Micro-channel reactor design method, device and storage medium based on multi-objective bayesian optimization

Abstract

The application relates to the technical field of reactor design, and particularly discloses a micro-channel reactor design method and device based on multi-target Bayesian optimization and a storage medium, which comprises the following steps: performing a parameter sampling operation to generate an initial parameter combination; performing a physical field simulation operation for the micro-channel reactor based on the initial parameter combination to generate initial simulation information, wherein the initial simulation information comprises multiple simulation indexes; performing multi-target scoring calculation on the initial simulation information, adjusting the simulation indexes based on the scoring calculation result to generate adjusted simulation information; optimizing the adjusted simulation information based on a preset iterative optimization algorithm to generate optimized information; and configuring and designing the micro-channel reactor based on the optimized information. The existing micro-channel reactor is simulated and analyzed from multiple aspects, so that efficient and rapid micro-channel reactor design is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of reactor design, and in particular to a micro-channel reactor design method and device based on multi-objective Bayesian optimization and a storage medium. BACKGROUND

[0002] With the continuous development of technology, people's requirements for pharmaceutical technology are constantly improving. In order to overcome various defects of traditional kettle reactors in the pharmaceutical process, people propose to use micro-channel reactors for pharmaceutical control, which significantly improves reaction efficiency and safety by efficiently mixing, heat transfer and mass transfer of reactants in a small space, and is one of the core equipment of flow chemistry.

[0003] In the preparation of different drugs, it is often necessary to design a micro-channel reactor that meets the requirements. In the existing micro-channel reactor design work, the technical personnel find that the existing technology at least has the following technical problems:

[0004] In the first aspect, the traditional micro-channel reactor design relies on trial and error method and expert experience, which is low in efficiency and difficult to balance multi-objective conflicts (such as high conversion rate and low energy consumption); in the second aspect, the existing simulation optimization method (such as genetic algorithm) is usually aimed at a single target, and cannot achieve the expected optimization effect when facing multiple targets; the third aspect, high-precision simulation (such as CFD, chemical reaction kinetics coupling model) takes a long time, and the traditional optimization method needs a large number of simulation times, which is difficult to be applied in engineering, and cannot meet the actual needs of enterprises. SUMMARY

[0005] In order to overcome the above technical problems existing in the prior art, the embodiments of the present application provide a micro-channel reactor design method, device and storage medium based on multi-objective Bayesian optimization, which simulates and analyzes the existing micro-channel reactor from multiple aspects, thereby realizing efficient and rapid micro-channel reactor design.

[0006] In order to achieve the above purpose, the embodiments of the present application provide a micro-channel reactor design method based on multi-objective Bayesian optimization, which comprises: performing a parameter sampling operation to generate an initial parameter combination; performing a physical field simulation operation for the micro-channel reactor based on the initial parameter combination to generate initial simulation information, the initial simulation information comprising a plurality of simulation indexes; performing multi-objective scoring calculation on the initial simulation information, adjusting the simulation indexes based on the scoring calculation result to generate adjusted simulation information; optimizing the adjusted simulation information based on a preset iterative optimization algorithm to generate optimized information; and configuring and designing the micro-channel reactor based on the optimized information.

[0007] Preferably, the execution parameter sampling operation, generating an initial parameter combination, comprises: determining the geometric structure, operation condition and material characteristic of the micro-channel reactor; determining geometric parameters corresponding to the geometric structure; determining operation parameters based on the operation condition and material parameters based on the material characteristic; processing the geometric parameters, the operation parameters and the material parameters based on a Latin hypercube sampling algorithm to generate the initial parameter combination.

[0008] Preferably, the physical field simulation operation for the micro-channel reactor based on the initial parameter combination, generating initial simulation information, comprises: performing a fluid dynamics simulation analysis based on the geometric parameters to generate fluid simulation information; performing a chemical reaction simulation analysis based on the material parameters to generate chemical simulation information; performing a heat transfer simulation analysis based on the operation parameters to generate heat transfer simulation information; and generating the initial simulation information based on the fluid simulation information, the chemical simulation information and the heat transfer simulation information.

[0009] Preferably, the fluid dynamics simulation analysis based on the geometric parameters, generating fluid simulation information, comprises: performing a fluid dynamics simulation based on the geometric parameters to generate initial simulation information, the initial simulation information comprising inlet flow rate, pipe diameter, fluid property and wall distance simulation data; calculating a Reynolds value based on the inlet flow rate, the pipe diameter and the fluid property, determining the Reynolds value as a switching parameter of a laminar flow model / turbulent flow model, and performing a dynamic adjustment processing on a boundary layer grid based on the wall distance simulation data to generate a dynamically adjusted boundary; performing a non-circular cross-section processing based on the pipe diameter to generate a processed channel; performing a local resistance correction on the processed channel to obtain a corrected channel; establishing an energy consumption model containing pump efficiency, optimizing the energy consumption model based on the pipe diameter and pipe curvature to generate an optimized model; and generating the fluid simulation information based on the initial simulation information, the switching parameter, the dynamically adjusted boundary, the corrected channel and the optimized model.

[0010] Preferably, the chemical reaction simulation analysis based on the material parameters, generating chemical simulation information, comprises: performing a micro-reaction speed analysis based on the material parameters to generate a micro-reaction speed, performing a macro-mapping analysis on the micro-reaction speed to generate a macro-reaction speed; performing a catalytic activity gradient design on the micro-channel reactor based on the macro-reaction speed to generate gradient design information; determining an axial temperature gradient, and generating target product trend information based on the axial temperature gradient; and generating the chemical simulation information based on the macro-reaction speed, the gradient design information, a preset product concentration boundary and the target product trend information.

[0011] Preferably, the method further comprises: determining all reaction product in the micro-channel reactor before generating the macroscopic reaction speed; screening the reaction product based on a preset reaction concentration threshold to obtain screened reaction; performing weight analysis on the screened reaction based on a Bayesian network to generate a reaction weight; performing macroscopic mapping analysis on the microcosmic reaction speed based on the reaction weight to generate a macroscopic reaction speed.

[0012] Preferably, the heat transfer simulation analysis based on the operation parameters to generate heat transfer simulation information comprises: constructing a temperature field uniformity evaluation function based on the operation parameters, the temperature field uniformity evaluation function is characterized as: Wherein, N t is the number of time steps, N x is the number of spatial positions, ΔT allow is the allowable temperature difference, T(x, t) is the actual temperature value at spatial position x and time step t, T sct is the set target temperature value; the heating rod in the micro-channel reactor is distributedly divided to obtain a plurality of division units; the local heat flux of each division unit is determined based on a model predictive control algorithm, and the space-time distribution information of heating power is generated based on the local heat flux; the surface-to-surface radiation heat flow in the micro-channel reactor is calculated, and radiation-convection coupling correction information is generated based on the radiation heat flow; key energy consumption points in the micro-channel reactor are analyzed to generate a plurality of key energy consumption points; a counterflow heat exchanger is arranged at the outlet flow channel of the micro-channel reactor, and the heat recovery efficiency of the counterflow heat exchanger is determined; energy consumption optimization information is generated based on the key energy consumption points and the heat recovery efficiency; heat transfer simulation information is generated based on the temperature field uniformity evaluation function, the space-time distribution information, the radiation-convection coupling correction information and the energy consumption optimization information.

[0013] Preferably, the optimization of the adjusted simulation information based on a preset iterative optimization algorithm to generate optimized information comprises: constructing a Gaussian process regression model; generating a preset iterative optimization algorithm based on an expected hypervolume improvement algorithm and the Gaussian process regression model; iteratively optimizing the adjusted simulation information based on the preset iterative optimization algorithm to generate optimized information.

[0014] Correspondingly, the application further provides a micro-channel reactor design device based on multi-target Bayesian optimization, which is applied to the method according to the embodiment of the application, and the device comprises: a parameter generation unit configured to perform a parameter sampling operation to generate an initial parameter combination; a simulation unit configured to perform a physical field simulation operation on the micro-channel reactor based on the initial parameter combination to generate initial simulation information, wherein the initial simulation information comprises a plurality of simulation indexes; an adjustment unit configured to perform multi-target scoring calculation on the initial simulation information, and adjust the initial simulation information based on the scoring calculation result to generate adjusted simulation information; an optimization unit configured to optimize the adjusted simulation information based on a Bayesian algorithm to generate optimized information; and a design unit configured to configure and design the micro-channel reactor based on the optimized information.

[0015] In another aspect, the application further provides a computer readable storage medium having a computer program stored thereon, wherein the program is executed by a processor to implement the method provided in the embodiments of the application.

[0016] Through the technical solutions provided in the application, the application has at least the following technical effects:

[0017] By improving the design method of the existing micro-channel reactor, firstly, the actual physical characteristics are analyzed by simulation analysis, and corresponding simulation analysis and optimization are performed respectively, so that more accurate and reasonable simulation information is obtained; then the simulation information of the micro-channel reactor is further improved and optimized by combining the optimization algorithm, so that the optimization design scheme meeting the actual needs of the enterprise is obtained.

[0018] Other features and advantages of the embodiments of the application will be described in detail in the following specific implementation part. BRIEF DESCRIPTION OF DRAWINGS

[0019] The accompanying drawings are used to provide a further understanding of the embodiments of the application, and constitute a part of the specification, and are used to explain the embodiments of the application together with the following specific implementation, but do not constitute a limitation to the embodiments of the application. In the drawings:

[0020] Figure 1 is a specific implementation flowchart of the micro-channel reactor design method based on multi-target Bayesian optimization provided in the embodiments of the application;

[0021] Figure 2 is a structural schematic diagram of the micro-channel reactor design device based on multi-target Bayesian optimization provided in the embodiments of the application. DETAILED DESCRIPTION

[0022] The specific implementation of the embodiments of the present application is described in detail below with reference to the accompanying drawings. It should be understood that the specific implementation described herein is only used to illustrate and explain the embodiments of the present application, and is not used to limit the embodiments of the present application.

[0023] The terms "system" and "network" in the embodiments of the present application can be used interchangeably. "Multiple" means two or more, and in view of this, "multiple" can also be understood as "at least two" in the embodiments of the present application. "And / or" describes the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B can represent the following three cases: A exists alone, A and B exist together, and B exists alone. In addition, the character " / ", if not specially stated, generally represents that the associated objects before and after it are in an "or" relationship. In addition, it should be understood that in the description of the embodiments of the present application, "first", "second", and the like are only used for distinguishing purposes of description, and cannot be understood as indicating or implying relative importance, nor can it be understood as indicating or implying order.

[0024] Please refer to Figure 1 The embodiments of the present application provide a micro-channel reactor design method based on multi-objective Bayesian optimization, the method comprises:

[0025] S10: performing a parameter sampling operation to generate an initial parameter combination;

[0026] S20: performing a physical field simulation operation for a micro-channel reactor based on the initial parameter combination to generate initial simulation information, the initial simulation information comprising a plurality of simulation indexes;

[0027] S30: performing multi-objective scoring calculation on the initial simulation information, adjusting the initial simulation information based on the scoring calculation result to generate adjusted simulation information;

[0028] S40: optimizing the adjusted simulation information based on a Bayesian algorithm to generate optimized information;

[0029] S50: configuring and designing the micro-channel reactor based on the optimized information.

[0030] In one possible implementation, the parameter sampling operation is first performed, and an initial parameter combination is generated. In the embodiments of the present application, the performing of the parameter sampling operation to generate the initial parameter combination comprises: determining the geometric structure, operating conditions and material properties of the micro-channel reactor; determining the geometric parameters corresponding to the geometric structure; determining the operating parameters based on the operating conditions and the material parameters based on the material properties; processing the geometric parameters, the operating parameters and the material parameters based on a Latin hypercube sampling algorithm to generate the initial parameter combination.

[0031] Specifically, first, parameters of the micro-channel reactor are determined, for example, corresponding geometric parameters can be determined according to the geometric structure of the micro-channel reactor, including but not limited to channel width (0.1-2 mm), mixing unit spacing (1-10 mm), channel bending angle (30°-150°), and the like. Corresponding operation parameters can be determined according to the operation conditions thereof, including but not limited to flow rate (0.1-5 mL / min), temperature (25-200℃), and the like. Corresponding material parameters can be determined through material properties, including but not limited to catalyst type (Pt, Pd, Ni), carrier specific surface area (50-500 m 2 / g), and the like. In the process of designing, in order to obtain the best design effect, the above parameters often need to be arranged and combined and gradually screened out the best parameters, but the workload is very huge, which greatly reduces the work efficiency. In order to solve the technical problem, in the embodiment of the present application, the Latin hypercube sampling algorithm processes the geometric parameters, the operation parameters and the material parameters to generate an initial parameter combination. Specifically, the Latin hypercube sampling algorithm is used to optimally combine the above parameters, so as to greatly reduce the sample size on the basis of ensuring uniform coverage of the parameter space, and improve the work efficiency.

[0032] After determining the initial parameter combination, simulation analysis of the micro-channel reactor is started. In the embodiment of the present application, by analyzing the actual physical field of the micro-channel reactor in the reaction process, different physical fields are simulated and optimized to optimize the design and generate an optimal design scheme.

[0033] In the embodiment of the present application, the physical field simulation operation for the micro-channel reactor based on the initial parameter combination generates initial simulation information, including: performing fluid dynamics simulation analysis based on the geometric parameters to generate fluid simulation information; performing chemical reaction simulation analysis based on the material parameters to generate chemical simulation information; performing heat transfer simulation analysis based on the operation parameters to generate heat transfer simulation information; and generating initial simulation information based on the fluid simulation information, the chemical simulation information and the heat transfer simulation information.

[0034] In a possible implementation, the simulation analysis is performed on fluid dynamics, chemical reaction and heat transfer respectively. In the embodiment of the present application, the fluid dynamics simulation analysis based on the geometric parameters is performed to generate fluid simulation information, including: performing fluid dynamics simulation based on the geometric parameters to generate initial simulation information, the initial simulation information including inlet flow rate, pipe diameter, fluid properties and wall distance simulation data; calculating Reynolds number based on the inlet flow rate, the pipe diameter and the fluid properties, determining the Reynolds number as a switching parameter of laminar flow model / turbulent flow model, and performing dynamic adjustment processing on the boundary layer grid based on the wall distance simulation data to generate a dynamically adjusted boundary; performing non-circular cross-section processing based on the pipe diameter to generate a processed channel; performing local resistance correction on the processed channel to obtain a corrected channel; establishing an energy consumption model including pump efficiency, optimizing the energy consumption model based on the pipe diameter and pipe curvature to generate an optimized model; and generating fluid simulation information based on the initial simulation information, the switching parameter, the dynamically adjusted boundary, the corrected channel and the optimized model.

[0035] Specifically, in the fluid dynamics simulation analysis, simulation analysis is performed from three aspects of turbulent flow simulation analysis, pressure drop simulation analysis and energy consumption analysis. In the embodiment of the present application, simulation can be performed by COMSOL Multiphysics. First, fluid dynamics simulation is performed according to the geometric parameters to generate initial simulation information. Through the initial simulation information, parameters such as inlet flow rate, pipe diameter, fluid properties and wall distance simulation data of the micro-channel reactor can be obtained. The corresponding Reynolds number is calculated in real time according to the above parameters. For example, the Reynolds number is represented as wherein ρ represents the density of the fluid, u in represents the inlet flow rate, D represents the pipe diameter, and μ represents the dynamic viscosity of the fluid. By calculating the Reynolds number, when Re is less than a certain critical value (such as 2300), the fluid flow is in a laminar state, and the characteristics are that the fluid layers do not interfere with each other, and the flow lines are parallel. When Re exceeds the critical value, the fluid flow changes to turbulent flow, and at this time the motion of fluid particles becomes chaotic and there is strong mixing and vortex phenomenon. Based on this, automatic switching of the laminar flow model / turbulent flow model can be realized. For example, in COMSOL, the wall distance simulation data is dimensionless data y+, by obtaining the wall distance simulation data, when y+> 30 (high Reynolds number turbulent flow), the standard wall function can be automatically activated; when y+< 5 (low Reynolds number or near-wall flow), the scalable wall function is switched to the enhanced wall treatment (such as Scalable wall function), so as to realize the self-adaptation of the wall function and the grid.

[0036] Then, pressure drop simulation analysis is performed. Specifically, first, non-circular section processing is performed, and for rectangular / ring-shaped channels, the original circular pipe formula is optimized by analyzing the hydraulic diameter, so as to optimize the existing micro-channel reactor pipe; then, the processed pipe is locally corrected, and specifically, the local resistance coefficient K can be set in advance by the technician at the elbow, valve and other components, so that the total pressure drop is corrected as follows: Wherein, f represents the friction factor, L represents the pipe length, D h represents the hydraulic diameter, and ∑K represents the sum of the local resistance coefficients, ρ represents the density of the fluid, and u represents the average flow rate of the fluid. By considering the variation characteristics of the friction factor under different flow conditions, the correction method can effectively improve the accuracy and applicability of the pressure drop calculation in complex pipe systems.

[0037] Finally, the simulation analysis is performed from the power consumption aspect, and specifically, the energy consumption model including the pump efficiency is established, for example, the model is as follows: Wherein, P total represents the total energy consumption of the system, Δp represents the total pressure drop generated by the fluid flowing in the pipe, Q represents the volume flow rate of the fluid, η pump represents the efficiency of the pump, and P leakage represents the leakage energy consumption power of the system. In the embodiment of the present application, after the model is established, P total is minimized to preliminarily optimize the model under the condition of meeting the uniformity constraint of the flow rate distribution, and then the pipe diameter and the pipe curvature are used to further optimize the energy consumption model. Specifically, the pipe diameter / curvature is used as a design variable, and the energy consumption optimal flow channel (such as the resistance reduction design of the gradually expanding pipe) is automatically generated by the adjoint optimization algorithm, and the optimized model is generated. In this way, the fluid simulation information after optimization in the fluid simulation process is completed.

[0038] In the embodiment of the present application, the chemical reaction simulation analysis based on the material parameters is performed to generate chemical simulation information, including: performing micro-reaction speed analysis based on the material parameters to generate a micro-reaction speed, performing macro-mapping analysis on the micro-reaction speed to generate a macro-reaction speed; performing catalytic activity gradient design on the micro-channel reactor based on the macro-reaction speed to generate gradient design information; determining an axial temperature gradient, and generating target product trend information based on the axial temperature gradient; and generating chemical simulation information based on the macro-reaction speed, the gradient design information, a preset product concentration boundary and the target product trend information.

[0039] In a possible implementation, first, the micro-reaction speed is analyzed according to the material and time, specifically, first, the energy of the molecule adsorbed on the surface of the catalyst is calculated based on the density functional theory, then the micro-reaction speed is derived based on the Eley-Rideal mechanism, and then the micro-reaction speed is connected to the macro-reaction speed constant according to the transition state theory to generate the macro-reaction speed. Through such a multi-scale dynamic coupling technology, the chemical reaction process can be accurately simulated, laying a solid foundation for subsequent processing of complex reaction systems.

[0040] At this time, further, in order to further improve the conversion rate of the reaction and increase the product amount, the boundary conditions of the micro-channel reactor are further optimized. Specifically, in the simulation process, the catalytic activity of each position in the micro-channel reactor is set, and the catalytic activity changes with the gradient change of the position, thereby generating corresponding gradient design information. On the basis of the gradient design information, the chemical reaction can be converted along the length direction of the micro-channel reactor, and the conversion rate can be relatively uniform, and there is no place where the reaction is too violent and there is no place where the reaction is too slow. Further, for a reversible reaction, like a back-and-forth switch, the forward direction generates products, and the reverse direction changes back to raw materials. For this, by setting a preset equilibrium concentration value, for example, the preset equilibrium concentration value is 80%, when the outlet product concentration reaches 80%, the reverse gas flow is provided through the micro-channel reactor to blow away the product, reduce the product concentration, and inhibit the reverse reaction to allow the forward reaction to continue to generate more products.

[0041] Finally, temperature simulation analysis is performed. Specifically, first, the axial temperature gradient (for example, the inlet is 500K and the outlet is 450K) is determined, and corresponding target product trend information is generated, so that the different reactions have different sensitivities to temperature, the main reaction runs fast at high temperature, and the side reaction slows down at low temperature, thereby improving the selectivity of the target product. Further, for a reaction system with a magnetic catalyst, an electromagnetic module is also configured in the micro-channel reactor, and the electromagnetic module generates a Lorentz force to generate a stirring force in the micro-channel reactor, so that the fluid moves more violently, reduces the mass transfer boundary layer thickness, and makes the reactants and catalysts contact faster, thereby making the reaction more efficient. The chemical simulation information is generated according to the macro-reaction speed, the gradient design information, the preset product concentration boundary, and the target product trend information.

[0042] In the embodiment of the present application, the method further comprises: determining all reaction products in the micro-channel reactor before generating the macro-reaction speed; screening the reaction product based on a preset reaction concentration threshold to obtain a screened reaction; performing weight analysis on the screened reaction based on a Bayesian network to generate a reaction weight; and performing macro-mapping analysis on the micro-reaction speed based on the reaction weight to generate a macro-reaction speed.

[0043] Specifically, due to the existence of a large number of by-products in the chemical reaction process, and the yield of some by-products is small, a large amount of computing resources will be wasted. In the embodiment of the present application, after the above macroscopic reaction rate is preliminarily generated, the reaction in the micro-channel reactor is further analyzed, when there are multiple reactions, a concentration threshold is set, for example, the concentration threshold is 1e-5 mol / m 3 If the concentration of a product is lower than the concentration threshold, the side reactions are ignored. At the same time, the importance of each reaction path is analyzed based on the Bayesian network, so as to allocate reasonable weights and corresponding computing resources to each reaction path, and avoid wasting computing power. The effective processing of side reactions and the reasonable allocation of computing resources further improve the practicability and accuracy of the multi-scale kinetic coupling technology in complex reaction systems.

[0044] Finally, the heat transfer simulation analysis is performed according to the operation parameters. In the embodiment of the present application, the heat transfer simulation analysis is performed based on the operation parameters to generate heat transfer simulation information, including: constructing a temperature field uniformity evaluation function based on the operation parameters, the temperature field uniformity evaluation function is characterized by: Wherein, N t is the number of time steps, N x is the number of spatial positions, ΔT allow is the allowable temperature difference, T(x, t) is the actual temperature value at spatial position x and time step t, T sct is the set target temperature value; the heating rod in the micro-channel reactor is distributedly divided to obtain a plurality of division units; the local heat flux of each division unit is determined based on the model predictive control algorithm, and the time-space distribution information of the heating power is generated based on the local heat flux; the surface-to-surface radiation heat flow in the micro-channel reactor is calculated, and the radiation-convection coupling correction information is generated based on the radiation heat flow; the key energy consumption points of the micro-channel reactor are analyzed to generate a plurality of key energy consumption points; the counterflow heat exchanger is arranged at the outlet flow channel of the micro-channel reactor, and the heat recovery efficiency of the counterflow heat exchanger is determined; the energy consumption optimization information is generated based on the key energy consumption points and the heat recovery efficiency; the heat transfer simulation information is generated based on the temperature field uniformity evaluation function, the time-space distribution information, the radiation-convection coupling correction information and the energy consumption optimization information.

[0045] In one possible implementation, a temperature field uniformity evaluation function containing time-space dimensions is first constructed according to the operation parameters, and the temperature field uniformity evaluation function is characterized by: Wherein, N t is the number of time steps, N x is the number of spatial positions, ΔT allowTo allow temperature difference, T(x, t) is the actual temperature value at spatial position x, time step t, T sct is the set target temperature value. On this basis, in order to further improve the uniformity of the temperature in the micro-channel reactor, a PCM (phase change material) layer is embedded in the reactor wall, and latent heat storage is simulated by a porous medium heat transfer model, so as to utilize the isothermal characteristics of the phase change process to suppress the local overheating / undercooling phenomenon.

[0046] Then the heating rod in the micro-channel reactor is distributedly divided, for example, divided into 10+ independent division units, and the corresponding local heat flux can be set through a pre-set grid group. At this time, the heat flux of each division unit is analyzed and optimized according to the model predictive control algorithm, so as to realize the spatio-temporal optimal distribution of the heating power. On the other hand, in order to further improve the accuracy of the heat analysis in the micro-channel reactor, surface-to-surface radiation is enabled in the high-temperature reactor, and specifically, the radiation heat flow is calculated through a grey body approximation algorithm to automatically identify the radiation heat exchange between the high-temperature wall and the environment, and correct the error of the traditional only convection model.

[0047] Finally, the micro-channel reactor is simplified as a thermal resistance equivalent circuit, and the thermal resistance of each component is defined, and on this basis, a plurality of key energy consumption points in the micro-channel reactor are analyzed to identify the key energy consumption nodes (such as the thickness of the insulation layer, the flow rate of the fluid). At the same time, a counter-flow heat exchanger is arranged at the outlet flow channel, the reaction product waste heat is used to preheat the feed, and the recovery efficiency is calculated through multi-physical field coupling calculation, for example, the calculation rule is: Wherein, η recovery represents the waste heat recovery efficiency, m represents the mass flow rate of the fluid, c p represents the constant-pressure specific heat capacity of the fluid, T out,in represents the temperature of the reaction product when entering the heat exchanger, T out,out represents the temperature of the reaction product when leaving the heat exchanger, T in,hot represents the temperature of the feed when leaving the heat exchanger, T in,cold represents the temperature of the feed when entering the heat exchanger. Through automatic optimization of the heat exchange area and the flow direction, the energy consumption can be reduced by more than 30%. The energy consumption optimization information is generated according to the above key energy consumption points and the heat recovery efficiency, and finally the heat transfer simulation information is generated.

[0048] Finally, the initial simulation information is generated by combining the simulation information of the above aspects, and then the initial simulation information is further designed and optimized. Specifically, multi-objective scoring calculation is performed on the initial simulation information, the simulation indexes are adjusted based on the scoring calculation result, and adjusted simulation information is generated. For example, first, define the parameter range, such as defining the maximum conversion rate (0-100%), the minimum energy consumption (kWh / kg), and the minimum manufacturing cost (yuan / unit), and then use the Hypervolume index to evaluate the quality of the Pareto frontier to avoid the deviation caused by manual weight setting. On this basis, multi-objective scoring calculation is performed on the initial simulation information, and each simulation index is adjusted to generate adjusted simulation information. At this time, the adjusted simulation information is further optimized according to the Bayesian algorithm.

[0049] In the embodiment of the present application, the preset iterative optimization algorithm is used to optimize the adjusted simulation information to generate optimized information, including: constructing a Gaussian process regression model; generating a preset iterative optimization algorithm based on an expected hypervolume improvement algorithm and the Gaussian process regression model; and iteratively optimizing the adjusted simulation information based on the preset iterative optimization algorithm to generate optimized information.

[0050] In a possible implementation, a Gaussian process regression model is constructed with Matérn 5 / 2 as the kernel function, an expected hypervolume improvement (EHVI) is used as the acquisition function to construct a preset iterative optimization algorithm, and then the preset iterative optimization algorithm is used to iteratively optimize the adjusted simulation information. Specifically, the Pareto optimal solution set is iteratively updated, and the termination condition is, for example, that the hypervolume change rate is <1% or the number of iterations is ≥100 times, to generate optimized information. Finally, the micro-channel reactor is configured and designed according to the optimized information to obtain a micro-channel reactor with the best configuration effect.

[0051] In the embodiment of the present application, by improving the design method of the existing micro-channel reactor, by respectively simulating and optimizing the actual physical characteristics, and by combining a better optimization algorithm, the convergence speed is improved by 40% and the simulation times are reduced by 60% compared with the traditional NSGA-II algorithm in terms of efficiency. Meanwhile, the Pareto optimal solution set is output to support users to select the best balanced scheme (such as "high conversion rate-medium cost" or "low energy consumption-low cost") according to the demand. In addition, by parameter dimension reduction and noise modeling, the high-dimensional parameter space and simulation error in actual engineering are adapted, the industrial applicability is greatly improved, and the actual needs of enterprises are met.

[0052] See Figure 2Based on the same inventive concept, the embodiment of the present application provides a micro-channel reactor design device based on multi-objective Bayesian optimization, which is applied to the method according to the embodiment of the present application, and characterized in that the device comprises: a parameter generation unit configured to perform a parameter sampling operation to generate an initial parameter combination; a simulation unit configured to perform a physical field simulation operation for the micro-channel reactor based on the initial parameter combination to generate initial simulation information, wherein the initial simulation information comprises a plurality of simulation indexes; an adjustment unit configured to perform multi-objective scoring calculation on the initial simulation information, and adjust the initial simulation information based on the scoring calculation result to generate adjusted simulation information; an optimization unit configured to optimize the adjusted simulation information based on a Bayesian algorithm to generate optimized information; and a design unit configured to configure and design the micro-channel reactor based on the optimized information.

[0053] Further, the embodiment of the present application further provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to implement the method according to the embodiment of the present application.

[0054] The above describes optional embodiments of the embodiment of the present application in detail with reference to the drawings, but the embodiment of the present application is not limited to the specific details in the above embodiments, and various simple modifications can be made to the technical solutions of the embodiment of the present application within the technical concept of the embodiment of the present application, and these simple modifications all belong to the protection scope of the embodiment of the present application.

[0055] In addition, it should be noted that each specific technical feature described in the above specific embodiments can be combined in any appropriate manner without contradiction. In order to avoid unnecessary repetition, the embodiment of the present application will not further describe various possible combinations.

[0056] Those skilled in the art can understand that all or part of the steps of the methods described in the above embodiments can be completed by programs instructing related hardware. The programs are stored in a storage medium, and include a plurality of instructions for causing a single-chip microcomputer, a chip or a processor to execute all or part of the steps of the methods described in the embodiments of the present application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0057] In addition, various different embodiments of the embodiment of the present application can also be combined in any appropriate manner, as long as they do not contradict the technical concept of the embodiment of the present application, and they should be considered as disclosed by the embodiment of the present application.

Claims

1. A microchannel reactor design method based on multi-objective Bayesian optimization, characterized in that, The method includes: Perform parameter sampling operations to generate initial parameter combinations; Based on the initial parameter combination, a physical field simulation operation is performed for the microchannel reactor to generate initial simulation information, which includes multiple simulation indicators. The initial simulation information is subjected to multi-objective scoring calculation, and the simulation indicators are adjusted based on the scoring calculation results to generate adjusted simulation information; The adjusted simulation information is optimized based on a preset iterative optimization algorithm to generate optimized information; The microchannel reactor is configured and designed based on the optimized information. The parameter sampling operation generates an initial parameter combination, including: Determine the geometry, operating conditions, and material properties of the microchannel reactor; Determine the geometric parameters corresponding to the geometric structure; The operating parameters are determined based on the operating conditions, and the material parameters are determined based on the material properties. The geometric parameters, operational parameters, and material parameters are processed using the Latin hypercube sampling algorithm to generate an initial parameter combination; The physical field simulation operation for the microchannel reactor based on the initial parameter combination is performed to generate initial simulation information, including: Based on the geometric parameters, perform fluid dynamics simulation analysis to generate fluid simulation information; Based on the material parameters, perform chemical reaction simulation analysis to generate chemical simulation information; Perform heat transfer simulation analysis based on the aforementioned operating parameters to generate heat transfer simulation information; Initial simulation information is generated based on the fluid simulation information, the chemical simulation information, and the heat transfer simulation information; The process of performing chemical reaction simulation analysis based on the material parameters to generate chemical simulation information includes: Microscopic reaction rate analysis is performed based on the material parameters to generate microscopic reaction rates. Macroscopic mapping analysis is then performed on the microscopic reaction rates to generate macroscopic reaction rates. Based on the macroscopic reaction rate, a catalytic activity gradient is designed for the microchannel reactor to generate gradient design information. Determine the axial temperature gradient, and generate target product trend information based on the axial temperature gradient; Chemical simulation information is generated based on the macroscopic reaction rate, the gradient design information, the preset product concentration boundary, and the target product trend information. The method further includes: Before generating the macroscopic reaction rate, all reactant products in the microchannel reactor are determined; The reactant products are screened based on a preset reactant concentration threshold to obtain the screened reactants; Weight analysis of the screened reactants is performed based on a Bayesian network to generate reaction weights; Based on the reaction weights, a macroscopic mapping analysis is performed on the microscopic reaction rates to generate macroscopic reaction rates.

2. The method according to claim 1, characterized in that, The process of performing fluid dynamics simulation analysis based on the geometric parameters to generate fluid simulation information includes: Based on the geometric parameters, perform fluid dynamics simulation to generate initial simulation information, which includes simulation data of inlet velocity, pipe diameter, fluid properties and wall distance. The Reynolds value is calculated based on the inlet velocity, the pipe diameter, and the fluid properties. The Reynolds value is then used as the switching parameter between the laminar flow model and the turbulent flow model. Based on the wall distance simulation data, the boundary layer mesh is dynamically adjusted to generate a dynamically adjusted boundary. Based on the pipe diameter, a non-circular cross-section is processed to generate a processed channel; Perform local resistance correction on the processed channel to obtain the corrected channel; An energy consumption model incorporating pump efficiency is established, and the model is optimized based on the pipe diameter and pipe curvature to generate an optimized model. Fluid simulation information is generated based on the initial simulation information, the switching parameters, the dynamically adjusted boundary, the corrected channel, and the optimized model.

3. The method according to claim 1, characterized in that, The step of performing heat transfer simulation analysis based on the operating parameters and generating heat transfer simulation information includes: A temperature field uniformity evaluation function is constructed based on the aforementioned operating parameters, and the temperature field uniformity evaluation function is characterized as follows: in, For time steps, The number of spatial locations, To allow for temperature differences, This represents the actual temperature value at spatial location x and time step t. The set target temperature value; The heating rods in the microchannel reactor are distributed and divided to obtain multiple partitioning units; The local heat flux of each partition unit is determined based on the model predictive control algorithm, and the spatiotemporal distribution information of heating power is generated based on the local heat flux. Calculate the surface-to-surface radiative heat flux in the microchannel reactor, and generate radiation-convection coupling correction information based on the radiative heat flux; A key energy consumption point analysis was performed on the microchannel reactor, generating multiple key energy consumption points; A countercurrent heat exchanger is installed in the outlet channel of the microchannel reactor, and the heat recovery efficiency of the countercurrent heat exchanger is determined. Energy consumption optimization information is generated based on the key energy consumption points and the heat recovery efficiency. Heat transfer simulation information is generated based on the temperature field uniformity evaluation function, the spatiotemporal allocation information, the radiation-convection coupling correction information, and the energy consumption optimization information.

4. The method according to claim 1, characterized in that, The optimization of the adjusted simulation information based on a preset iterative optimization algorithm to generate optimized information includes: Construct a Gaussian process regression model; A preset iterative optimization algorithm is generated based on the expected hypervolume improvement algorithm and the Gaussian process regression model; The adjusted simulation information is iteratively optimized based on the preset iterative optimization algorithm to generate optimized information.

5. A microchannel reactor design device based on multi-objective Bayesian optimization, characterized in that, The method according to any one of claims 1-4, characterized in that the apparatus comprises: The parameter generation unit is used to perform parameter sampling operations and generate initial parameter combinations. The simulation unit is used to perform physical field simulation operations for the microchannel reactor based on the initial parameter combination, and generate initial simulation information, which includes multiple simulation indicators. An adjustment unit is used to perform multi-objective scoring calculation on the initial simulation information, adjust the initial simulation information based on the scoring calculation results, and generate adjusted simulation information. An optimization unit is used to optimize the adjusted simulation information based on a Bayesian algorithm to generate optimized information. The design unit is used to configure and design the microchannel reactor based on the optimized information. The parameter generation unit is specifically used for: Determine the geometry, operating conditions, and material properties of the microchannel reactor; Determine the geometric parameters corresponding to the geometric structure; The operating parameters are determined based on the operating conditions, and the material parameters are determined based on the material properties. The geometric parameters, operational parameters, and material parameters are processed using the Latin hypercube sampling algorithm to generate an initial parameter combination; The simulation unit is specifically used for: Based on the geometric parameters, perform fluid dynamics simulation analysis to generate fluid simulation information; Based on the material parameters, perform chemical reaction simulation analysis to generate chemical simulation information; Perform heat transfer simulation analysis based on the aforementioned operating parameters to generate heat transfer simulation information; Initial simulation information is generated based on the fluid simulation information, the chemical simulation information, and the heat transfer simulation information; The process of performing chemical reaction simulation analysis based on the material parameters to generate chemical simulation information includes: Microscopic reaction rate analysis is performed based on the material parameters to generate microscopic reaction rates. Macroscopic mapping analysis is then performed on the microscopic reaction rates to generate macroscopic reaction rates. Based on the macroscopic reaction rate, a catalytic activity gradient is designed for the microchannel reactor to generate gradient design information. Determine the axial temperature gradient, and generate target product trend information based on the axial temperature gradient; Chemical simulation information is generated based on the macroscopic reaction rate, the gradient design information, the preset product concentration boundary, and the target product trend information. The method further includes: Before generating the macroscopic reaction rate, all reactant products in the microchannel reactor are determined; The reactant products are screened based on a preset reactant concentration threshold to obtain the screened reactants; Weight analysis of the screened reactants is performed based on a Bayesian network to generate reaction weights; Based on the reaction weights, a macroscopic mapping analysis is performed on the microscopic reaction rates to generate macroscopic reaction rates.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the method described in any one of claims 1-4.

Images