Parameter optimization method for preparation of hollow silica microspheres based on flow field simulation

Abstract

The present invention relates to the technical field of hollow silica microsphere preparation, and discloses a method for optimizing hollow silica microsphere preparation parameters based on flow field simulation. The method first plans the optimization process based on target physical property parameters and preparation conditions, specifies the reactor flow field distribution simulation, obtains microsphere morphology data for different parameter combinations, and determines parameter combination types such as high temperature, specific physical properties, and general physical properties. The method then generates a parameter optimization process set and execution sequence, and generates a list of simulation inputs and experimental configurations for each parameter combination, which are linked to form a complete solution. The method also includes determining parameter optimization boundaries, extracting optimized data subsets, and implementing targeted adjustments, corrections, and stabilization strategies for high-efficiency, low-efficiency, and critical parameter combinations. The method also includes a real-time data update process, experimental rules (dynamic adjustment of the number of experiments, triggering supplementary experiments, and equipment calibration compensation), and database management and analysis. The method improves optimization efficiency and preparation stability and is suitable for industrial production.

Description

Technical Field

[0001] The invention relates to the technical field of hollow silica microsphere preparation, in particular to a method for optimizing preparation parameters of hollow silica microspheres based on flow field simulation. Background Art

[0002] Hollow silica microspheres, due to their unique hollow structure, excellent chemical stability, large specific surface area, and superior optical properties, offer broad application prospects in a wide range of fields, including catalytic supports, biomedicine, nanomaterials, electronic packaging, and coating additives. For example, in catalysis, their hollow structure can provide more active sites and storage space for reactants; in biomedicine, they can serve as drug carriers for controlled drug release.

[0003] Currently, traditional methods for optimizing hollow silica microsphere preparation parameters rely primarily on empirical trial-and-error. For example, in the spray-drying method for preparing hollow silica microspheres disclosed in patent CN103288093A, the traditional method requires repeated preparation experiments, observing the microsphere morphology and physical properties under different parameter combinations, and gradually adjusting the parameters to obtain the optimal solution. This approach has significant limitations: on the one hand, the large number of experiments and long cycles consume significant manpower, material resources, and time costs; on the other hand, due to the lack of in-depth understanding of the inherent physical mechanisms such as flow field distribution during the preparation process, parameter adjustments are often unreliable, making it difficult to achieve efficient parameter optimization. This results in unstable and inconsistent microsphere quality, making it difficult to meet the demand for high-quality, highly stable products for industrial large-scale production.

[0004] With the advancement of computer technology and numerical simulation methods, flow field simulation has become increasingly widely used in fields such as chemical engineering and materials preparation. Flow field simulation can intuitively visualize the flow characteristics, mass transfer, and energy transfer processes of fluids within a reactor, providing a theoretical basis for optimizing the preparation process. However, research integrating flow field simulation with parameter optimization for hollow silica microsphere preparation is still in its infancy, lacking a systematic and comprehensive parameter optimization method. For example, existing research on silica microsphere preparation (such as the sol-gel method and emulsion polymerization method mentioned in the "Preparation of Silica Microspheres" courseware) lacks the ability to effectively combine flow field simulation with parameter optimization. The key technical challenge currently facing this field is how to organically integrate flow field simulation results with preparation experiments, construct a scientific and efficient parameter optimization process, and achieve precise control of hollow silica microsphere preparation parameters.

[0005] Furthermore, existing technologies lack scientific and flexible approaches to parameter combination classification and optimization strategies. Different parameter combinations have varying degrees of influence on microsphere properties. Traditional methods often employ a unified optimization strategy, failing to tailor optimization to different parameter combinations, resulting in low optimization efficiency. Furthermore, during the preparation process, the utilization of real-time data and the dynamic adjustment mechanisms for parameter optimization processes are inadequate, making it difficult to adapt to the various uncertainties involved in the preparation process, further impacting the effectiveness of parameter optimization and the stability of microsphere preparation. Summary of the Invention

[0006] The purpose of the present invention is to provide a method for optimizing the preparation parameters of hollow silica microspheres based on flow field simulation to solve the problems raised in the above background technology.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for optimizing the preparation parameters of hollow silica microspheres based on flow field simulation, the method comprising:

[0008] Planning the parameter optimization process based on the target physical properties and preparation conditions of the hollow silica microspheres, and determining the flow field simulation and preparation experiment content required for optimization. The flow field simulation content includes the simulation of the flow field distribution in the reactor, and the preparation experiment content includes obtaining microsphere morphology data under different parameter combinations;

[0009] Determining different parameter combination types according to the target physical properties and preparation conditions of the hollow silica microspheres;

[0010] Generate a parameter optimization process set according to the different parameter combination types, flow field simulation contents and preparation experiment contents, and determine the execution order of the parameter optimization process set based on a dynamic sorting principle;

[0011] Generating a simulation input list for each parameter combination under each parameter combination according to a preset optimization cycle, the different parameter combination types, and the preparation process index requirements, and generating an experimental configuration list for each parameter combination under each parameter combination according to preset experimental rules, the different parameter combination types, and the preparation process index requirements;

[0012] Based on the execution order of the parameter optimization process set, the simulation input list, the experimental configuration list and the preparation optimization content of each parameter combination are associated to generate a complete solution for the preparation parameter optimization of the hollow silica microspheres.

[0013] Preferably, the parameter optimization process is planned based on the target physical properties and preparation conditions of the hollow silica microspheres, and the flow field simulation and preparation experiment contents required for optimization are determined, including:

[0014] For a current parameter combination, determining a parameter optimization boundary of the current parameter combination based on historical morphology data of the current parameter combination, historical morphology data of adjacent parameter combinations, and a preset parameter division threshold;

[0015] Extracting an optimized data subset of the current parameter combination according to the parameter optimization boundary and the preset parameter window;

[0016] The contents of the flow field simulation and preparation experiment related data in the optimization data subset are determined as the preparation optimization contents of the current parameter combination.

[0017] Preferably, different parameter combination types are determined according to the target physical properties and preparation conditions of the hollow silica microspheres, including:

[0018] For a current parameter combination, if the reaction temperature of the current parameter combination is greater than a preset temperature threshold, determining the current parameter combination as a high-temperature parameter combination;

[0019] When the reaction temperature of the current parameter combination is not greater than the preset temperature threshold, classifying the particle size distribution and porosity data of the current parameter combination based on a first preset optimization algorithm to obtain data of various physical property categories, and determining core features of the data of various physical property categories;

[0020] Determining an optimization potential level of the current parameter combination according to the preparation conditions of the current parameter combination;

[0021] When the matching degree between the core features of the physical property category data and the preset physical property category is greater than a preset matching threshold, determining the current parameter combination as a specific physical property parameter combination;

[0022] When the matching degree is not greater than the preset matching threshold, the current parameter combination is determined to be a common physical property parameter combination.

[0023] Preferably, the preparation optimization content also includes reactor structural parameters and raw material ratio data, and the different parameter combination types include high-efficiency parameter combination, low-efficiency parameter combination and critical parameter combination. According to the different parameter combination types, flow field simulation content and preparation experiment content, a parameter optimization process set is generated, including:

[0024] For the current parameter combination, the raw material ratio data of the current parameter combination is clustered based on the second preset optimization algorithm to obtain data of each ratio category; for the current ratio category data, the current ratio category data is screened based on the influence weight and concentration gradient of each raw material in the current ratio category data to obtain a key parameter subset of the current ratio category data, and the influence label of the current ratio category data is determined according to the correlation characteristics of the key parameter subset and the mean value of the preparation process indicators.

[0025] Preferably, after determining whether the parameter combination type of the current parameter combination is an efficient parameter combination, the method further includes:

[0026] If the parameter combination type of the current parameter combination is an efficient parameter combination, then matching the impact labels of the respective ratio category data with the preparation process indicators to generate a directional adjustment strategy;

[0027] If the parameter combination type of the current parameter combination is an inefficient parameter combination, generating a parameter correction strategy based on a preset correction rule and a physical property improvement target;

[0028] If the parameter combination type of the current parameter combination is a critical parameter combination, a process stabilization strategy is generated according to historical data fluctuations and process stabilization rules.

[0029] Preferably, after generating the parameter optimization process set, the method further includes:

[0030] Updating the preparation optimization content of each parameter combination based on real-time preparation data;

[0031] Recalculate the parameter combination type of each parameter combination according to the updated preparation optimization content;

[0032] The execution order of the parameter optimization process set is dynamically adjusted according to the recalculated parameter combination type.

[0033] Preferably, the preset experimental rules include:

[0034] Dynamically adjust the number of repetitions and sample size of the preparation experiment based on the confidence interval of historical experimental data;

[0035] When the deviation between the measured morphology data and the flow field simulation results exceeds the preset tolerance threshold, the supplementary experiment module is automatically triggered.

[0036] Preferably, the preset experimental rules also include:

[0037] Establish a calibration and compensation mechanism for detection equipment, and calibrate the detection equipment for error compensation based on the benchmark values ​​of the standard particle size analyzer and electronic balance before each experiment.

[0038] Preferably, after generating the complete scheme for optimizing the preparation parameters of the hollow silica microspheres, the method further comprises:

[0039] Establishing a preparation parameter optimization database, storing the simulation input list, experimental configuration list, preparation optimization content, and parameter combination type data in the complete scheme in the database;

[0040] Regularly clean and correct abnormal data in the database and eliminate invalid data records;

[0041] Based on this database, the preparation parameter optimization results are periodically compared and analyzed to evaluate the evolution of the parameter optimization effect.

[0042] Preferably, the second preset optimization algorithm is a particle swarm optimization algorithm, and the specific steps of clustering the raw material ratio data of the current parameter combination include:

[0043] Initialize the particle swarm position and velocity vector, calculate the fitness value of each particle and update the individual optimal solution;

[0044] The particle flight direction is adjusted according to the swarm optimal solution, and the final optimized solution set is determined as the data for each ratio category by presetting the number of iterations or fitness convergence threshold.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] In terms of parameter optimization planning, a comprehensive process was developed based on the target physical properties and preparation conditions of the hollow silica microspheres, clarifying the specific content of flow field simulation and preparation experiments. By analyzing historical morphological data for the current parameter combination, historical morphological data for adjacent parameter combinations, and preset parameter partitioning thresholds, the parameter optimization boundary was determined. An optimized data subset was extracted from this data, and the flow field simulation and preparation experiment data were confined to this subset. This targeted planning focused on the key parameter range, avoiding blind exploration, reducing unnecessary simulations and experiments, significantly shortening the parameter optimization cycle, and lowering R&D costs.

[0047] In determining parameter combination types, the system comprehensively considers factors such as reaction temperature, particle size distribution, porosity data, preparation conditions, and compatibility with pre-set physical property categories. Parameter combinations are categorized into high-temperature parameter combinations, specific physical property parameter combinations, and general physical property parameter combinations. Furthermore, the system introduces classifications for efficient, inefficient, and critical parameter combinations. A second pre-set optimization algorithm (such as particle swarm optimization) is then used to cluster and filter raw material ratio data to identify key parameter subsets and influencing labels. This refined classification approach can deeply reveal the characteristics and optimization potential of different parameter combinations, providing a solid foundation for the subsequent development of personalized optimization strategies.

[0048] In terms of optimization strategy formulation and execution, differentiated optimization strategies are adopted for different types of parameter combinations. For efficient parameter combinations, by matching the impact labels of the ratio category data with the preparation process indicators, a targeted adjustment strategy is generated to achieve precise fine-tuning of key parameters and further improve the performance of the microspheres; for inefficient parameter combinations, a parameter correction strategy is generated based on preset correction rules and physical property improvement goals to specifically address the problem of unreasonable parameter settings and quickly improve the quality of microspheres; for critical parameter combinations, a process stability strategy is generated based on historical data fluctuations and process stability rules to ensure the stability of the preparation process and avoid large fluctuations in the physical properties of the microspheres. In addition, after generating the parameter optimization process set, the preparation optimization content is dynamically updated based on real-time preparation data, the parameter combination type is recalculated, and the process execution order is adjusted, so that the optimization process can respond to changes in the preparation process in real time, improving the flexibility and adaptability of parameter optimization.

[0049] In terms of experimental rule design, the preset experimental rules fully consider the reliability and accuracy of the experiment. Based on the confidence interval of historical experimental data, the number of repetitions and sample size of the preparation experiment are dynamically adjusted to ensure the statistical significance of the experimental results. When the deviation between the measured morphology data and the flow field simulation results exceeds the preset tolerance threshold, the supplementary experiment module is automatically triggered to promptly detect and correct the discrepancy between the simulation and the experiment, ensuring the credibility of the optimization results. At the same time, a calibration and compensation mechanism for testing equipment is established, and error compensation calibration is performed on the testing equipment before each experiment, effectively reducing the impact of equipment errors on experimental results and improving the accuracy and reliability of the data.

[0050] In terms of data management and optimization effect evaluation, a preparation parameter optimization database was established to store, clean, and calibrate all types of data in the complete solution, and periodically compare and analyze the optimization results. This not only helps accumulate preparation experience and data, providing a reference for subsequent parameter optimization, but also allows for the evaluation of the evolution of parameter optimization effects, timely identification of problems in the optimization process, and providing a basis for continuous improvement of the preparation process, thereby promoting the continuous development and improvement of hollow silica microsphere preparation technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 This is a working principle diagram of the hollow silica microsphere preparation parameter optimization method based on flow field simulation according to the present invention;

[0052] Figure 2 A working principle diagram for determining preparation optimization content based on historical morphological data and parameter division thresholds;

[0053] Figure 3 Schematic diagram of the working principle for the classification of parameter combination types of hollow silica microspheres;

[0054] Figure 4Working principle diagram generated for the preparation strategy corresponding to different parameter combination types. DETAILED DESCRIPTION

[0055] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0056] See also Figures 1-4 The present invention relates to a method for optimizing the preparation parameters of hollow silica microspheres based on flow field simulation, and the specific implementation steps are as follows:

[0057] Based on the target physical properties of hollow silica microspheres (such as particle size distribution and porosity) and preparation conditions (such as reaction temperature, raw material ratio, and reactor structure), a systematic plan for parameter optimization was developed to clearly define the required flow field simulation and preparation experiments. The flow field simulation involved simulating the flow field distribution within the reactor using computational fluid dynamics (CFD) and other simulation methods, including the distribution of parameters such as flow velocity, pressure, and turbulence intensity. The preparation experiments involved obtaining microsphere morphology data under different parameter combinations through actual experiments, such as sphere roundness, surface roughness, and hollow structure integrity.

[0058] According to the target physical properties and preparation conditions of hollow silica microspheres, different parameter combination types are divided to provide a classification basis for subsequent optimization processes.

[0059] By combining different parameter combination types, flow field simulation content, and preparation experiment content, a parameter optimization process set is generated. The execution order of each process is determined based on the dynamic sorting principle (such as prioritizing parameter combinations that have a significant impact on the target physical properties) to improve optimization efficiency.

[0060] According to the preset optimization cycle (such as the optimization frequency set by day, week or batch), different parameter combination types and preparation process index requirements (such as particle size uniformity index, porosity range, etc.), a corresponding simulation input list (such as the initial parameter value set for flow field simulation) is generated under each parameter combination; at the same time, according to the preset experimental rules (such as experimental repeatability requirements, sample size standards, etc.), different parameter combination types and preparation process index requirements, an experimental configuration list for each parameter combination is generated (such as experimental equipment parameter settings, raw material feeding order, etc.).

[0061] Following the execution order of the parameter optimization process set, the simulation input list, experimental configuration list, and preparation optimization content (such as the adjustment direction and goals for a specific parameter combination) for each parameter combination are linked to form a complete solution for optimizing the preparation parameters of hollow silica microspheres, achieving coordinated optimization of simulation and experiment.

[0062] The present invention will be further described below in conjunction with Examples 1 to 5:

[0063] Example 1:

[0064] In the parameter optimization process planning phase, the implementation method for determining the parameter optimization boundary and the optimization data subset for the current parameter combination is as follows:

[0065] Obtain historical morphological data for the current parameter combination. This data is derived from the test results of the microsphere physical properties in previous preparation experiments, including but not limited to statistical values ​​of particle size distribution (such as average particle size and particle size standard deviation), porosity measurements, sphere roundness coefficient, surface roughness parameters, and hollow structure integrity indicators (such as the ratio of hollow cavity diameter to microsphere outer diameter). Simultaneously, obtain historical morphological data for adjacent parameter combinations. Adjacent parameter combinations are defined based on the range of differences in key parameters in the preparation conditions. For example, with the reaction temperature of the current parameter combination as the center, a temperature fluctuation range of ±5°C is set as the adjacent temperature parameter combination; with the silicon source concentration in the raw material ratio as the benchmark, a concentration fluctuation range of ±10% is set as the adjacent concentration parameter combination; and for reactor structural parameters (such as agitator speed), a speed fluctuation range of ±10% is set as the adjacent speed parameter combination. By collecting historical data for these adjacent parameter combinations under the same test indicators, a multidimensional data matrix centered around the current parameter combination is formed.

[0066] The historical morphological data for the current parameter combination is compared and analyzed with the historical morphological data for adjacent parameter combinations. The impact of parameter changes on microsphere morphology is determined using a preset parameter threshold. The preset parameter threshold is a critical value pre-set based on the target physical properties of the hollow silica microspheres and the tolerance range of the preparation process. For example, the particle size change threshold can be set to ±15% of the target average particle size, the porosity change threshold can be set to ±10% of the target porosity, and the sphere roundness coefficient change threshold can be set to ±0.05 (based on the roundness coefficient of a standard circle of 1). The specific judgment process is as follows: If a key parameter (e.g., reaction temperature) in the current parameter combination changes the corresponding microsphere morphological indicator (e.g., average particle size) by more than the preset parameter threshold when adjusted to the value of the adjacent parameter combination, the current adjustment range of the parameter is considered to significantly affect microsphere performance, and the parameter change boundary needs to be included in the parameter optimization boundary. If the morphological indicator change does not exceed the threshold, the parameter is considered to have a minimal impact on microsphere performance within the current range, and the parameter adjustment range can be expanded or excluded from the key optimization parameters. By analyzing all key parameters (such as reaction temperature, silicon source concentration, template ratio, stirring speed, etc.) one by one, the parameter optimization boundary of the current parameter combination is finally determined, that is, the adjustable range of each key parameter. For example, the optimization boundary of reaction temperature is [70℃, 90℃], and the optimization boundary of silicon source concentration is [0.1mol / L, 0.3mol / L].

[0067] After determining the parameter optimization boundary, it is necessary to extract an optimization data subset from within the boundary according to the preset parameter window. The preset parameter window is a parameter sampling rule set to balance computational efficiency and optimization accuracy. It is usually centered on the key parameter values ​​of the current parameter combination and selects discrete parameter points according to a certain step size. For example, if the reaction temperature of the current parameter combination is 80°C and the step size of the preset parameter window is set to 5°C, then the temperature parameter optimization data subset extracted within the parameter optimization boundary [70°C, 90°C] is {70°C, 75°C, 80°C, 85°C, 90°C}; if the silicon source concentration is 0.2 mol / L and the step size is set to 0.05 mol / L, then the concentration parameter optimization data subset is {0.1 mol / L, 0.15 mol / L, 0.2 mol / L, 0.25 mol / L, 0.3 mol / L}. For the case of multiple parameter combinations, the optimized data subsets of each key parameter are combined into a complete parameter combination set through the Cartesian product method. For example, the optimized data subset containing two key parameters, temperature and concentration, will generate 5×5=25 different parameter combinations.

[0068] After extracting the optimized data subset, the flow field simulation and preparation experiments are confined to that subset. For flow field simulation, for each parameter combination in the optimized data subset, a corresponding reactor geometry model (e.g., a three-dimensional model based on the actual reactor dimensions) is established. The corresponding parameter combinations (e.g., reaction temperature, flow rate, pressure, etc.) are then input. Computational fluid dynamics (CFD) methods are then used to simulate the flow field distribution within the reactor. Data such as velocity vector diagrams, pressure contours, and turbulence intensity distributions are obtained. The effects of the fluid flow characteristics under different parameter combinations on the hollow silica microsphere formation process are analyzed. For example, by simulating the flow field distribution at different stirring speeds, it is possible to determine whether high turbulence regions will cause microsphere collisions and breakage, or whether low flow rate regions will cause uneven raw material distribution, thereby affecting the formation of hollow structures.

[0069] For the preparation experiments, experimental designs were conducted based on parameter combinations from the optimized data subset. Experimental conditions were strictly controlled, and microsphere morphology data for each parameter combination was recorded. During the experiment, a raw material solution was first prepared according to the parameter combination. For example, a silicon source (such as tetraethyl orthosilicate), a template (such as polyethylene glycol), and a solvent (such as ethanol) were mixed in a predetermined ratio at a specific temperature. A uniform reaction system was then formed by stirring or ultrasonic treatment. The reaction system was then transferred to a reactor and allowed to react at the set reaction temperature and stirring speed. After the reaction, the hollow silica microspheres were isolated through centrifugation, washing, and drying. Finally, the microsphere size distribution was measured using a laser particle size analyzer, the porosity was determined by nitrogen adsorption, and the sphere roundness, surface roughness, and hollow structure integrity were observed using a scanning electron microscope (SEM). These data were used as the preparation results for this parameter combination.

[0070] Example 2:

[0071] When determining the parameter combination type, the implementation method based on reaction temperature and physical property data as the core judgment basis is as follows:

[0072] For the current parameter combination, its reaction temperature parameter value is obtained. This value comes from the temperature data in the reactor monitored in real time during the preparation process, and is collected and recorded in real time by thermocouples or temperature sensors. The setting of the preset temperature threshold is based on the analysis of the thermodynamic characteristics of the hollow silica microsphere preparation process. For example, when the reaction temperature exceeds a certain critical value, it may trigger side reactions or cause a significant change in the decomposition rate of the template, thereby affecting the formation of the microsphere structure. This threshold is usually determined through preliminary exploratory experiments, such as conducting preliminary experiments at different temperatures to observe whether the microspheres have abnormal morphology (such as agglomeration, fragmentation, or collapse of the hollow structure). The temperature value at which obvious abnormal morphology first appears is set as the preset temperature threshold, assuming it is 100°C (this is only an example, the actual threshold needs to be determined according to the specific process).

[0073] If the reaction temperature of the current parameter combination is greater than the preset temperature threshold (such as 105°C), the parameter combination is directly determined to be a high-temperature parameter combination. Under high temperature conditions, the reaction kinetics rate is significantly improved, which may lead to accelerated nucleation and growth of microspheres, thereby forming a special structure different from that under normal temperature or low temperature conditions. For example, high temperature may promote the rapid hydrolysis and condensation of the silicon source to form a denser silicon oxide shell, or cause the template to quickly release gas, affecting the formation and stability of the hollow cavity. For high-temperature parameter combinations, a separate optimization process needs to be established, focusing on the effect of temperature on the reaction rate and product structure. For example, in the flow field simulation, the analysis of fluid viscosity changes and heat transfer characteristics at high temperature is added, and in the preparation experiment, a reactor made of high-temperature resistant materials is used and the heating rate is strictly controlled.

[0074] If the reaction temperature of the current parameter combination is not greater than the preset temperature threshold (e.g., ≤100°C), the physical property data classification process based on the first preset optimization algorithm is entered. The first preset optimization algorithm can use unsupervised learning methods such as the K-means clustering algorithm to perform cluster analysis on the particle size distribution and porosity data of the current parameter combination. First, multiple sets of particle size distribution data (e.g., the average particle size and particle size distribution width of microspheres in different batches of experiments) and porosity data (e.g., the porosity value converted from the BET specific surface area measured by the nitrogen adsorption method) accumulated in previous experiments for the current parameter combination are collected to form a two-dimensional data matrix (particle size-related indicators are one dimension, and porosity is the other dimension).

[0075] In the K-means clustering process, the number of clusters, K, is first determined. This value can be preset based on the distribution characteristics of the actual physical property data. For example, setting K=3 divides the data into three categories: high porosity and low dispersion, medium porosity and medium dispersion, and low porosity and high dispersion. Next, K cluster centers are randomly initialized. The Euclidean distance from each data point to each cluster center is calculated. The data point is assigned to the cluster with the closest cluster center, and the mean of each cluster is recalculated as the new cluster center. This process is repeated until the cluster centers no longer change significantly or a preset number of iterations (e.g., 50) is reached, ultimately obtaining data for each physical property category.

[0076] After obtaining the clustering results, it is necessary to extract the core features of the data for each physical property category. For particle size distribution, core features include average particle size, particle size standard deviation (reflecting dispersion), and particle size distribution peak position; for porosity, core features include average porosity and porosity coefficient of variation. For example, the core features of data for a physical property category might be: average particle size 500nm, particle size standard deviation 20nm, average porosity 35%, and porosity coefficient of variation 5%, indicating that the microspheres in this category have uniform particle size and high and stable porosity.

[0077] At the same time, the optimization potential level of the current parameter combination is evaluated based on the preparation conditions. Preparation conditions include reactor type (such as autoclave reactor, microfluidic reactor), stirring speed, order of raw material addition, reaction time, etc. The evaluation of the optimization potential level is achieved by establishing a multi-factor scoring model. For example, when the reactor type is a microfluidic reactor, it has better mass transfer and heat transfer performance, so the score is higher, corresponding to high optimization potential; when the stirring speed is lower than the critical value, it may cause uneven mixing of raw materials, and the score is lower, corresponding to low optimization potential. The scoring criteria for each factor are formulated based on process experience and preliminary experimental data, and the optimization potential level is finally obtained through weighted summation, which is divided into three levels: high, medium, and low.

[0078] The matching degree of the core features of each physical property category data is calculated with the core features of the preset physical property category. The preset physical property category is an ideal category predefined based on the target product requirements of hollow silica microspheres. For example, if the target product requires an average particle size of 400-600nm, a particle size standard deviation of ≤25nm, and an average porosity of 30%-40%, the core features of the preset physical property category can be set to an average particle size of 500nm, a particle size standard deviation of 20nm, and an average porosity of 35%. The matching degree is calculated using the cosine similarity algorithm, and the formula is:

[0079]

[0080] in, is the core feature vector of the physical property category data, is the core feature vector of a preset property category. If the calculated similarity value is greater than a preset matching threshold (e.g., 80%), the current parameter combination is considered a specific property parameter combination, indicating that its physical properties are close to the target product requirements. It is necessary to conduct precise optimization to address the subtle differences in the preset property category, such as adjusting the raw material ratio to further reduce the particle size standard deviation. If the similarity value is not greater than the preset matching threshold, it is considered a common property parameter combination, requiring a general optimization strategy to comprehensively adjust the particle size distribution and porosity.

[0081] Throughout the implementation process, the determination of reaction temperature provides a preliminary basis for the classification of parameter combination types. The separate classification of high-temperature parameter combinations helps to deal with optimization problems under special process conditions. The classification of physical property data based on clustering algorithms enables the refined grouping of parameter combinations under normal or low temperature conditions, making the optimization strategy more targeted. The evaluation of the optimization potential level combines the actual limitations of the preparation conditions to avoid proposing optimization solutions that are out of touch with process feasibility. The matching calculation directly links the physical property performance with the target requirements, ensuring that the determination of the parameter combination type is closely centered around the product goal. Through the above steps, a parameter combination type determination system based on temperature threshold, physical property clustering, potential evaluation, and matching analysis has been formed, providing a key basis for the subsequent generation of differentiated optimization processes.

[0082] All thresholds involved in the above process (such as preset temperature thresholds and matching thresholds), algorithm parameters (such as the number of clusters K in K-means), and evaluation model weight coefficients must be pre-calibrated and verified based on the specific preparation process and target product requirements to ensure the scientific and accurate determination of parameter combination types. Furthermore, as preparation data continues to accumulate, thresholds and model parameters can be regularly updated to adapt to changes in process conditions and adjustments to optimization goals.

[0083] Example 3:

[0084] When generating a parameter optimization process set, the implementation method for analyzing the raw material ratio data and parameter combination types is as follows:

[0085] For the current parameter combination, a second preset optimization algorithm (e.g., particle swarm optimization) is used to perform cluster analysis on the raw material ratio data. The raw material ratio data includes key parameters such as the concentration of the silicon source (e.g., tetraethyl orthosilicate), the ratio of the template (e.g., polyethylene glycol), the solvent (e.g., the volume ratio of ethanol to water), and the concentration of the catalyst (e.g., ammonia). The implementation steps of the particle swarm optimization algorithm are as follows:

[0086] 1. Particle Swarm Initialization

[0087] Determine the size of the particle group (e.g. set to 30 particles), each particle represents a set of raw material ratio parameter combinations, and its position vector X is composed of the ratio parameters of each raw material. For example, if the raw materials include silicon source concentration , template ratio and solvent ratio , then the particle's position vector can be expressed as X= At the same time, the particle's velocity vector V is initialized. This vector determines the particle's search direction and step size in the parameter space. The initial velocity is usually randomly generated within the parameter value range.

[0088] 2. Fitness value calculation

[0089] Define the fitness function to evaluate the quality of particle position. The fitness function needs to comprehensively consider the target physical properties of hollow silica microspheres, such as particle size distribution uniformity, porosity compliance rate, etc. Assume that the fitness function for:

[0090]

[0091] in: and is the weight coefficient ( ), reflecting the importance of particle size distribution uniformity and porosity, respectively; is the particle size standard deviation, which characterizes the dispersion of particle size distribution. The smaller the value, the better the uniformity. It is the matching score between the measured porosity value and the target value (range 0-1), obtained through normalization.

[0092] Calculate the fitness value corresponding to the current position of each particle. The higher the score, the closer the raw material ratio is to the optimization target.

[0093] 3. Individual Optimal Solution and Group Optimal Solution Update

[0094] Each particle records its own best historical position (the individual best solution, pbest) and its corresponding fitness value. The entire particle swarm records the best solution among all particles' best historical positions (the group best solution, gbest). Initially, the individual best solution is the particle's initial position, and the group best solution is the position with the highest fitness among all particles' initial positions.

[0095] 4. Particle flight direction adjustment

[0096] Update the particle's velocity and position according to the following formula:

[0097]

[0098]

[0099] in: and Respectively The particle in The speed and position at the iteration; is the inertia weight, which controls the degree to which the particle inherits the historical velocity; and is the acceleration constant, usually around 2, which is used to adjust the step size of the particle moving towards the individual optimal solution and the group optimal solution; and is a random number in the interval [0,1], introducing randomness to avoid local optimality.

[0100] Through the above formula, particles continuously adjust their flight direction in the search space, gradually approaching the individual optimal solution and the group optimal solution.

[0101] 5. Iteration termination and clustering result generation

[0102] The iterative process continues until a preset number of iterations is reached (e.g., 100) or the fitness value converges (e.g., the fitness of the swarm's optimal solution changes by less than 1% over 10 consecutive iterations). Ultimately, the positions of all particles constitute the optimized solution set, i.e., the data for each ratio category. For example, the optimized solution set might include different ratio combinations, such as high silicon source and low template, medium silicon source and medium template, and low silicon source and high template. Each category corresponds to a group of raw material ratios with similar fitness characteristics.

[0103] 6. Key Parameter Subset Screening and Impact Label Assignment

[0104] For each ratio category data, the influence weight of each raw material is first determined through sensitivity analysis. Sensitivity analysis can use the single factor variable method, fix other parameters, change a certain raw material ratio alone, and measure the change range of the microsphere morphology index. The greater the change range, the higher the influence weight of the raw material. For example, if the particle size increases by 15% on average when the silicon source concentration increases by 10%, and the porosity increases by 8% on average when the template ratio increases by 10%, then the influence weight of the silicon source concentration is Can be set to 0.6, the influence weight of the template ratio It is set to 0.3, and the sum of other parameter weights is 0.1.

[0105] We also analyzed the concentration gradients of each raw material within the ratio category data, specifically the magnitude of the difference between adjacent ratio parameter values. For example, consider a silicon source concentration gradient of 0.05 mol / L and a template ratio gradient of 5%. By combining the influence weights and concentration gradients, we identified a subset of key parameters that significantly impact microsphere morphology. For example, if the silicon source concentration had the highest influence weight and a large concentration gradient, we included it in the key parameter subset.

[0106] Based on the correlation characteristics of the key parameter subset (e.g., the silicon source concentration is negatively correlated with the template ratio, i.e., high silicon source concentration is often accompanied by a low template ratio) and the mean value of the preparation process index (e.g., the average particle size target value is 500nm), an impact label is assigned to each ratio category data. The impact label reflects the main impact direction of the raw material ratio in this category, for example:

[0107] If the key parameter is silicon source concentration, and the particle size is larger when the silicon source concentration is higher under this category, the label is "particle size sensitive";

[0108] If the key parameter is the template ratio, and the porosity is higher when the template ratio is higher in this category, the label is "porosity-dominated";

[0109] If the influence weights of each parameter are similar and there is no significant dominant factor, the label is "comprehensive impact type".

[0110] 7. Relationship with parameter optimization process

[0111] The selected subset of key parameters and impact labels were incorporated into the parameter optimization process set as the core elements of the preparation optimization content. For example, for the "particle size sensitive" ratio category data, the flow field simulation focused on analyzing the impact of silicon source concentration on the solute diffusion rate within the reactor, and a gradient adjustment of silicon source concentration was designed in the preparation experiment. For the "porosity dominated" category data, the impact of gas release during template decomposition on the flow field turbulence structure was simulated, and the efficiency of hollow structure formation under different template ratios was verified through experiments.

[0112] Example 4:

[0113] After determining the parameter combination type of the current parameter combination, the implementation methods of different optimization strategies need to be implemented according to the type difference as follows:

[0114] When the parameter combination type of the current parameter combination is determined to be an efficient parameter combination, it indicates that the performance of the microspheres of this combination is close to the target physical property parameter requirements, or that it has shown high optimization efficiency in the early optimization. At this time, it is necessary to deeply match the impact labels of each ratio category data with the preparation process indicators to generate a targeted adjustment strategy. The impact label reflects the core influence direction of the raw material ratio parameters on the microsphere morphology (such as "particle size sensitive" and "porosity dominated", etc.), and the preparation process indicators include the specific range of the target physical property parameters (such as average particle size 450-550nm, porosity 30%-40%) and process stability requirements (such as particle size standard deviation ≤25nm).

[0115] During the implementation process, the data for each ratio category and its impact label corresponding to the high-efficiency parameter combination were first extracted. For example, a high-efficiency parameter combination contained three ratio categories: Category A, with the impact label "particle size sensitive," had a high silicon source concentration and a particle size close to the upper target limit; Category B, with the label "porosity dominant," had a moderate template ratio and a porosity close to the lower target limit; and Category C, with the label "comprehensive impact," had a balanced ratio of parameters, but all indicators were slightly below the target value. Next, the key parameters corresponding to each category label were compared and analyzed with the preparation process indicators: For Category A, since the particle size is close to the target upper limit, it is necessary to determine whether there is a risk of microsphere agglomeration due to excessive particle size. If the process indicators require particle size uniformity to be prioritized, an adjustment plan is formulated to reduce the silicon source concentration, and the concentration is gradually reduced in steps of 0.05 mol / L. At the same time, the effect of reduced concentration on fluid shear force and microsphere growth rate is predicted through flow field simulation; For Category B, since the porosity is close to the target lower limit, it is necessary to analyze whether there is still room for improvement in the template ratio (limited by the solubility or cost of the template). If adjustment is allowed, the ratio is increased in steps of 2%, and the impact of the amount of gas generated by the decomposition of the template on the flow field stability is evaluated through simulation; For Category C, since all indicators have room for improvement, a multi-parameter coordinated adjustment strategy is adopted, such as simultaneously increasing the silicon source concentration and the template ratio, and controlling the adjustment amplitude within 5% each time to avoid performance deterioration due to excessive parameter fluctuations.

[0116] The core of the targeted adjustment strategy is to accurately approximate the target physical property parameters while minimizing the range of parameter changes. The "single factor priority" principle must be followed during the adjustment process, that is, only one key parameter is adjusted each time. After the simulation or experimental results after the parameter adjustment are stable, the next parameter adjustment is carried out to ensure that the independent influence of each parameter on the performance of the microspheres can be accurately identified. For example, when adjusting the silicon source concentration of category A, the template ratio and solvent composition are fixed, and the predicted values ​​of the microsphere particle size at different concentrations are obtained through flow field simulation to form a concentration-particle size relationship curve, and the optimal adjustment range is determined accordingly; in the experimental verification stage, preparation is carried out according to the parameter combination recommended by the simulation, and the particle size distribution and hollow structure integrity are monitored simultaneously using a laser particle size analyzer and a scanning electron microscope to ensure that no new quality problems are introduced during the adjustment process.

[0117] If the parameter combination type of the current parameter combination is determined to be an inefficient parameter combination, it means that the performance of its microspheres deviates significantly from the target physical property parameters, or shows low optimization efficiency in the early optimization. At this time, it is necessary to generate a parameter correction strategy based on the preset correction rules and physical property improvement goals. The preset correction rules are formulated based on the physical and chemical principles of the preparation process and early experimental experience, such as "prioritizing the adjustment of parameters with high influence weights" and "avoiding the simultaneous adjustment of strongly correlated parameters". The physical property improvement goals are clearly defined according to the target product requirements (such as reducing the particle size standard deviation to below 25nm and increasing the porosity to more than 30%).

[0118] The specific implementation steps are as follows: A sensitivity analysis is performed to determine the influence weights of each raw material ratio parameter in the current parameter combination (e.g., silicon source concentration, 40%, template ratio, 30%, stirring speed, 20%, and reaction temperature, 10%). The parameters are then ranked from highest to lowest according to their influence weights to determine the priority sequence for adjustment. For example, if the particle size standard deviation of an inefficient parameter combination reaches 40nm (far exceeding the target value of 25nm), and the sensitivity analysis shows that silicon source concentration has the greatest impact on particle size uniformity, silicon source concentration is selected as the primary parameter for adjustment. Next, a step-by-step correction plan is developed based on the physical property improvement goals and the feasible adjustment range of the parameters. For example, if the current silicon source concentration is 0.4 mol / L, the target average particle size is 500 nm, and the measured average particle size is 600 nm (too large), the concentration is reduced in stages with a step size of 0.08 mol / L, and adjusted to 0.32 mol / L in the first stage. The particle size distribution at this concentration is predicted through flow field simulation. If the simulation results show that the standard deviation can be reduced to 30 nm, experimental verification is carried out; if the experimental results deviate greatly from the simulation (such as the standard deviation is still 35 nm), the supplementary simulation module is triggered to analyze whether there are any unconsidered parameter correlation effects (such as the synergistic effect of silicon source concentration and stirring speed), and adjust the correction plan.

[0119] During the parameter correction process, special attention should be paid to the interaction between parameters. For example, a decrease in the silicon source concentration may lead to a decrease in the viscosity of the solution. If the stirring speed remains unchanged, it may cause insufficient turbulence intensity in the flow field, thereby affecting the uniformity of raw material mixing and causing porosity fluctuations. Therefore, while adjusting the silicon source concentration, it is necessary to simultaneously evaluate the adaptability of the stirring speed. If necessary, a coordinated adjustment should be made according to the preset parameter association rules (such as for every 10% decrease in viscosity, the stirring speed should be increased by 5%). In addition, for inefficient parameter combinations, the number of experimental repetitions and sample capacity should be increased (such as repeating the experiment 3 times and testing 50 microsphere samples each time) to ensure the reliability of the correction strategy and avoid misjudgment due to accidental errors.

[0120] When the current parameter combination is determined to be critical, its microsphere performance indicators are close to the qualified threshold (e.g., the average particle size just reaches the upper limit of 550nm, and the porosity is close to the lower limit of 30%). However, their stability is insufficient and susceptible to parameter fluctuations. In this case, a process stabilization strategy needs to be generated based on historical data fluctuations and process stability rules. Historical data fluctuation analysis analyzes the performance indicator variation range of the current parameter combination over the past 5-10 batches (e.g., particle size standard deviation fluctuates between 20-30nm, porosity fluctuates between 28%-32%) to identify the main sources of fluctuation (e.g., raw material purity changes, insufficient reactor temperature control accuracy, and detection equipment errors). Process stability rules include the allowable parameter fluctuation range (e.g., key parameter fluctuation range ≤5%), equipment calibration frequency (e.g., weekly), and process monitoring frequency (e.g., hourly recording of reaction temperature and pressure).

[0121] Specific implementation methods include: first, reducing the adjustment step of key parameters, such as shortening the adjustment step of silicon source concentration from 0.05mol / L to 0.02mol / L, and shortening the adjustment step of template ratio from 2% to 1%, so as to reduce performance fluctuations caused by large changes in parameters. Secondly, strengthen process monitoring and increase the frequency of online detection during the reaction process, such as using a laser particle size analyzer to monitor the changes in microsphere particle size in real time. When the detection value approaches the qualified threshold boundary (such as the particle size reaches 540nm), the parameter fine-tuning mechanism is automatically triggered (such as reducing the stirring speed by 10rpm to slow down the growth rate). In addition, establish an equipment abnormality early warning mechanism, such as monitoring the pressure drop change in the reactor through a pressure sensor. If the pressure drop suddenly increases and exceeds the preset threshold (such as 10kPa), it indicates that there may be microsphere agglomeration blocking the flow channel, and the cleaning program is immediately started and the stirring strategy is adjusted.

[0122] Calibration and compensation of testing equipment are particularly important in a process stabilization strategy. Before each experiment, the testing equipment is calibrated for error compensation using a standard particle size analyzer (standard microspheres of known diameter) and an electronic balance (calibrated reference weights). For example, if the nominal particle size of the standard microspheres is 500nm, but the laser particle size analyzer indicates 510nm, a -2% error compensation factor is applied to all subsequent test results. Furthermore, a raw material batch management ledger is established to record key indicators for each batch of raw materials (such as the degree of hydrolysis of the silicon source and the molecular weight distribution of the template). When changing raw material batches, pilot experiments are conducted in advance to assess the impact of batch differences on microsphere performance and ensure that process stability is not affected by raw material fluctuations.

[0123] Regardless of whether it is an efficient, inefficient, or critical parameter combination, after executing the corresponding optimization strategy, the adjusted parameter combination must be re-incorporated into the parameter optimization process set, its parameter combination type must be updated based on the simulation data and experimental data generated in real time, and the execution order in the process set must be dynamically adjusted. For example, if the performance of an inefficient parameter combination is significantly improved after correction, its type can be re-determined as a critical parameter combination or a general physical property parameter combination, and adjusted to a mid-to-late position in the process set accordingly; if the critical parameter combination achieves continuous performance indicators through a stable strategy, it can be upgraded to an efficient parameter combination to reduce subsequent optimization resource investment. Through this dynamic feedback mechanism, it is ensured that the entire parameter optimization process always focuses on the parameter combination that is most in need of improvement, thereby maximizing optimization efficiency.

[0124] Example 5:

[0125] After generating the parameter optimization process set, the implementation methods for establishing a dynamic update mechanism and supporting data management measures are as follows:

[0126] Implement real-time updates of preparation optimization content. During the preparation of hollow silica microspheres, a sensor network (such as temperature sensors, pressure sensors, and concentration detectors) collects real-time reaction environment data. Microsphere morphology and physical property data are also acquired through offline detection methods (such as laser particle size analyzers, scanning electron microscopes, and nitrogen adsorption analyzers). This real-time data is synchronously transmitted to the system's data center for comparison and analysis with historical preparation optimization content for each parameter combination (such as parameter optimization boundaries, simulation input parameters, and experimental configuration schemes). For example, if the real-time reaction temperature of a parameter combination exceeds the preset optimization boundary upper limit (e.g., set at 90°C and measured at 92°C), the system automatically triggers a boundary correction process. Based on the temperature-morphology correlation model (trained based on historical data), the temperature optimization boundary for this parameter combination is adjusted to [72°C, 92°C], and the input parameter range for the flow field simulation and the temperature gradient of the experimental design are updated accordingly.

[0127] Based on the updated preparation optimization content, the parameter combination type of each parameter combination is recalculated. The logic for determining the parameter combination type is consistent with the initial classification, namely, comprehensive consideration of reaction temperature, physical property data clustering results, optimization potential level, and matching degree with the preset physical property category. For example, a parameter combination was initially determined to be a common physical property parameter combination. However, in real-time data, its measured porosity values ​​were found to have exceeded the range corresponding to the preset matching threshold for three consecutive batches. The system then re-clustered its physical property data (particle size distribution, porosity) using the K-means algorithm and found that the matching degree of its core features with the preset physical property category of "high porosity, low dispersion" increased from 75% to 85%, thereby re-determining its parameter combination type as a specific physical property parameter combination. For another example, the reaction temperature of a parameter combination was reduced from 95°C to 90°C (below the preset temperature threshold of 100°C) due to equipment commissioning. Its type was adjusted from a high-temperature parameter combination to a common physical property parameter combination, and subsequent operations must be performed according to the optimization process corresponding to the new type.

[0128] According to the recalculated parameter combination type, the execution order of the parameter optimization process set is dynamically adjusted. The dynamic sorting principle is based on the optimization urgency and potential level of the parameter combination: specific physical property parameter combinations are given priority for fine-tuning because they are close to the target state; high-temperature parameter combinations require priority processing during equipment idle periods because they involve special process conditions; inefficient parameter combinations require a higher priority in optimization resources because of large performance deviations. For example, when a critical parameter combination is re-determined to be an inefficient parameter combination due to performance fluctuations, its execution order in the process set is advanced from 10th to 3rd to ensure timely allocation of simulation computing resources and experimental equipment resources. The sorting adjustment is achieved through an intelligent scheduling algorithm, which comprehensively considers factors such as parameter combination type, historical optimization time, equipment load status, etc. to generate the optimal execution sequence to avoid resource conflicts and inefficient waiting.

[0129] The calibration and compensation mechanism for testing equipment in the pre-set experimental rules is implemented as follows: Before each experiment, the operator must calibrate the testing equipment using a standard particle size analyzer (equipped with standard microspheres of a known diameter of 500 nm) and an electronic balance (equipped with a 100g reference weight calibrated by a metrology institute). During the calibration process, the standard microsphere sample is first injected into the laser particle size analyzer, the test program is run, and the particle size displayed by the instrument is recorded. If the measured value deviates from the nominal value by more than ±2% (for example, a display of 510 nm indicates a +2% deviation), the system automatically generates an error compensation factor (-2% in this example) and applies this factor to all subsequent particle size measurement data. For the electronic balance, if the displayed value deviates from the nominal value by more than ±0.1% (for example, a display of 100.1 g indicates a +0.1% deviation) by weighing the reference weight, the balance undergoes zero point calibration and linearity correction until the error meets the required tolerance. The calibration results are electronically recorded and stored in the preparation parameter optimization database, serving as a traceable basis for experimental data validity.

[0130] After generating a complete optimization plan for hollow silica microsphere preparation parameters, the process of establishing and maintaining the preparation parameter optimization database begins. This database utilizes a relational database architecture and contains core data tables such as the simulation input list, experimental configuration list, preparation optimization content, and parameter combination type table. The simulation input list stores the initial flow field simulation parameters for each parameter combination (such as reaction temperature, flow rate, and reactor dimensions); the experimental configuration list records experimental equipment parameter settings (such as stirring speed and raw material addition sequence) and sample size; the preparation optimization content stores parameter optimization boundaries, key parameter subsets, and impact labels; and the parameter combination type table records the type determination results and dynamic change history for each combination. For example, the simulation input list for a parameter combination includes temperature 80°C, silicon source concentration 0.2 mol / L, and template ratio 10%. The experimental configuration list includes stirring speed 200 rpm and sample size 50. The preparation optimization content includes temperature optimization boundaries [70°C, 90°C], key parameters are silicon source concentration and template ratio, and the parameter combination type is a common physical property parameter combination.

[0131] Database maintenance includes regular data cleanup and correction. Abnormal data detection is performed once a week. By setting data logic verification rules (such as the reaction temperature must be greater than 0°C and the porosity must be between 0% and 100%), abnormal records (such as entries with a temperature value of -5°C and a porosity of 120%) are identified and marked. For marked abnormal data, the original experimental records or simulation logs are first traced back to confirm whether they are caused by input errors or equipment failures. If it is an input error, the data is corrected directly; if it is caused by equipment failure, the record is deleted and the equipment maintenance process is triggered. At the same time, data redundancy cleaning is performed once a month to delete duplicate parameter combination records (such as duplicate entries of the same parameter combination with the same configuration) to free up storage resources.

[0132] Periodic comparative analysis based on the database is conducted monthly, using time series analysis to compare the evolving trends in the optimization effects of various parameter combinations. Specific analysis includes: changes in the number of various parameter combinations (e.g., the proportion of high-efficiency parameter combinations increased from 15% to 20%), the frequency and amplitude distribution of key parameter adjustments (e.g., the average silicon source concentration adjustment step size decreased from 0.08 mol / L to 0.05 mol / L), and changes in the fluctuation range of microsphere morphology indicators (e.g., the average particle size standard deviation decreased from 30 nm to 25 nm). Trend analysis is used to evaluate the effectiveness of the current optimization strategy. For example, if the optimization of a high-temperature parameter combination is found to be taking a prolonged period of time, consideration is given to optimizing the flow field simulation model under high-temperature conditions. If the conversion rate of an inefficient parameter combination is low, the rationality of the preset correction rules is reviewed, and the parameter influence weights or correction step sizes are adjusted.

[0133] The entire implementation process utilizes a dynamic update mechanism driven by real-time data to ensure that the parameter optimization process consistently adapts to actual changes in the preparation process. A calibration and compensation mechanism for testing equipment ensures the accuracy of experimental data from the source. The establishment and maintenance of a database provides reliable data support for the optimization process, while periodic analysis provides a scientific basis for strategic adjustments. These measures together form a closed-loop parameter optimization management system, enabling automated and refined management of the entire process, from data collection, type determination, process adjustment, to effect evaluation, effectively improving the efficiency and reliability of parameter optimization for hollow silica microsphere preparation.

[0134] Example 6:

[0135] This embodiment is used to analyze the implementation effect of parameter optimization for the preparation of hollow silica microspheres, and is supplemented with specific examples, comparative examples and analysis. The example section covers different scenarios: "Deep optimization of high-temperature environment parameter combinations" analyzes and optimizes the influence of high temperature on fluid properties for high-temperature parameter combinations, thereby improving the performance of microspheres; "Precise fine-tuning of high-efficiency parameter combinations" precisely adjusts the ratio for high-efficiency parameter combinations to improve various indicators of microspheres; "Stable guarantee of critical parameter combinations" reduces the adjustment step size, strengthens monitoring, and ensures stable performance for critical parameter combinations. The comparative examples include traditional empirical trial and error methods, unclassified optimization methods, and optimization methods without a dynamic update mechanism, which show their shortcomings such as low efficiency and unstable quality. Comparison of table data shows that the optimization method of this application can accurately control parameters, improve microsphere performance and preparation efficiency, excel in indicators such as average particle size, standard deviation, and porosity, has a short preparation cycle, and achieves advantages through multi-link collaboration, with significant technological progress. Specifically including:

[0136] Example 1: Deep optimization of high temperature environment parameter combinations

[0137] In this example, we selected a parameter combination with a relatively high reaction temperature for optimization. The reaction temperature of the current parameter combination reached 110°C, and the preset temperature threshold was 100°C, so this parameter combination was determined to be a high-temperature parameter combination.

[0138] In terms of flow field simulation, the focus was on analyzing the changes in fluid viscosity and heat transfer characteristics under high-temperature conditions. Simulation results show that high temperatures reduce fluid viscosity and accelerate heat transfer, significantly impacting the nucleation and growth of microspheres. During the preparation experiments, a reactor made of high-temperature-resistant materials was used, and the heating rate was strictly controlled to 5°C / min.

[0139] After multiple experiments and optimizations, the resulting hollow silica microspheres have a denser silica shell and enhanced stability within the hollow cavity. The microspheres have an average particle size of 550 nm, a standard deviation of 22 nm, and a porosity of 32%.

[0140] Example 2: Efficient parameter combination and precise fine-tuning

[0141] Parameter combinations identified as highly effective were selected for optimization. With this combination, microsphere performance approached the target physical property requirements. The impact labels for each ratio category indicate that silicon source concentration has the greatest impact on particle size, while template ratio has the greatest impact on porosity.

[0142] Following a targeted adjustment strategy, the silicon source concentration was first gradually decreased in steps of 0.03 mol / L, while the template ratio was simultaneously increased in 1% increments. The adjustment process strictly adhered to the "single factor priority" principle, adjusting only one key parameter at a time. The optimal adjustment range was determined through flow field simulation and experimental verification.

[0143] After the adjustment, the average particle size of the microspheres reached 505nm, the standard deviation of the particle size was reduced to 18nm, the porosity was increased to 35%, and all performance indicators were further improved.

[0144] Example 3: Critical Parameter Combination Stability Guarantee

[0145] A critical parameter combination was selected for optimization. The microsphere performance indicators of this parameter combination were close to the qualified threshold, but the stability was poor. Analysis of historical data fluctuations showed that the main sources of fluctuation were changes in raw material purity and insufficient reactor temperature control precision.

[0146] To improve process stability, the following measures were taken: shortening the adjustment steps of key parameters, reducing the silicon source concentration adjustment step from 0.05 mol / L to 0.02 mol / L, and the template ratio adjustment step from 2% to 1%; strengthening process monitoring, increasing the frequency of online testing, and recording the reaction temperature and pressure every 30 minutes; establishing an equipment abnormality early warning mechanism, and immediately starting the cleaning procedure and adjusting the stirring strategy when it is detected that the pressure drop in the reactor suddenly increases by more than 8 kPa.

[0147] After a period of stable operation, the performance indicators of the microspheres have been effectively controlled, the standard deviation of the particle size has stabilized between 20-22nm, and the porosity has stabilized between 30-32%.

[0148] Comparative Example 1: Traditional Trial and Error Method

[0149] The traditional trial-and-error method was used to optimize the preparation parameters of hollow silica microspheres. Repeated preparation experiments were conducted, and the microsphere morphology and physical properties under different parameter combinations were observed, and the parameters were gradually adjusted. Flow field simulations were not performed during the experiments, and no parameter combinations were classified or optimized specifically.

[0150] After multiple experiments, the average particle size of the microspheres obtained was 600nm, the standard deviation of the particle size was 35nm, and the porosity was 28%. The preparation cycle was long, consuming a lot of manpower, material resources, and time costs, and the microspheres were unstable and inconsistent.

[0151] Comparative Example 2: Optimization method without parameter combination classification

[0152] During the optimization process, parameter combinations were not classified, and a unified optimization strategy was used to optimize all parameter combinations. The flow field simulation and preparation experiments did not consider the characteristics and optimization potential of different parameter combinations, resulting in low optimization efficiency.

[0153] After optimization, the average particle size of the microspheres was 580 nm, the standard deviation was 30 nm, and the porosity was 30%. Although performance improved, the optimization results were not satisfactory compared to the example.

[0154] Comparative Example 3: Optimization method without dynamic update mechanism

[0155] The optimization process did not establish a dynamic update mechanism, did not update the preparation optimization content based on real-time preparation data, and did not recalculate the parameter combination type and adjust the process execution order.

[0156] As the preparation process progressed, the performance of the parameter combination changed, but the optimization strategy was not adjusted in a timely manner, resulting in a gradual decline in the optimization effect. The final microspheres had an average particle size of 570nm, a standard deviation of 28nm, and a porosity of 31%.

[0157] Comparison results table

[0158] category Average particle size (nm) Particle size standard deviation (nm) Porosity (%) Preparation cycle Optimize efficiency Example 1 550 22 32 Shorter high Example 2 505 18 35 Shorter high Example 3 520 (stable range) 20 - 22 30 - 32 Shorter high Comparative Example 1 600 35 28 long Low Comparative Example 2 580 30 30 Longer Low Comparative Example 3 570 28 31 Longer Low

[0159] Tabular data analysis

[0160] As can be seen from the table, the examples in this application outperform the comparative examples in terms of performance indicators such as average particle size, particle size standard deviation, and porosity. The examples have a short preparation cycle and high optimization efficiency, demonstrating that the parameter optimization method in this application can effectively improve the preparation quality and efficiency of hollow silica microspheres.

[0161] Comparative Example 1 adopts the traditional trial-and-error method. Due to the lack of in-depth understanding of the inherent physical mechanisms such as flow field distribution during the preparation process, the parameter adjustment is largely blind, resulting in unstable microsphere quality, poor consistency, long preparation cycle and low optimization efficiency.

[0162] Comparative Example 2 does not classify the parameter combinations and adopts a unified optimization strategy, which cannot perform personalized optimization for different types of parameter combinations, resulting in unsatisfactory optimization results.

[0163] Comparative Example 3 did not establish a dynamic update mechanism and could not respond to changes in the preparation process in real time, resulting in a gradual decline in the optimization effect.

[0164] In summary, the parameter optimization method of this application achieves precise control of the preparation parameters of hollow silica microspheres through parameter optimization planning, parameter combination type determination, optimization strategy formulation and execution, experimental rule design, data management and optimization effect evaluation, etc., which can significantly improve the performance and preparation efficiency of microspheres and has obvious technical advantages.

[0165] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0166] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for optimizing the preparation parameters of hollow silica microspheres based on flow field simulation, characterized in that: The method comprises: The parameter optimization process is planned based on the target physical properties and preparation conditions of the hollow silica microspheres. The flow field simulation and preparation experiment content required for optimization are determined. Computational fluid dynamics (CFD) simulation is used to establish a three-dimensional geometric model based on the actual reactor dimensions. The corresponding parameters of reaction temperature, flow rate, and pressure are input to simulate the flow field distribution of flow rate, pressure, and turbulence intensity parameters within the reactor. The preparation experiment content includes obtaining microsphere morphology data under different parameter combinations. Determining different parameter combination types according to the target physical properties and preparation conditions of the hollow silica microspheres; Generate a parameter optimization process set according to the different parameter combination types, flow field simulation contents and preparation experiment contents, and determine the execution order of the parameter optimization process set based on a dynamic sorting principle; Generating a simulation input list for each parameter combination under each parameter combination according to a preset optimization cycle, the different parameter combination types, and the preparation process index requirements, and generating an experimental configuration list for each parameter combination under each parameter combination according to preset experimental rules, the different parameter combination types, and the preparation process index requirements; Based on the execution order of the parameter optimization process set, the simulation input list, the experimental configuration list and the preparation optimization content of each parameter combination are associated to generate a complete solution for the preparation parameter optimization of the hollow silica microspheres.

2. The method for optimizing hollow silica microsphere preparation parameters based on flow field simulation according to claim 1, characterized in that: Based on the target physical properties and preparation conditions of hollow silica microspheres, the parameter optimization process is planned to determine the flow field simulation and preparation experiment content required for optimization, including: For a current parameter combination, determining a parameter optimization boundary of the current parameter combination based on historical morphology data of the current parameter combination, historical morphology data of adjacent parameter combinations, and a preset parameter division threshold; Extracting an optimized data subset of the current parameter combination according to the parameter optimization boundary and the preset parameter window; The contents of the flow field simulation and preparation experiment related data in the optimization data subset are determined as the preparation optimization contents of the current parameter combination.

3. The method for optimizing hollow silica microsphere preparation parameters based on flow field simulation according to claim 1, characterized in that: Different parameter combination types are determined according to the target physical properties and preparation conditions of the hollow silica microspheres, including: For a current parameter combination, if the reaction temperature of the current parameter combination is greater than a preset temperature threshold, determining the current parameter combination as a high-temperature parameter combination; When the reaction temperature of the current parameter combination is not greater than the preset temperature threshold, classifying the particle size distribution and porosity data of the current parameter combination based on a first preset optimization algorithm to obtain data of various physical property categories, and determining core features of the data of various physical property categories; Determining an optimization potential level of the current parameter combination according to the preparation conditions of the current parameter combination; When the matching degree between the core features of the physical property category data and the preset physical property category is greater than a preset matching threshold, determining the current parameter combination as a specific physical property parameter combination; When the matching degree is not greater than the preset matching threshold, the current parameter combination is determined to be a common physical property parameter combination.

4. The method for optimizing hollow silica microsphere preparation parameters based on flow field simulation according to claim 3, characterized in that: The preparation optimization content also includes reactor structural parameters and raw material ratio data. The different parameter combination types include high-efficiency parameter combination, low-efficiency parameter combination and critical parameter combination. According to the different parameter combination types, flow field simulation content and preparation experiment content, a parameter optimization process set is generated, including: For the current parameter combination, the raw material ratio data of the current parameter combination is clustered based on the second preset optimization algorithm to obtain data of each ratio category; for the current ratio category data, the current ratio category data is screened based on the influence weight and concentration gradient of each raw material in the current ratio category data to obtain a key parameter subset of the current ratio category data, and the influence label of the current ratio category data is determined according to the correlation characteristics of the key parameter subset and the mean value of the preparation process indicators.

5. The method for optimizing hollow silica microsphere preparation parameters based on flow field simulation according to claim 4, wherein: After determining whether the parameter combination type of the current parameter combination is an efficient parameter combination, the method further includes: If the parameter combination type of the current parameter combination is an efficient parameter combination, then matching the impact labels of the respective ratio category data with the preparation process indicators to generate a directional adjustment strategy; If the parameter combination type of the current parameter combination is an inefficient parameter combination, generating a parameter correction strategy based on a preset correction rule and a physical property improvement target; If the parameter combination type of the current parameter combination is a critical parameter combination, a process stabilization strategy is generated according to historical data fluctuations and process stabilization rules.

6. The method for optimizing hollow silica microsphere preparation parameters based on flow field simulation according to claim 5, characterized in that: After generating the parameter optimization process set, the following steps are further included: Updating the preparation optimization content of each parameter combination based on real-time preparation data; Recalculate the parameter combination type of each parameter combination according to the updated preparation optimization content; The execution order of the parameter optimization process set is dynamically adjusted according to the recalculated parameter combination type.

7. The method for optimizing hollow silica microsphere preparation parameters based on flow field simulation according to claim 1, characterized in that: The preset experimental rules include: Dynamically adjust the number of repetitions and sample size of the preparation experiment based on the confidence interval of historical experimental data; When the deviation between the measured morphology data and the flow field simulation results exceeds the preset tolerance threshold, the supplementary experiment module is automatically triggered.

8. The method for optimizing hollow silica microsphere preparation parameters based on flow field simulation according to claim 7, characterized in that: The preset experimental rules also include: Establish a calibration and compensation mechanism for detection equipment, and calibrate the detection equipment for error compensation based on the benchmark values ​​of the standard particle size analyzer and electronic balance before each experiment.

9. The method for optimizing hollow silica microsphere preparation parameters based on flow field simulation according to claim 1, wherein: After generating the complete scheme for optimizing the preparation parameters of the hollow silica microspheres, it also includes: Establishing a preparation parameter optimization database, storing the simulation input list, experimental configuration list, preparation optimization content, and parameter combination type data in the complete scheme in the database; Regularly clean and correct abnormal data in the database and eliminate invalid data records; Based on this database, the preparation parameter optimization results are periodically compared and analyzed to evaluate the evolution of the parameter optimization effect.

10. The method for optimizing hollow silica microsphere preparation parameters based on flow field simulation according to claim 4, characterized in that: The second preset optimization algorithm is a particle swarm optimization algorithm, and the specific steps of clustering the raw material ratio data of the current parameter combination include: Initialize the particle swarm position and velocity vector, calculate the fitness value of each particle and update the individual optimal solution; The particle flight direction is adjusted according to the swarm optimal solution, and the final optimized solution set is determined as the data for each ratio category by presetting the number of iterations or fitness convergence threshold.

Images