flue gas flow and fuel combustion of the target pyrolysis furnace.

[0052] Specifically, in some embodiments, the present invention may select the standard k-epsilon model as the turbulence model. Because Re is large in ethylene cracking furnace combustion, the flow of combustion flue gas is turbulent. This is because the cracking furnace involves complex combustion and heat transfer processes, requiring the solution of momentum, energy, mass, turbulent kinetic energy dissipation rate, kinetic energy, and component transport equations. Therefore, in non-swirling flames, the standard k-epsilon model exhibits superior performance.

[0053] Furthermore, this invention can also select the eddy dissipation concept (EDC) model as the combustion model. Since the NOx and CO formation process in pyrolysis furnace combustion is a slow reaction with a low reaction rate, the EDC model assumes that molecular mixing and subsequent reactions occur in small turbulent structures at the Kolmogorov scale, where kinetic energy is dissipated into heat. Given that the EDC model considers the detailed combustion chemical reaction mechanism within the turbulent structure, it can more accurately track the chemical reaction process, and its coupling with the turbulence model better reflects the actual physical phenomena within the pyrolysis furnace.

[0054] Furthermore, this invention can also select the Discrete Coordinate (DO) radiation model as the flue gas radiation model. Since the primary form of heat transfer in an ethylene cracking furnace is radiation, by using the DO radiation model to model radiation and employing the Weighted Multi-Grey Gas (WSGGM) model to calculate the absorption coefficient, this invention can divide the emissivity of the real gas into a weighted sum of several grey gases, thus achieving higher computational accuracy and efficiency.

[0055] Furthermore, this invention can also determine the detailed reaction kinetics mechanism of the turbulent combustion coupling model based on the currently common GRI3.0 mechanism for methane combustion.

[0056] Please continue to refer to this. Figure 1 After establishing the turbulent combustion coupling model, the multi-objective optimization system can perform step S2: simplify the methane combustion mechanism of the turbulent combustion coupling model through the directed relation graph method and the computational singular perturbation method to obtain a simplified model describing the simplified mechanism of combustion reaction kinetics, and determine the input and output variables of the surrogate model for predicting NOx emissions and furnace thermal efficiency.

[0057] Please refer to Figure 3 ,exist Figure 3 A schematic diagram of a simplified process of reaction kinetics provided according to some embodiments of the present invention is shown.

[0058] like Figure 3 As shown, in the process of simplifying the methane combustion mechanism of the turbulent combustion coupling model, the present invention can first use the PSR model in Chemkin software to conduct ignition experiments to obtain the reaction rates of each elementary reaction in the methane combustion GRI3.0 mechanism, and summarize the reaction rates of each elementary reaction and related parameters in an Excel spreadsheet.

[0059] Subsequently, this invention can use DRG and its derivative methods to simplify the skeletal reaction mechanism to remove redundant components. The DGR algorithm can be briefly expressed as follows:

[0060]

[0061]

[0062]

[0063] Among them, R AB Let ν be the orthogonal contribution rate of substance B to substance A. A,i ω represents the stoichiometric coefficient of component A in the detailed chemical mechanism. i k represents the reaction rate of the i-th elementary reaction. fi The forward reaction rate of the i-th elementary reaction can be expressed as:

[0064]

[0065] In the formula, A i Let n be the pre-exponential factor of the elementary reaction, T be the reaction temperature, and n be the... i E represents the temperature coefficient. i The activation energy is represented by F, where R is the gas reaction constant and F is the activation energy. i This is a correction item.

[0066] k bi The backward reaction rate of the i-th elementary reaction can be calculated using the following formula:

[0067] k bi =k fi / K ci

[0068] Among them, K ci It can be expressed as concentration and can be obtained from the phase equilibrium constant of the thermodynamic properties of pressure.

[0069] In this experiment, the threshold ε is set to 0.01. When R AB When the threshold is less than ε, substance B can be considered a substance that contributes little to the reaction and thus eliminated. A source program designed using the VS platform allows for the rapid elimination of unimportant components simply by inputting the target component and threshold. This simplifies the GRI3.0 reaction mechanism, which involves 53 components and 325 reactions, into a first-order simplified reaction kinetic mechanism involving only 27 components and 132 reactions.

[0070] Furthermore, the present invention can further simplify this first-order simplified reaction kinetic mechanism using the CSP method. The CSP method is briefly expressed as follows:

[0071] For a reaction system containing R elementary reactions, its N unknown quantities, such as the concentration of chemical reactants or temperature, are represented by a column vector y = (y 1 ,y 2 ,…,y N ) T It satisfies the quasi-linear ordinary differential equation:

[0072]

[0073] Where g(y,t) is the sum of contributions from R elementary reactions, called the total reaction rate vector, and its physical expression is:

[0074]

[0075] Among them, s r With F r These are the generalized stoichiometric vector and reaction rate of the r-th elementary reaction, respectively. Let an (t) represents N linearly independent column basis vectors, b n (t) are N row basis vectors, and have Then g(y,t) can be expressed as the sum of N reaction modes:

[0076]

[0077]

[0078] In the formula, a n and f n These represent the direction and amplitude of the nth mode, respectively, and their physical meanings are the effective generalized stoichiometry vector and effective reaction rate of the nth mode. It is called the augmented stoichiometric coefficient.

[0079] Define the participation index:

[0080]

[0081] In the formula, Δy is the allowable error vector of the unknown quantity, and Δt is the time scale.

[0082] Given a threshold ε p If the participation index of a certain elementary reaction is less than ε p If so, it can be considered a reaction with low participation and can be removed.

[0083] Specifically, after simplifying the first-level skeletal reaction mechanism, elementary reactions involving components with small contributions to the reaction have been removed. This invention can obtain the Jacobian matrix using the stoichiometric coefficient matrix and reaction rate vector corresponding to this first-level simplified mechanism, as well as the expressions for the reaction system component vector and component reaction rate vector. Then, this invention can input the component concentration at the time point of interest into this Jacobian matrix to obtain the eigenvalues ​​and eigenvectors of the Jacobian matrix at this point. Based on the eigenvalues, the number of fast reaction modes at this point is determined, and the eigenvectors are considered as basis vectors. Next, this invention can use a two-step correction method to correct the trial basis vectors, thereby achieving the separation of fast and slow modes. Finally, this invention can obtain the fast spatial mapping matrix and determine the quasi-steady-state components based on the corrected basis vectors. Based on the identified quasi-steady-state components, the corresponding reactions are obtained, and the quasi-steady-state components and related reactions are then removed from the entire reaction system to eliminate the rigidity problem of the reaction. The primary simplified mechanism obtained in the previous step is further simplified into a second-level simplified reaction kinetic mechanism involving only 27 components and 82 reactions.

[0084] Subsequently, the present invention can verify the accuracy of the secondary simplified mechanism obtained through the above two steps, and determine the simplified reaction kinetic mechanism after the accuracy meets the requirements.

[0085] Please continue to refer to this. Figure 1 After obtaining the simplified model describing the simplified mechanism of combustion reaction kinetics, the multi-objective optimization system can perform step S3: use the Latin hypercube sampling method to obtain experimental sample points for the input and output variables, and perform CFD calculations based on the simplified model to obtain the corresponding simulation results.

[0086] Specifically, the multi-objective optimization system can first determine the fuel gas flow rate, excess air coefficient and air preheating temperature as input variables based on prior knowledge and industrial practice, and then obtain experimental sample points by sampling using the Latin hypercube sampling method.

[0087] Furthermore, to improve simulation efficiency, the multi-objective optimization system can preferably remove points that contribute little to the model fitting and sparsely select values ​​for points in the sample space where the excess air coefficient is outside the range of 0.95-1.15. Then, based on the optimized sample space, the multi-objective optimization system can perform CFD simulation using the turbulent combustion coupling model with embedded simplified mechanisms established above, and statistically calculate the results. The specific formula for calculating furnace thermal efficiency is shown below:

[0088]

[0089] Among them, Q combustion Q represents the total heat release of the fuel gas. in The amount of heat required to preheat the air, Q out This includes the heat carried away by the flue gas and the heat lost through the furnace walls.

[0090] Please continue to refer to this. Figure 1 After determining the experimental sample points and simulation results of the input and output variables, the multi-objective optimization system can execute step S4: based on the experimental sample points and simulation results, establish a surrogate model via a radial basis function (RBF) neural network.

[0091] Specifically, in the process of establishing a surrogate model for predicting NOx emissions and furnace thermal efficiency via a radial basis function neural network, this invention can locally optimize the experimental sample space obtained by Latin hypercube sampling, removing at least one experimental sample point that contributes little to the fitting. Then, this invention can use fuel gas flow rate, the excess air coefficient, and air preheating temperature as input variables, and NOx emissions and thermal efficiency as output variables to establish the surrogate model, whose regression expression is:

[0092]

[0093] In the formula, G i(x) is the neuron transfer function, expressed as:

[0094]

[0095] Among them, C j σ represents the j-th center vector of the hidden layer within the proxy model. j F is the base width parameter of node j. j (X) represents the linear output value of the proxy model, ω j0 ω ji G(‖Xt) represents the connection weights of the output unit. i || ci ) represents the i-th output unit of the j-th node in the hidden layer.

[0096] Subsequently, the present invention can train the established surrogate model using a radial basis function neural network based on a locally optimized experimental sample space to determine a surrogate model that accurately predicts NOx emissions and furnace thermal efficiency.

[0097] Therefore, after extensive computation, the multi-objective optimization system can construct a database using the corresponding inputs and outputs as a dataset. Through extensive training with an RBF neural network using Matlab software, a multi-feature fast-response surrogate model is built. The final surrogate model can be represented by the following expression:

[0098]

[0099]

[0100] in For NOx emissions, η firebox For furnace thermal efficiency.

[0101] Please continue to refer to this. Figure 1 After establishing the surrogate model, the multi-objective optimization system can execute step S5: Based on the surrogate model, the NSGA-II algorithm is used to perform multi-objective optimization of low-NOx combustion in the pyrolysis furnace, and the Pareto solution set of NOx value and thermal efficiency is obtained within the allowable range of input variables to determine the optimal input variable parameters for balancing nitrogen emissions and thermal efficiency.

[0102] Specifically, this study aims to maximize the furnace thermal efficiency of the target pyrolysis furnace while maintaining low NOx concentrations. However, existing low-NOx combustion strategies often reduce the temperature of the combustion zone, thereby decreasing furnace thermal efficiency. Therefore, this invention uses the NSGA-II fast non-dominated sorting genetic algorithm with an elitist strategy to perform multi-objective optimization on the multi-feature fast response model constructed in the previous step, in order to obtain operating parameters that combine low NOx emissions with high furnace thermal efficiency.

[0103] In the design process of this study, the multi-objective optimization system can take the NOx emission at the furnace outlet and the furnace thermal efficiency as optimization objectives, and combine the multi-feature fast response surrogate model of the pyrolysis furnace and the selected multi-objective optimization algorithm to continuously optimize the selected operating parameters such as fuel gas flow rate, excess air coefficient and air preheating temperature.

[0104] Furthermore, the objective function of this study is to minimize NOx emission concentration while maximizing furnace thermal efficiency. Constraints are added to improve the optimization effect; these constraints are the reasonable control ranges of each decision variable, and the decision variables are the operating parameters selected during the surrogate model establishment process. By iteratively approaching the optimal solution through the NSGA-II evolutionary algorithm, this invention can ultimately obtain the optimal design scheme for the structural parameters of the ultra-low NOx burner.

[0105] Furthermore, the constraints of the multi-objective decision-making model in step S5 can be specifically defined as follows:

[0106] q min ≤q≤q max

[0107] a min ≤a≤a max

[0108] temp min ≤temp≤temp max

[0109] Where q is the fuel gas flow rate, q min q max Here, represents the minimum and maximum fuel flow rates; 'a' represents the excess air coefficient. min a max The minimum and maximum values ​​of the excess air coefficient; temp is the air preheating temperature. min temp max These are the minimum and maximum values ​​of the air preheating temperature.

[0110] The multi-objective decision model is as follows:

[0111]

[0112]

[0113]

[0114] Subsequently, the multi-objective optimization system can use the NSGA-II evolutionary algorithm based on this multi-objective decision model to find the optimized burner structural parameters. Please refer to [link / reference needed] for details. Figure 4 , Figure 4A flowchart of a fast non-dominated sorting genetic algorithm with an elitist strategy, provided according to some embodiments of the present invention, is shown.

[0115] like Figure 4 As shown, in using the NSGA-II evolutionary algorithm to find the optimal design scheme for the structural parameters of an ultra-low NOx burner, the multi-objective optimization system can first set parameters related to the NSGA-II evolutionary algorithm, such as the initial population size N, maximum number of iterations, crossover probability, mutation probability, crossover distribution index, and mutation distribution index, to obtain the parent population P. t (t=1). Afterwards, the multi-objective optimization system can optimize the parent population P. t Multiple individuals in the population are sorted in a non-dominated manner to obtain multiple first non-dominated sets. The crowding degree of each individual in each non-dominated set is calculated. Then, the offspring population Q is obtained through selection, crossover, and mutation. t Subsequently, the multi-objective optimization system can reduce the offspring population Q. t With parent population P t Merge to form a population R of size 2N. t And based on the elite strategy, the population R t Perform a non-dominated sort to generate multiple second non-dominated sets, and calculate the crowding degree of each individual in each second non-dominated set to produce a new offspring population Q. t+1 Next, the multi-objective optimization system can determine whether the current iteration count has reached the preset maximum iteration count. If the current iteration count has not reached the maximum iteration count, the multi-objective optimization system can set t = t + 1 and proceed to the next iteration count. Conversely, if the current iteration count has reached the maximum iteration count, the multi-objective optimization system can output the optimal solution for the decision variables and terminate the algorithm.

[0116] Please refer to the reference. Figure 5 And Table 1. Figure 5 A schematic diagram of an approximate Pareto front provided according to some embodiments of the present invention is shown. Table 1 gives the operating parameter values ​​corresponding to the optimized concentration boundary.

[0117] Table 1. Operational parameters and results corresponding to the optimization boundary.

[0118]

[0119] Compared to traditional back propagation (BP) neural networks, the RBF neural network used in this invention has advantages such as fast learning speed, high data selection efficiency, and strong classification ability. Especially when dealing with multiple variables, it possesses strong error correction and classification capabilities, and can handle difficult-to-analyze patterns within the system, making it a superior model construction method. Furthermore, the NSGA-II algorithm, as a practical, efficient, and robust optimization algorithm, is particularly adept at solving multi-objective optimization problems. Therefore, by implementing the multi-objective optimization method for low-NOx combustion in pyrolysis furnaces based on simplified reaction mechanisms provided by this invention, decision-makers can select a set of non-dominated solutions to guide actual industrial operating parameters according to their needs and preferences.

[0120] In summary, the multi-objective optimization method for low-NOx combustion in pyrolysis furnaces based on a simplified reaction mechanism provided by this invention has the following specific optimization effects:

[0121] (1) Using the geometric model of the pyrolysis furnace combined with the turbulence and combustion models, we can complete the effective CFD calculation of the turbulent combustion mechanism and achieve accurate prediction of the outlet NOx and flue gas component concentrations.

[0122] (2) Innovatively, a two-step simplification method is used to simplify the detailed reaction kinetic mechanism on both spatial and temporal scales, which greatly reduces the computation time while ensuring that the accuracy still meets the requirements;

[0123] (3) A fast response proxy model for multi-feature pyrolysis furnace was constructed based on radial basis function (RBF) neural network, which greatly reduced the time spent on frequent CFD calculations;

[0124] (4) NOx and furnace thermal efficiency are taken as target values, and the NSGA-II evolutionary algorithm is used to search within the allowable range of the cracking furnace operating parameters to solve the multi-objective optimization problem model of the cracking furnace combustion process, so as to propose a design scheme for the operating parameters of the low-NOx burner, so that decision-makers can select the corresponding dominant solution according to their needs and preferences to guide actual production.

[0125] Although the methods described above are illustrated and depicted as a series of actions for the sake of simplicity, it should be understood and appreciated that these methods are not limited by the order of the actions, as some actions may occur in a different order and / or concurrently with other actions from the illustrations and descriptions herein or not illustrated and described herein but which may be understood by those skilled in the art, according to one or more embodiments.

[0126] Those skilled in the art will understand that information, signals, and data can be represented using any of a variety of different techniques and arts. For example, the data, instructions, commands, information, signals, bits, symbols, and chips described throughout the above description can be represented by voltage, current, electromagnetic waves, magnetic fields or magnetic particles, light fields or optical particles, or any combination thereof.

[0127] 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 can be implemented as electronic hardware, computer software, or a combination of both. To clearly illustrate this interchangeability between hardware and software, the various illustrative components, blocks, modules, circuits, and steps are described above in a generalized manner 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. Those skilled in the art may implement the described functionality in different ways for each specific application, but such implementation decisions should not be construed as departing from the scope of the invention.

[0128] The prior description of this disclosure is provided to enable any person skilled in the art to make or use this disclosure. Various modifications to this 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 this disclosure. Therefore, this disclosure is not intended to be limited to the examples and designs described herein, but should be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A multi-objective optimization method for low-NOx combustion in a pyrolysis furnace, characterized in that, Includes the following steps: Establish a turbulent combustion coupled model to describe the flue gas flow and fuel combustion in the furnace of the target pyrolysis furnace; The methane combustion mechanism of the turbulent combustion coupling model is simplified by using the directed relation graph method and the computational singular perturbation method to obtain a simplified model describing the simplified mechanism of combustion reaction kinetics. Determine the input and output variables of the surrogate model for predicting NOx emissions and furnace thermal efficiency; The Latin hypercube sampling method is used to obtain experimental sample points for the input and output variables, and computational fluid dynamics calculations are performed based on the simplified model to obtain the corresponding simulation results. Based on the experimental sample points and the simulation results, the surrogate model is established via a radial basis function neural network; and Based on the surrogate model, the NSGA-II algorithm is used for multi-objective optimization of low-NOx combustion in the pyrolysis furnace. Within the allowable range of the input variables, the Pareto solution set of NOx value and thermal efficiency is obtained to determine the optimal input variable parameters for balancing nitrogen emissions and thermal efficiency.

2. The multi-objective optimization method as described in claim 1, characterized in that, The steps for establishing a turbulent combustion coupling model describing the flue gas flow and fuel combustion in the target pyrolysis furnace include: A geometric model of the target pyrolysis furnace is constructed and a mesh is generated to determine the mesh model of the target pyrolysis furnace; and Computational fluid dynamics simulations were performed on the mesh model to determine the turbulent combustion coupling model.

3. The multi-objective optimization method as described in claim 2, characterized in that, The turbulent combustion coupling model is determined based on the turbulence model, radiation heat transfer model, flue gas radiation characteristics model, combustion model, and / or combustion reaction kinetics model. The turbulence model is the standard k-ε model, the radiation heat transfer model is the discrete coordinate model, the flue gas radiation characteristics model is the multi-ash gas weighted model, the combustion model is the eddy dissipation concept model, and the combustion reaction kinetics model is determined based on the GRI3.0 methane combustion mechanism.

4. The multi-objective optimization method as described in claim 3, characterized in that, The steps of simplifying the methane combustion mechanism of the turbulent combustion coupling model using the directed relation graph method and the singular perturbation calculation method to obtain a simplified model describing the simplified mechanism of combustion reaction kinetics include: The GRI3.0 methane combustion mechanism is initially simplified by using the directed relation graph method, and insensitive reaction components in the reaction mechanism are spatially eliminated to obtain a first-order simplified mechanism. The first-order simplified mechanism is further simplified using the computational singular perturbation method to decouple fast and slow responses and eliminate reactions that are too fast or too slow in the first-order simplified mechanism, thereby eliminating the rigidity problem in the first-order simplified mechanism; and The accuracy of the second-level simplification mechanism after the second simplification is analyzed and verified, and the simplified model is determined based on the second-level simplification mechanism that meets the accuracy requirements.

5. The multi-objective optimization method as described in claim 4, characterized in that, The steps for determining the input and output variables of the surrogate model for predicting NOx emissions and furnace thermal efficiency include: Based on prior knowledge and actual engineering operation parameters, the input variables of the surrogate model are determined to be fuel gas flow rate, excess air coefficient, and air preheating temperature, and the output variables of the simplified model are determined to be NOx emissions and furnace thermal efficiency. The furnace thermal efficiency is used to represent the combustion efficiency of the target pyrolysis furnace and is expressed as: In the formula, Q out Q is the outlet heat of the furnace. in Q represents the inlet heat of the furnace. combustion This represents the total heat released by the fuel.

6. The multi-objective optimization method as described in claim 5, characterized in that, The step of establishing a surrogate model for predicting NOx emissions and furnace thermal efficiency via a radial basis function neural network based on the experimental sample points and the simulation results includes: Local optimization is performed on the experimental sample space obtained by the Latin hypercube sampling to remove at least one experimental sample point that has a low contribution to the fitting. Using the fuel gas flow rate, the excess air coefficient, and the air preheating temperature as input variables, and the NOx emission and thermal efficiency as output variables, the surrogate model is established, and its regression expression is: In the formula, G i (x) is the neuron transfer function, expressed as: Among them, C j σ represents the j-th center vector of the hidden layer within the proxy model. j F is the base width parameter of node j. j (X) represents the linear output value of the proxy model, ω j0 ω ji G(||Xt) represents the connection weights of the output unit. i || ci ) represents the i-th output unit of the j-th node in the hidden layer; and Based on the locally optimized experimental sample space, the established surrogate model is trained using a radial basis function neural network to determine a surrogate model that accurately predicts NOx emissions and furnace thermal efficiency.

7. The multi-objective optimization method as described in claim 6, characterized in that, The steps of performing multi-objective optimization of low-NOx combustion in a pyrolysis furnace based on the surrogate model and using the NSGA-II algorithm to obtain the Pareto solution set of NOx value and thermal efficiency within the allowable range of the input variables, and determining the optimization input variable parameters for balancing nitrogen emissions and thermal efficiency, include: The constraints for the multi-objective decision-making model based on the NSGA-II algorithm are determined as follows: q min ≤q≤q max a min ≤a≤a max temp min ≤temp≤temp max Where q is the fuel gas flow rate, q min q max Let a be the minimum and maximum values ​​of the fuel gas flow rate, and a be the excess air coefficient. min a max The minimum and maximum values ​​of the excess air coefficient are given, where temp is the air preheating temperature. min temp max These are the minimum and maximum values ​​of the air preheating temperature; Based on the aforementioned constraints, the multi-objective decision-making model is determined as follows: as well as Based on the surrogate model, the NSGA-II algorithm is used to determine the Pareto solution set of the multi-objective optimization model within the allowable range of the constraints, so as to determine the optimal input variable parameters for balancing nitrogen emissions and thermal efficiency.

8. The multi-objective optimization method as described in claim 7, characterized in that, The step of determining the Pareto solution set of the multi-objective optimization model within the allowable range of the constraints based on the surrogate model, and thus determining the optimization input variable parameters for balancing nitrogen emissions and thermal efficiency, includes: Set the initial population size N, maximum number of iterations, crossover probability, mutation probability, crossover distribution index, and mutation distribution index for the NSGA-II evolutionary algorithm to obtain the parent population P. t , where t = 1; For the parent population P t Multiple individuals in the population are sorted in a non-dominated manner to obtain multiple first non-dominated sets. The crowding degree of each individual in each first non-dominated set is calculated. Then, the offspring population Q is obtained through selection, crossover, and mutation. t ; The offspring population Q t With the parent population P t Merge to form a population R of size 2N. t According to the elite strategy, the population R t A non-dominated sort is performed to generate multiple second non-dominated sets, and the crowding degree of each individual in each second non-dominated set is calculated to produce the offspring population Q. t+1 ; Determine whether the current iteration count has reached the preset maximum iteration count; In response to the determination that the current iteration number has not reached the maximum iteration number, the next round of iteration calculation is performed; and In response to the determination that the current iteration number has reached the maximum iteration number, the optimized solution of the input variable in the current round is output as the parameter of the optimized input variable.

9. A multi-objective optimization system for low-NOx combustion in a pyrolysis furnace, characterized in that, include: Memory, on which computer instructions are stored; and A processor, connected to the memory, is configured to execute computer instructions stored in the memory to implement the multi-objective optimization method for low-NOx combustion in a pyrolysis furnace as described in any one of claims 1 to 8.

10. A computer-readable storage medium storing computer instructions thereon, characterized in that, When the computer instructions are executed by the processor, the multi-objective optimization method for low-NOx combustion in a pyrolysis furnace as described in any one of claims 1 to 8 is implemented.

Images