Method for producing sodium methoxide based on reaction heat regulation

Abstract

This invention discloses a method for producing sodium methoxide based on reaction heat regulation, belonging to the field of fine chemical process control technology. The method includes the following steps: continuously collecting temperature data at various spatial measurement points within the reactor and performing time-series gradient difference calculations to obtain the instantaneous heat flux direction and intensity between adjacent measurement points, thereby constructing a three-dimensional dynamic heat distribution matrix. This invention integrates three-dimensional thermal field modeling and bubble image recognition to achieve multi-source spatial mapping of temperature, bubbles, and heat flux, accurately identifying potential hot spot regions, and constructing a risk index based on multi-factor features to predict hot spot migration trends. When the risk index exceeds a threshold, it triggers periodic directional perturbation and pulsed stirring operations to break bubble aggregation, improve heat diffusion efficiency, effectively suppress hot spots, and significantly improve the thermal control accuracy and safety of the sodium methoxide reaction. This method is suitable for intelligent thermal management of highly exothermic reactions.

Description

Technical Field

[0001] This invention relates to the field of fine chemical process control technology, specifically to a method for producing sodium methoxide based on reaction heat regulation. Background Technology

[0002] Sodium methoxide production based on reaction heat regulation refers to a process method that dynamically monitors and precisely controls the exothermic reaction process during the preparation of sodium methoxide from methanol and metallic sodium. This method intelligently adjusts the feed rate, stirring intensity, and cooling medium flow rate by real-time acquisition of temperature, heat flow, and related thermophysical parameters in the reaction system, combined with changes in reaction rate and heat release intensity. This achieves effective management of reaction temperature and heat release. Precise control of the reaction heat not only prevents safety risks such as localized overheating and explosions but also improves the production rate and purity of sodium methoxide, reduces energy consumption, and achieves efficient, safe, and green sustainable development in sodium methoxide production.

[0003] The existing technology has the following shortcomings: In the reaction system of high-concentration methanol and metallic sodium, due to the high reaction rate, a large number of bubbles are often rapidly generated within the system. When the bubble generation rate is too fast and the distribution is uneven, dense bubble aggregation areas are easily formed in local areas within the reaction liquid. This obstructs the effective conduction path of local heat, preventing the heat released by the reaction from dissipating to the outer layer of the system in a timely manner, resulting in the migration of "hot spots" in the bubble aggregation area. Because these "hot spots" are covered by bubbles, they are difficult to capture in real time by conventional temperature monitoring methods, thus forming hidden high-temperature areas. If the bubbles in these high-temperature areas burst or the bubble clusters dissipate due to external disturbances, the accumulated heat will be released instantaneously into the surrounding medium, which can easily cause violent local boiling, material splashing, and even lead to drastic fluctuations in temperature and pressure throughout the reaction system. In severe cases, this can cause serious consequences such as system thermal control failure, equipment damage, or production interruption.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a sodium methoxide production method based on reaction heat regulation. By integrating three-dimensional thermal field modeling and bubble image recognition, a multi-source spatial mapping of temperature, bubbles, and heat flux is achieved, accurately identifying potential hot spot regions. A risk index is constructed based on multi-factor features to predict hot spot migration trends. When the risk index exceeds a threshold, a combined periodic directional perturbation and pulsating stirring operation is activated to break bubble aggregation, improve heat diffusion efficiency, effectively suppress hot spots, and significantly improve the thermal control accuracy and safety of the sodium methoxide reaction. This method is suitable for intelligent thermal management of highly exothermic reactions, thus solving the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for producing sodium methoxide based on reaction heat regulation, comprising the following steps:

[0007] The temperature of each spatial measuring point inside the reactor is continuously collected, and time-series gradient difference calculation is performed to obtain the instantaneous heat flux direction and heat flux intensity between adjacent measuring points, thereby constructing a three-dimensional dynamic heat distribution matrix.

[0008] Based on the three-dimensional dynamic thermal distribution matrix, spatial regions with high temperature fluctuation frequency and discontinuous heat flux direction are identified as high-frequency thermal disturbance regions; at the same time, bubble images of the surface and interior of the reaction medium are acquired to identify regions with abnormally high local bubble concentration.

[0009] The high-frequency thermal disturbance region and the bubble concentration anomaly region are spatially mapped. Regions that simultaneously exhibit severe temperature fluctuations and high-density bubble distribution characteristics in the same spatial location are extracted, marked as potential hot spot regions, and hot spot spatial labels are established for each potential hot spot region.

[0010] For each hot spot spatial label, the temperature fluctuation amplitude sequence, the frequency sequence of bubble concentration change per unit area, and the heat retention duration per unit volume within a preset time window are statistically analyzed and standardized to generate hot spot temperature characteristic factor, hot spot bubble density characteristic factor, and hot spot heat retention characteristic factor, respectively.

[0011] The generated feature factors are combined into a hot spot feature parameter vector, and the hot spot feature parameter vector is input into the hot spot migration risk modeling function established based on historical hot spot instability events to calculate the hot spot migration risk index.

[0012] When the hot spot migration risk index exceeds the set safety threshold, the direction of the stirring paddle is periodically switched between forward and reverse rotation based on the current spatial location and migration trend of the hot spot, and intermittent pulsating stirring operation is implemented to break the local bubble aggregation state and promote the uniform dispersion and rapid breakup of bubbles.

[0013] The steps for constructing a three-dimensional dynamic matrix of heat distribution covering the entire reaction space are as follows:

[0014] Before the reaction begins, a fixed sampling period is preset, and multiple sets of thermocouple measuring points are evenly distributed in different spatial locations inside the reactor. The real-time temperature values ​​of each measuring point are obtained through a high-frequency synchronous acquisition system in each sampling period to establish a complete time series temperature dataset.

[0015] Based on continuously acquired time-series temperature data, frame-by-frame temporal gradient difference calculation is performed to analyze the temperature change trend between adjacent time nodes. For each measuring point, an instantaneous heat flux direction vector and heat flux intensity vector model are constructed in its neighboring direction to reflect local heat transfer behavior.

[0016] Spatial location matching and data fusion are performed on the heat flux vector models of all measuring points. The heat flow information of discrete measuring points is continuously reconstructed through a three-dimensional interpolation algorithm, thereby forming a three-dimensional dynamic heat distribution matrix that can reflect the heat distribution and change law inside the reactor in real time.

[0017] The steps for identifying high-frequency thermal disturbance regions are as follows:

[0018] Based on the continuous temperature change data of each spatial grid in the three-dimensional thermal distribution dynamic matrix, the temperature fluctuation frequency curve of each grid within a preset time window is extracted.

[0019] By performing Fourier spectrum analysis on the temperature fluctuation frequency of each grid, spatial regions with a high frequency component ratio higher than a set threshold are extracted, and the degree of variation of the heat flux direction vector in the current region is calculated.

[0020] The spatial regions that simultaneously satisfy the condition that the temperature fluctuation frequency is higher than the set value and the heat flux direction vector is highly discontinuous in space are identified and marked as high-frequency thermal disturbance regions.

[0021] The steps to identify areas with abnormal bubble concentration are as follows:

[0022] During the reaction, a sequence of dynamic images of bubbles on the surface and inside the reaction medium is continuously acquired, and grayscale normalization and edge enhancement preprocessing are performed on each frame of the image.

[0023] The contour boundaries of each bubble are identified by an image edge reconstruction algorithm, and the bubble area, position and morphological features are extracted to generate a bubble distribution feature map for each frame of the image.

[0024] Based on morphological clustering algorithms, bubble features in continuous images are clustered and analyzed to identify spatial regions where the bubble density is significantly higher than that of the surrounding area per unit time and per unit area. These regions are identified as areas with abnormally high bubble concentration and are used for spatial correlation and fusion analysis with high-frequency thermal disturbance regions.

[0025] The steps for constructing potential hot spot regions and establishing hot spot spatial labels are as follows:

[0026] Based on the high-frequency thermal disturbance regions identified in the three-dimensional thermal distribution dynamic matrix, their spatial location information in the three-dimensional coordinate system is extracted, and the spatial coordinate set of the abnormal bubble concentration regions obtained from the bubble image clustering analysis is acquired simultaneously.

[0027] One-to-one spatial matching was performed on the two types of spatial data. Overlapping voxel analysis was used to determine whether there were both drastic temperature fluctuations and high-density bubble distribution characteristics in the same three-dimensional coordinate range. Overlapping regions that met both conditions were then selected.

[0028] Each overlapping spatial region that meets the conditions is defined as a potential hot spot region, and a unique hot spot spatial label is assigned to it. The label contains a spatial coordinate index, an identification timestamp, and a corresponding perturbation intensity factor.

[0029] The specific steps for generating the hot spot temperature characteristic factor, hot spot bubble density characteristic factor, and hot spot thermal retention characteristic factor are as follows:

[0030] Within a preset time window, the temperature change sequence, bubble recognition image sequence, and heat flux distribution data of each hot spot spatial label are extracted within the three-dimensional coordinate range. The temperature fluctuation amplitude curve, the bubble concentration change frequency curve per unit area, and the heat retention time sequence per unit volume are calculated respectively.

[0031] The acquired raw data sequence is numerically normalized to ensure that the parameters are comparable in the subsequent modeling process;

[0032] Based on the normalization results, the root mean square value of temperature fluctuation amplitude is extracted as the hot spot temperature feature factor, the maximum frequency peak value of bubble concentration change frequency is extracted as the hot spot bubble density feature factor, and the average duration of heat retention time series is extracted as the hot spot heat retention feature factor. The three together constitute the multidimensional feature expression of the hot spot region.

[0033] The specific steps for extracting the temperature change sequence, bubble recognition image sequence, and heat flux distribution data at the corresponding location within a preset time window, around the three-dimensional coordinate range of each hot spot spatial label, and calculating the temperature fluctuation amplitude curve, the bubble concentration change frequency curve per unit area, and the heat retention time sequence per unit volume are as follows:

[0034] Based on the three-dimensional coordinate position recorded in each hot spot spatial label, the temperature data corresponding to the spatial region within the entire preset time window is extracted from the three-dimensional thermal distribution dynamic matrix, and a continuous temperature time series is constructed.

[0035] Synchronously call the bubble recognition image sequence obtained by the image recognition system within the same spatial coordinate range, count the number of bubbles per unit area in each frame image, and calculate the bubble density change frequency curve per unit time.

[0036] Based on the local heat conduction velocity and direction data in the heat flux vector model, the cumulative heat flux residence time per unit volume in the hot spot spatial region is calculated, and a continuous heat residence time series is constructed to provide an accurate data foundation for subsequent feature factor extraction and risk modeling.

[0037] The steps for calculating the hot spot migration risk index are as follows:

[0038] The hot spot temperature feature factor, hot spot bubble density feature factor, and hot spot thermal retention feature factor corresponding to each hot spot spatial label are assembled into a hot spot feature parameter vector in a predetermined order.

[0039] The hot spot feature parameter vector is input into the hot spot migration risk modeling function constructed based on historical hot spot instability events. The pre-trained empirical weight coefficients are called in the hot spot migration risk modeling function to assign different importance levels to the three feature factors to reflect the differentiated influence of each factor on the hot spot migration trend.

[0040] A multi-index fusion algorithm is used to nonlinearly combine the weighted feature factors to output a unique hot spot migration risk index.

[0041] When the hot spot migration risk index exceeds the set safety threshold, based on the current spatial location and migration trend of the hot spot, the specific steps for periodically switching the forward and reverse rotation directions of the agitator and implementing intermittent pulsating stirring operation are as follows:

[0042] Based on the hot spot migration risk index corresponding to each hot spot spatial label and the preset hot spot migration risk index reference threshold, the disturbance response frequency adjustment coefficient for adjusting the stirring response frequency is calculated, and the calculation expression is as follows:

[0043] ,

[0044] In the formula, R d It is a hot spot migration risk index. R t This is the reference threshold for the hot spot migration risk index. l f It is the disturbance response frequency adjustment coefficient;

[0045] Obtaining the disturbance response frequency adjustment coefficient l f Then, a periodic signal is generated to control the rotation direction of the agitator, which drives the agitator to switch between forward and reverse rotation at a dynamic frequency. The generation formula is as follows:

[0046] ,

[0047] In the formula, i (t () is the stirring direction control signal. t It is a time variable. f It is the initial phase angle. sign It is a sign determination function;

[0048] Combined with disturbance response frequency adjustment coefficient l f With stirring direction control signal i ( t The rotational speed of the agitator is adjusted in real time to achieve pulsating agitation. The adjustment formula is as follows:

[0049] ,

[0050] In the formula, A ( t () represents the rotational speed range of the agitator at the current time. A 0 This is the reference stirring speed. α It is the pulsation enhancement coefficient.

[0051] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0052] This invention, through the deep integration of 3D thermal field modeling and bubble image analysis, achieves for the first time spatial mapping and identification of multi-source data on "temperature-bubble-heat flux," accurately locating potential hot spot regions. Furthermore, it constructs a risk index using multi-factor feature parameters to predict hot spot migration trends. Then, when the risk index exceeds a threshold, it actively disrupts bubble aggregation structures using a combined mechanism of periodic directional perturbation and pulsating stirring, enhancing heat diffusion rates and effectively suppressing hot spot formation and expansion. Overall, this method significantly improves the thermal control accuracy and safety stability of the sodium methoxide reaction process, providing a predictable, interventionist, and sustainable intelligent thermal management solution for the industrialization of high-risk, strongly exothermic reactions. Attached Figure Description

[0053] Figure 1 This is a flowchart of the sodium methoxide production method based on reaction heat regulation according to the present invention. Detailed Implementation

[0054] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0055] This invention provides, for example Figure 1The sodium methoxide production method shown includes the following steps:

[0056] The temperature of each measuring point in the reactor is continuously collected at a fixed time period, and the temporal gradient difference calculation is performed to obtain the instantaneous heat flux direction and heat flux intensity between adjacent measuring points, thereby constructing a three-dimensional dynamic heat distribution matrix covering the entire reaction space.

[0057] This step aims to achieve real-time, comprehensive, and accurate monitoring of heat transfer behavior during the reaction of methanol and metallic sodium, providing fundamental dynamic thermal field information support for the entire thermal control system. Because this type of reaction is a violently exothermic reaction, accompanied by the generation of a large amount of gas, the heat release rate is rapid, and local temperature fluctuations are large, posing a risk of forming localized overheating areas (i.e., hot spots). Therefore, relying solely on static temperature values ​​from a few measuring points cannot fully reflect the complex thermal distribution state inside the reactor, nor can it promptly detect potential heat accumulation hazards. This step, by deploying multiple sets of thermocouple measuring points inside the reactor and synchronously collecting the temperature changes of each measuring point at fixed time intervals, combined with time-series gradient difference calculations, can effectively identify the direction and intensity of heat transfer between each measuring point and its neighboring measuring points. By establishing a spatial heat flux vector model and further employing a three-dimensional interpolation algorithm, these heat flux data are continuously reconstructed, generating a three-dimensional dynamic thermal distribution matrix covering the entire reaction space. This matrix not only reflects the current temperature level of each region, but also plots the heat conduction path and accumulation trend in space in real time, providing key support for subsequent identification of thermal disturbance areas, bubble interference areas, potential hot spots and implementation of intelligent control. It is the basic core step in building a refined reaction heat management system.

[0058] The steps for constructing a three-dimensional dynamic matrix of heat distribution covering the entire reaction space are as follows:

[0059] Before the reaction begins, a fixed sampling period is preset, and multiple sets of thermocouple measuring points are evenly distributed in different spatial locations inside the reactor. The real-time temperature values ​​of each measuring point are obtained through a high-frequency synchronous acquisition system in each sampling period to establish a complete time series temperature dataset.

[0060] The placement of thermocouple measuring points inside the reactor should adhere to three principles: spatial uniformity, thermal response sensitivity, and structural compatibility, ensuring the representativeness and dynamic response capability of the acquired temperature data. Specific placement methods include: establishing a three-dimensional coordinate grid system along the reactor's axial, height, and radial directions; uniformly arranging thermocouple measuring points at the grid intersections, prioritizing coverage of key thermally sensitive areas such as those near the sodium dosing port, the agitator, areas prone to bubble accumulation, and around cooling pipes. Each thermocouple group should be securely suspended using fixed brackets or corrosion-resistant support arms to prevent displacement due to liquid disturbance or bubble effects, while ensuring that the measuring points do not obstruct the agitator's movement path. To improve data synchronization, all measuring points must be connected to the same high-speed multi-channel data acquisition system to achieve millisecond-level synchronous acquisition, forming a temperature measurement network structure with high spatial resolution and high temporal accuracy.

[0061] Based on continuously acquired time-series temperature data, frame-by-frame temporal gradient difference calculation is performed to analyze the temperature change trend between adjacent time nodes. For each measuring point, an instantaneous heat flux direction vector and heat flux intensity vector model are constructed in its neighboring direction to reflect local heat transfer behavior.

[0062] To construct a model of the instantaneous heat flux direction vector and heat flux intensity vector for each measuring point in its vicinity, it is necessary to base this model on the temperature gradient change relationship between the measuring point and several neighboring measuring points at continuous time nodes. Specifically, the method involves first selecting multiple nearest-neighbor measuring points in three-dimensional space for each target measuring point, calculating their temperature change rate per unit time, and then, based on Fourier's law of heat conduction, determining the direction and intensity of heat transfer with the measuring point as the center, combined with relative position coordinates and temperature differences. This results in the construction of a vector model pointing in the direction of heat flow transfer from the measuring point. The purpose of this vector model is to realistically depict the instantaneous heat conduction path and velocity in the reaction system, achieving a visual representation of the heat flow field, and providing fundamental data support and precise physical basis for subsequent hot spot identification, thermal risk analysis, and control decisions.

[0063] Spatial location matching and data fusion are performed on the heat flux vector models of all measuring points. The heat flow information of discrete measuring points is continuously reconstructed through a three-dimensional interpolation algorithm, thereby forming a three-dimensional dynamic heat distribution matrix that can reflect the heat distribution and change law inside the reactor in real time.

[0064] Three-dimensional interpolation is a numerical computation method used to smoothly fit and continuously complete discrete data points in three-dimensional space. Its purpose is to predict and generate estimated values ​​at arbitrary spatial locations based on known data from finite spatial measurement points, thereby constructing a continuous spatial distribution field. In sodium methoxide production, due to the limited number of thermocouple measurement points, only locally discrete heat flow data can be provided. Therefore, three-dimensional interpolation can mathematically fit the heat flow information between these discrete points, completing the unknown regions in space and forming a continuous and smooth heat distribution map. This algorithm can reflect the transitional trend of heat conduction in different spatial regions, eliminate information gaps caused by the spacing between measurement points, and improve the accuracy and completeness of thermal field modeling. The specific steps include: First, constructing a three-dimensional spatial coordinate system using the three-dimensional coordinate position of each measuring point and the corresponding heat flux vector as input data; second, selecting an appropriate interpolation algorithm such as cubic spline interpolation, inverse distance weighted interpolation, or Kriging interpolation, and generating intermediate point data based on the neighbor point weight function; finally, numerically filling the entire reactor space grid through the interpolation function to output continuous three-dimensional heat flux distribution results, thereby forming a three-dimensional heat distribution dynamic matrix that can be updated in real time, which is used to support subsequent hot spot identification and risk control operations.

[0065] Based on the three-dimensional thermal distribution dynamic matrix, spatial regions with high temperature fluctuation frequency and discontinuous heat flux direction are identified as high-frequency thermal disturbance regions; at the same time, bubble images of the surface and interior of the reaction medium are acquired, and regions with abnormally high local bubble concentration are identified through image edge reconstruction algorithm and morphological clustering algorithm.

[0066] This step aims to identify key spatial regions with potential thermal control risks in the reaction system through multi-source data fusion analysis, providing a foundation for subsequent hotspot localization and dynamic control. In the strongly exothermic reaction of sodium methoxide, rapid local temperature fluctuations and abnormal bubble aggregation are the main causes of thermal instability and equipment safety accidents. Simply relying on temperature monitoring or image recognition often suffers from insufficient accuracy or reaction lag, making it difficult to fully reflect the complex dynamic behavior of the heat-flow-gas coupling. This step first identifies regions with high-frequency temperature fluctuations in the time dimension and disordered heat flux direction in the spatial dimension through frequency analysis of the three-dimensional thermal distribution dynamic matrix, reflecting the potential for violent and unbalanced heat release in these regions. Subsequently, a high-frame-rate image acquisition system is used to identify bubble behavior on and inside the reaction medium, using image edge reconstruction algorithms to extract bubble boundaries, and then morphological clustering algorithms to determine regions with abnormal bubble density. Finally, spatial location matching of the data from these two dimensions accurately pinpoints regions that simultaneously exhibit rapid thermal disturbance and high bubble concentration characteristics, which are the potential locations of high-risk hotspots. This identification mechanism not only improves the spatial resolution and timeliness of thermal control early warning, but also provides quantitative and intelligent basis for subsequent hot spot intervention. It is an indispensable key technology link for achieving safe, efficient and green sodium methoxide production.

[0067] The steps for identifying high-frequency thermal disturbance regions are as follows:

[0068] Based on the continuous temperature change data of each spatial grid in the three-dimensional thermal distribution dynamic matrix, the temperature fluctuation frequency curve of each grid within a preset time window is extracted.

[0069] By performing Fourier spectrum analysis on the temperature fluctuation frequency of each grid, spatial regions with a high frequency component ratio higher than a set threshold are extracted, and the degree of variation of the heat flux direction vector in the current region is calculated.

[0070] This threshold is a key criterion for determining whether temperature fluctuations are in a high-frequency disturbance state. Its setting needs to be determined comprehensively based on experimental data analysis, historical hotspot formation cases, and the thermal stability parameters of the reaction system. Specifically, firstly, Fourier spectrum analysis is performed on temperature sampling data from a large number of typical reaction batches to statistically analyze the spectral characteristics of each spatial grid under stable and unstable states, extracting the proportion of high-frequency components (e.g., above 0.5 Hz) in the total spectral energy. Secondly, based on a comparison of the spectral characteristics at the time of hotspot appearance, the critical high-frequency proportion threshold that frequently occurs in historical thermal control instability events is identified, for example, exceeding 20%-30% of the total energy. Finally, considering the current reactor volume, stirring speed, thermal inertia, and heat transfer delay characteristics, a dynamically adjustable high-frequency proportion threshold range is set using a combination of empirical adjustment and safety margin control. This serves as the baseline for identifying high-frequency thermal disturbance regions in real-time analysis, thereby improving the sensitivity and accuracy of thermal control early warning.

[0071] Spatial regions that simultaneously satisfy the conditions of temperature fluctuation frequency being higher than a set value and heat flux direction vector exhibiting a highly discontinuous state in space are identified and marked as high-frequency thermal disturbance regions, and used for subsequent hot spot determination and analysis.

[0072] This step aims to accurately identify potentially hazardous areas within the reactor exhibiting signs of thermal instability, providing crucial early warning information for subsequent hotspot detection and risk control. By performing Fourier spectrum analysis on the temperature change data of each spatial grid, high-frequency temperature fluctuations occurring in localized areas within a short period can be effectively captured. These fluctuations are often precursors to instability phenomena such as violent reactions, localized overheating, or bubble disturbances. Further analysis of the spatial variability of the heat flux direction vector reveals whether the heat conduction path within the region is disordered, reversed, or abnormally shifted, thus identifying high-risk areas exhibiting both rapid temperature fluctuations and abrupt changes in heat flow direction. This dual-indicator coupling-based judgment mechanism is more sensitive and forward-looking than traditional single-point temperature exceedance monitoring, enabling early identification and intervention before large-scale hotspot formation, providing real-time dynamic support for stable system operation and safe production.

[0073] The steps to identify areas with abnormal bubble concentration are as follows:

[0074] During the reaction, a sequence of dynamic images of bubbles on the surface and inside the reaction medium is continuously acquired, and grayscale normalization and edge enhancement preprocessing are performed on each frame of the image.

[0075] Acquiring dynamic image sequences of bubbles on and inside the reaction medium typically relies on high-frame-rate industrial camera modules or endoscopic image acquisition devices integrated at specific locations within the reactor. First, high-temperature resistant and corrosion-resistant imaging equipment is placed near the transparent observation window or optical acquisition channel of the reactor. Through multi-angle viewing channels combined with a backlighting system, continuous imaging of the bubble generation process on the reaction liquid surface and inside the liquid is achieved, forming a dynamic image sequence at a rate of tens of frames per second or higher. Second, each acquired frame undergoes grayscale normalization processing, converting the original color image into a grayscale image under a uniform brightness standard, compressing color interference, and enhancing grayscale contrast. Finally, edge enhancement algorithms such as Sobel or Canny edge detection are applied to strengthen the identification of bubble boundaries, improving the accuracy of subsequent image segmentation and feature extraction. Through this image acquisition and preprocessing process, high-precision, high-definition capture of bubble dynamic behavior can be achieved, providing a reliable data foundation for subsequent bubble identification and aggregation area analysis.

[0076] The contour boundaries of each bubble are identified by an image edge reconstruction algorithm, and the bubble area, position and morphological features are extracted to generate a bubble distribution feature map for each frame of the image.

[0077] Image edge reconstruction algorithms are image processing methods that utilize edge information from abrupt changes in grayscale to reconstruct the complete outer contour of objects (such as bubbles) in an image. The core of these algorithms lies in accurately separating bubbles from complex backgrounds through steps such as edge detection, boundary tracking, and contour closure. This algorithm can be implemented based on existing classic methods, such as Canny edge detection, the Sobel operator, the Laplace operator, and edge gradient enhancement combined with contour tracking (such as Suzuki-Abe contour finding). The implementation process is as follows: First, the Canny algorithm is applied to the image after grayscale normalization and edge enhancement to extract an initial set of edge points, combined with binarization to highlight the bubble boundary structure. Second, a boundary tracking algorithm (such as OpenCV's findContours function) is used to reconstruct the contour of the edge point set, forming a closed bubble boundary curve. Finally, geometric parameters are calculated for each closed contour, including calculating the pixel area within the contour to determine the bubble size, calculating the centroid position to determine the bubble distribution coordinates, and analyzing the bubble's morphological characteristics through parameters such as the contour's aspect ratio, roundness, and edge smoothness. By mapping the area, position, and morphological parameters of all bubbles to image coordinates, a bubble distribution feature map of the frame image can be generated, providing accurate data support for identifying high-density clusters and dynamic evolution trends.

[0078] Based on morphological clustering algorithms, bubble features in continuous images are clustered and analyzed to identify spatial regions where the bubble density is significantly higher than that of the surrounding area per unit time and per unit area. These regions are identified as areas with abnormally high bubble concentration and are used for spatial correlation and fusion analysis with high-frequency thermal disturbance regions.

[0079] Morphological clustering algorithms are clustering analysis methods based on the geometric shape, spatial distribution, and structural features of target objects in an image. They are commonly used to classify objects with similar morphological features into the same category, thereby identifying high-density clustered regions. In bubble recognition, this algorithm can be implemented using existing algorithms such as density-based clustering (DBSCAN), K-means clustering, MeanShift clustering, or graph-based segmentation clustering (such as Graph-Based Image Segmentation). The specific implementation steps are as follows: First, the bubble contours extracted from consecutive image frames are parameterized, extracting features such as the center coordinates, area, and boundary morphology of each bubble, and constructing these feature data into a spatial point set; second, the DBSCAN algorithm is used to perform clustering analysis on this point set. Based on the set spatial proximity distance and minimum sample size, bubbles densely distributed within a unit time and unit area are grouped into the same cluster, while isolated bubbles are removed as noise points; finally, statistical analysis is performed on the formed bubble clusters to identify spatial clusters with bubble density much higher than the overall average, and their locations are marked as areas with abnormally high bubble concentration. This method can dynamically identify high-density bubble aggregation areas that may cause local heat transfer failure and hot spot risks during the reaction process, providing spatial basis for thermal control regulation and safety early warning.

[0080] This step aims to achieve early identification and spatial localization of abnormal bubble behavior during the reaction process through intelligent image analysis, thereby providing real-time and visualized auxiliary judgment for the reaction thermal control system. Bubbles are not only reaction byproducts in the reaction system but also directly affect heat conduction and local temperature equilibrium. Especially in high-concentration reaction systems, bubble aggregation easily forms thermal stagnation zones, leading to hot spots or local instability. By extracting bubble contour information through high-frame-rate image acquisition and edge reconstruction algorithms, and using morphological clustering algorithms to identify areas with high-density aggregation per unit time and unit area, potential areas of abnormal bubble accumulation can be effectively located. These areas are often key points where heat conduction paths are blocked and heat is difficult to dissipate, and they highly overlap with areas of high-frequency thermal disturbance. Spatially fusing this identification result with the heat distribution matrix helps enhance the robustness and spatial resolution of the hot spot early warning model, improves the system's response speed and intervention accuracy to local instability phenomena, and ensures the continuity, safety, and intelligent level of thermal control in the sodium methoxide production process.

[0081] Under current industrial technology, the feasibility of multi-point temperature acquisition and bubble behavior image monitoring inside the reactor during the strongly exothermic reaction of metallic sodium and methanol has a relatively mature engineering implementation path. Specifically, high-temperature explosion-proof industrial endoscopic imaging systems widely used in current chemical plants, such as endoscopic fiber optic camera modules with corrosion resistance, non-conductivity, and high-pressure steam resistance, have been successfully applied to reaction systems under various high-temperature, high-pressure, or corrosive environments. These systems can achieve real-time image acquisition of specific spatial areas inside the reactor by setting optical observation windows (such as quartz glass observation windows) or miniature probe inlets on the stainless steel reactor body, thereby supporting bubble identification and concentration estimation. In particular, setting the optimal viewing path at the top of the reactor or around the stirring shaft is more helpful in observing the dynamic behavior of bubble generation, aggregation, and collapse, providing key information for hot spot identification.

[0082] For temperature acquisition, a corrosion-resistant thermocouple array can be uniformly arranged in three-dimensional space inside the reactor, or a flexible thin-film thermistor grid can be encapsulated in an alkali-resistant polymer material, with temperature signals extracted through directional perforation or thermal insulation structures. These measuring points, through insulation and grounding protection, can achieve millisecond-level synchronous data acquisition without affecting reaction stability, and a temperature time series and spatial thermal field model can be constructed using a high-speed acquisition card.

[0083] The high-frequency thermal disturbance region and the bubble concentration anomaly region are spatially mapped. Regions that simultaneously exhibit severe temperature fluctuations and high-density bubble distribution characteristics in the same spatial location are extracted, marked as potential hot spot regions, and hot spot spatial labels are established for each potential hot spot region.

[0084] This step aims to accurately identify key spatial regions within the reactor that may pose a severe risk of thermal runaway by fusing multidimensional heterogeneous data. These high-risk regions are then digitally identified and dynamically managed using hotspot spatial labels, providing clear targets and response guidelines for subsequent thermal control strategies. In the highly exothermic reaction of methanol and metallic sodium, localized heat accumulation and bubble aggregation are often the core causes of safety accidents such as hotspots, boiling splashes, and even explosions. Single-dimensional monitoring methods (such as relying solely on temperature or bubble images) are insufficient to fully reflect the true risk state under the triple coupling mechanism of heat, flow, and gas. Therefore, this step uses spatial location mapping to perform spatial overlap analysis in a three-dimensional coordinate system on high-frequency thermal disturbance regions derived from the three-dimensional thermal distribution dynamic matrix and regions with abnormal bubble concentrations identified through image processing and clustering algorithms. Regions exhibiting both severe temperature fluctuations and abnormal bubble density aggregation at the same spatial location are defined as potential hotspot regions. Subsequently, by assigning a unique hotspot spatial label to each potential hotspot region, containing its three-dimensional coordinate index, identification time, disturbance intensity, and other core information, continuous tracking, data archiving, and response triggering of hotspots are achieved. This labeling mechanism not only enhances the intelligence level of the thermal control system, but also transforms the risk identification of the entire reaction process from "overall monitoring" to "targeted management," laying a key foundation for precise intervention, risk prediction, and thermal safety assurance.

[0085] The steps for constructing potential hot spot regions and establishing hot spot spatial labels are as follows:

[0086] Based on the high-frequency thermal disturbance regions identified in the three-dimensional thermal distribution dynamic matrix, their spatial location information in the three-dimensional coordinate system is extracted, and the spatial coordinate set of the abnormal bubble concentration regions obtained from the bubble image clustering analysis is acquired simultaneously.

[0087] One-to-one spatial matching was performed on the two types of spatial data. Overlapping voxel analysis was used to determine whether there were both drastic temperature fluctuations and high-density bubble distribution characteristics in the same three-dimensional coordinate range. Overlapping regions that met both conditions were then selected.

[0088] Overlapping voxels refer to spatially overlapping regions in 3D rasterized modeling where two or more datasets with different sources or characteristics coexist on the same spatial voxel unit (i.e., 3D pixel block). Their function is to map multidimensional physical field features (such as temperature perturbation and bubble density) onto a unified spatial unit, thereby achieving precise spatial correspondence and logical judgment of heterogeneous data. In the implementation process, firstly, high-frequency thermal perturbation regions and anomalous bubble concentration regions are rasterized into 3D voxel data structures. Each voxel unit contains its position index in the X, Y, and Z coordinate systems and its corresponding physical feature value (such as temperature fluctuation amplitude, high-frequency proportion, or bubble density factor). Secondly, spatial overlap analysis is performed on the two datasets using the voxel index to find units where thermal perturbation voxels and anomalous bubble voxels coexist at the same coordinate position. Finally, attribute filtering is performed on the overlapping voxels, retaining only voxels whose temperature perturbation intensity exceeds a set threshold and whose bubble density index is higher than the overall field average, forming candidate hotspot regions that meet both conditions. This method can effectively integrate features from multiple data sources, improve the spatial accuracy and precision of hotspot identification, and provide a spatial decision-making basis for subsequent control strategies.

[0089] Each overlapping spatial region that meets the conditions is defined as a potential hot spot region and assigned a unique hot spot spatial label. The label contains a spatial coordinate index, an identification timestamp, and a corresponding disturbance intensity factor, which serves as the basic index unit for subsequent hot spot feature analysis and risk modeling.

[0090] This step uniquely identifies and structurally manages potential hotspot regions, giving them traceable, quantifiable, and dynamically updatable attributes. This facilitates precise intervention and command matching in subsequent monitoring, risk assessment, and thermal control response. By assigning a unique hotspot spatial label to each overlapping spatial region that meets both thermal disturbance and bubble aggregation characteristics, the system can not only pinpoint the hotspot's exact location in a three-dimensional coordinate system but also record its initial identification time and the numerical level of the disturbance intensity factor, enabling dynamic tracking and behavioral modeling of the hotspot's lifecycle. This labeling mechanism is equivalent to setting digital identification tags for runaway points in complex reaction spaces, allowing the system to dynamically adjust cooling medium flow, stirring methods, or raw material addition strategies based on the parameters carried by the tags. This achieves refined closed-loop thermal control management centered on the hotspot, effectively preventing localized overheating and improving reaction stability and safety.

[0091] For each hot spot spatial label, the temperature fluctuation amplitude sequence, the frequency sequence of bubble concentration change per unit area, and the heat retention duration per unit volume within a preset time window are statistically analyzed and standardized to generate hot spot temperature characteristic factor, hot spot bubble density characteristic factor, and hot spot heat retention characteristic factor, respectively.

[0092] The purpose of this step is to quantify the dynamic behavior of hot spot spatial regions from multiple dimensions, constructing a feature factor system under a unified dimension, and providing a precise numerical input basis for subsequent hot spot risk modeling and migration trend prediction. Hot spots in the sodium methoxide reaction process are often driven by multiple physical variables, including temperature perturbations, bubble aggregation behavior, and local heat retention effects. Relying on a single parameter alone is insufficient to comprehensively describe their risk status. Therefore, this step systematically extracts the temperature fluctuation amplitude sequence, the frequency sequence of bubble concentration changes per unit area, and the duration of heat retention per unit volume within a preset time window, around the three-dimensional coordinate region corresponding to each hot spot spatial label. These are then represented as time-varying curves, realistically reflecting the dynamic changes of the hot spot region under the coupling effects of heat, flow, and gas. Subsequently, the three types of data are standardized to eliminate differences in physical dimensions, orders of magnitude, and acquisition accuracy of the original data, making them comparable and fusionable. Furthermore, key statistical features from each curve—such as the root mean square value of temperature fluctuations, the peak value of the dominant frequency of bubbles, and the average duration of thermal retention—are extracted as three characteristic factors representing the intensity of thermal disturbance, the dynamic aggregation of bubbles, and the hysteresis of heat conduction, respectively. These factors not only retain the core information of the original physical process but are also embedded into the hot spot risk assessment model in a unified dimensional form, enhancing the model's interpretability and response sensitivity. This serves as a crucial intermediate layer for achieving risk prediction and precise control of the reaction process.

[0093] The specific steps for generating the hot spot temperature characteristic factor, hot spot bubble density characteristic factor, and hot spot thermal retention characteristic factor are as follows:

[0094] Within a preset time window, the temperature change sequence, bubble recognition image sequence, and heat flux distribution data of each hot spot spatial label are extracted within the three-dimensional coordinate range. The temperature fluctuation amplitude curve, the bubble concentration change frequency curve per unit area, and the heat retention time sequence per unit volume are calculated respectively.

[0095] The acquired raw data sequence is numerically normalized and uniformly converted into a dimensionless form to eliminate scale differences between different physical quantities and ensure that the parameters are comparable in the subsequent modeling process.

[0096] Based on the normalization results, the root mean square value of temperature fluctuation amplitude is extracted as the hot spot temperature feature factor, the maximum frequency peak value of bubble concentration change frequency is extracted as the hot spot bubble density feature factor, and the average duration of heat retention time series is extracted as the hot spot heat retention feature factor. The three together constitute the multidimensional feature expression of the hot spot region, which is used to support the quantitative input of the risk assessment model.

[0097] To achieve this feature factor extraction process, after normalizing the temperature fluctuation sequence, bubble density change frequency sequence, and heat retention time sequence, quantitative feature extraction is performed on each type of sequence. First, the root mean square value of the normalized temperature fluctuation amplitude sequence is calculated; that is, the square of the fluctuation value at each time point is averaged and the square root is taken to obtain the overall thermal disturbance intensity index of the hotspot region within the time window, which serves as the hotspot temperature feature factor. Second, spectral analysis is performed on the normalized bubble density change frequency sequence. The dominant frequency distribution is extracted using Fast Fourier Transform, and the peak with the highest frequency intensity is selected as the strongest disturbance index of the dynamic aggregation behavior of bubbles in the region, forming the hotspot bubble density feature factor. Finally, the average duration within the time window of the normalized heat retention time sequence is calculated to measure the stability and conduction hysteresis of heat accumulation in the region, serving as the hotspot heat retention feature factor. These three feature factors are combined into a hotspot feature parameter vector after unified dimension processing, serving as the input to the subsequent hotspot migration risk modeling function, enabling a quantitative evaluation of hotspot stability and diffusion risk.

[0098] The specific steps for extracting the temperature change sequence, bubble recognition image sequence, and heat flux distribution data at the corresponding location within a preset time window, around the three-dimensional coordinate range of each hot spot spatial label, and calculating the temperature fluctuation amplitude curve, the bubble concentration change frequency curve per unit area, and the heat retention time sequence per unit volume are as follows:

[0099] Based on the three-dimensional coordinate position recorded in each hot spot spatial label, the temperature data corresponding to the spatial region within the entire preset time window is extracted from the three-dimensional thermal distribution dynamic matrix, and a continuous temperature time series is constructed.

[0100] Synchronously call the bubble recognition image sequence obtained by the image recognition system within the same spatial coordinate range, count the number of bubbles per unit area in each frame image, and calculate the bubble density change frequency curve per unit time.

[0101] Based on the local heat conduction velocity and direction data in the heat flux vector model, the cumulative heat flux residence time per unit volume in the hot spot spatial region is calculated, and a continuous heat residence time series is constructed to provide an accurate data foundation for subsequent feature factor extraction and risk modeling.

[0102] To achieve the three-dimensional coordinate range around each hotspot spatial label, extract the temperature change sequence, bubble recognition image sequence, and heat flux distribution data of its corresponding location within a preset time window, and calculate three types of feature curves, a unified spatiotemporal indexing mechanism and data processing workflow are required. First, using the three-dimensional coordinate index recorded in the hotspot spatial label, the corresponding spatial location of the hotspot in the heat distribution matrix, image acquisition results, and heat flux model is located. Second, based on the time window parameters set by the thermal control system, the temperature value of the hotspot region within multiple consecutive sampling periods is extracted from the temperature acquisition data, the deviation amplitude of each time point relative to the average value is calculated, and a temperature fluctuation amplitude curve is constructed. Simultaneously, image frames acquired within the corresponding time period at this spatial location are extracted, bubble recognition processing is performed on each image frame, the number of bubbles per unit area is counted, and the frequency change curve in the time dimension is calculated. Finally, the heat flux intensity of the region at each time node is extracted from the heat flux vector model, the heat retention state in the region is determined, the retention time is accumulated and formed into a time series, ultimately generating a sequence of heat retention duration per unit volume. Through the above operations, the thermal dynamics of the hot spot region in the spatiotemporal dimension can be transformed into quantitatively comparable analytical data, providing solid data support for subsequent risk factor calculations and thermal control response decisions.

[0103] The generated feature factors are combined into a hot spot feature parameter vector, and the hot spot feature parameter vector is input into the hot spot migration risk modeling function established based on historical hot spot instability events. The hot spot migration risk index is calculated by using a multi-index fusion algorithm and preset empirical weight coefficients.

[0104] This step aims to quantitatively assess and predict the potential migration risk of hot spots, constructing an intelligent risk decision-making mechanism driven by multi-source feature factors and possessing nonlinear expression capabilities. This provides precise and real-time intervention basis for the thermal control system in the sodium methoxide reaction process. In the complex reaction system intertwined with intense exothermic reactions and bubble generation, the formation, evolution, and migration of hot spots are not driven by a single factor, but are the result of the combined effects of temperature perturbations, bubble density changes, and heat retention behavior. Therefore, relying solely on a single type of parameter for risk assessment often carries the risk of delayed response, inaccurate judgment, or missed detection. To overcome this problem, this step unifies the previously extracted hot spot temperature feature factors, bubble density feature factors, and heat retention feature factors into a structured hot spot feature parameter vector, achieving a standardized expression of multi-dimensional dynamic features. Subsequently, this vector is input into the hot spot migration risk modeling function constructed based on historical hot spot instability events. The empirical weight coefficients determined during the model training phase are invoked to assign differentiated influence strengths to different feature factors, reflecting their actual contribution to hot spot instability under historical operating conditions. This weighting mechanism enhances the model's ability to perceive real-world risk trends.

[0105] In the fusion phase, the system employs a multi-index fusion algorithm to nonlinearly combine the weighted feature factors. This algorithm amplifies the numerical representation of high-risk intervals by exponentially calculating the risk intensity of each factor and utilizes a polynomial structure to simulate the cross-coupling effect between factors, thereby constructing a nonlinear response model that better reflects hotspot behavior under actual operating conditions. Ultimately, the model outputs a normalized hotspot migration risk index, which not only provides an immediate quantitative evaluation of the current hotspot state but also predicts its short-term evolution trend. The magnitude of the risk index directly serves as the decision-making basis for whether the system initiates a control response strategy, enabling the entire thermal control system to shift from "static threshold judgment" to "dynamic prediction and early warning," significantly improving the initiative in safety assurance and the intelligence level of control operations. Therefore, this step plays a crucial role in the entire thermal control process, serving as a key computational link for achieving refined risk control and green, safe production.

[0106] The steps for calculating the hot spot migration risk index are as follows:

[0107] The hot spot temperature feature factor, hot spot bubble density feature factor, and hot spot thermal retention feature factor corresponding to each hot spot spatial label are assembled into a hot spot feature parameter vector in a predetermined order, and this parameter vector is incorporated into a unified data structure for input into the model.

[0108] Assembling hot spot feature parameter vectors in a predetermined order refers to arranging hot spot temperature feature factors, hot spot bubble density feature factors, and hot spot thermal retention feature factors into numerical vectors with fixed dimensions and semantic meaning according to a pre-defined data structure and positional order. The specific implementation is as follows: First, the arrangement structure of the hot spot feature parameter vectors is predefined in the thermal control modeling system. For example, the first position is fixed as the temperature feature factor, the second position as the bubble density feature factor, and the third position as the thermal retention feature factor. Second, for each hot spot spatial label, the three normalized and calculated feature factor values ​​corresponding to that hot spot are called and filled into the various positions of the three-dimensional vector in a strictly predetermined order, forming a hot spot feature parameter vector with a one-to-one correspondence. Finally, this parameter vector is stored in a standardized input dataset and bound to the corresponding hot spot label to ensure that semantic information and numerical meaning can be correctly transmitted in the subsequent risk modeling process, thereby providing a consistent, standardized, and interpretable model input format for risk index calculation.

[0109] The hot spot feature parameter vector is input into the hot spot migration risk modeling function constructed based on historical hot spot instability events. The pre-trained empirical weight coefficients are called in the hot spot migration risk modeling function to assign different importance levels to the three feature factors to reflect the differentiated influence of each factor on the hot spot migration trend.

[0110] The hot spot migration risk modeling function, constructed based on historical hot spot instability events, utilizes typical case data of hot spot migration, diffusion, or system instability that have actually occurred in past reactions. It extracts key influencing factors and hot spot evolution patterns, establishing a predictive mathematical model to assess the probability of current hot spot migration or runaway. This function performs multi-dimensional feature analysis on parameters such as temperature fluctuation intensity, bubble concentration dynamics, and heat residence time in historical hot spot samples, and models their correlation with the time, location, and evolution trend of actual thermal instability events. This allows for the understanding of the influence weights and changing patterns of various characteristic factors on system stability.

[0111] During the modeling process, the system uses machine learning, multivariate regression, or fuzzy inference to train the model with a large amount of historical hotspot data, extracting decision rules or response functions that can be used to predict future hotspot behavior. The resulting risk modeling function can receive the feature parameter vector of each current hotspot as input and automatically calculate the risk level of the hotspot's migration or expansion within a given time scale. This information is then provided to the control system to determine whether to initiate thermal control measures such as adjusting the stirring method, enhancing the cooling strategy, or intervening in the feeding rhythm. The core value of this function lies in transforming historical experience into a predictive mechanism, realizing intelligent thermal safety management from "post-event response" to "pre-event prediction."

[0112] The pre-trained empirical weight coefficients refer to the optimal weighting coefficients determined by a fitting algorithm during model building, based on a quantitative assessment of the importance of each input feature factor using historical sample data. This training process is typically implemented using least mean square error fitting, maximum likelihood estimation, or gradient optimization algorithms, ultimately yielding a set of differentiated weight parameters reflecting the influence of temperature, bubble, and heat retention factors on hot spot migration. For example, if historical data shows that heat retention time is more sensitive to the risk of hot spot spread, the model will automatically assign higher weights to the heat retention feature factor. These empirical weights are invoked during model inference to perform weighted calculations on each hot spot feature parameter vector, thereby improving the accuracy and discriminative power of risk assessment.

[0113] A multi-index fusion algorithm is used to nonlinearly combine the weighted feature factors and output a unique hot spot migration risk index. This index is used to quantify the possibility of the current hot spot region spreading, proliferating, or becoming unstable, and serves as the core criterion for triggering subsequent thermal control response strategies.

[0114] The multinomial index fusion algorithm is a numerical calculation method for nonlinearly combining multiple weighted feature factors. Essentially, it expands and transforms feature factors by introducing exponential functions, enhancing the expression of differences, and models the interaction relationships among factors through a polynomial structure to output a comprehensive index with risk indication significance. In implementation, each feature factor is first multiplied by its corresponding empirical weight coefficient to form a weighted feature value. Then, each weighted feature value is exponentially calculated to amplify its numerical influence in the high-risk range, resulting in three exponential terms. Next, these exponential terms are added in polynomial form and combined with cross-product terms to form a fusion function, which comprehensively reflects the nonlinear correlation and superposition effect of the three factors. Finally, the value output by the fusion function is normalized and mapped to a standard risk index range as the quantitative result of the current hotspot migration risk. This algorithm can fully express the dynamic instability trend driven by multiple factors under complex operating conditions, improving the sensitivity and accuracy of risk identification.

[0115] When the hot spot migration risk index exceeds the set safety threshold, based on the current spatial location and migration trend of the hot spot, the forward and reverse rotation direction of the stirring paddle is periodically switched, and intermittent pulsating stirring operation is implemented to break the local bubble aggregation state, promote the uniform dispersion and rapid breakage of bubbles, and increase the local heat diffusion rate, thereby inhibiting the generation and migration of hot spots and achieving precise control of the heat release behavior of the reaction.

[0116] The purpose of this step is to immediately respond to the spatial location and migration trend of the hot spot when the system detects that the hot spot migration risk index has exceeded the set safety threshold. By dynamically adjusting the stirring method, it actively intervenes in the local fluid structure and thermal distribution of the reaction system, thereby effectively suppressing the formation and expansion of the hot spot. Hot spots are usually caused by the combined effects of intense heat release, bubble aggregation, and heat retention. They are characterized by abnormally high local temperatures and obstructed heat diffusion, which can easily lead to local reaction instability or even systemic thermal runaway. To address this, this step utilizes the three-dimensional coordinate information and migration direction provided by the hot spot spatial tag to apply periodic rotation switching operations to the stirring system. By rapidly switching between forward and reverse rotation, it disrupts the ring-shaped bubble chains, shear accumulation zones, and heat transfer dead zones that may form in the liquid, thus disrupting the bubble aggregation pattern. At the same time, a pulsating stirring mechanism is introduced. By dynamically adjusting the rotation speed of the stirring paddle, it induces changes in the local flow field intensity, causing the bubbles to be subjected to nonlinear disturbance forces and rapidly break up and disperse, thereby improving the smoothness of the heat conduction path and the diffusion speed in the local area and reducing the heat retention time. This control method not only enables early intervention before hot spots develop into systemic thermal instability, but also avoids overcooling or waste of raw materials, achieving refined and targeted energy management and effectively improving the overall safety, thermal control efficiency and process stability of the sodium methoxide reaction process.

[0117] When the hot spot migration risk index exceeds the set safety threshold, based on the current spatial location and migration trend of the hot spot, the specific steps for periodically switching the forward and reverse rotation directions of the agitator and implementing intermittent pulsating stirring operation are as follows:

[0118] Based on the hot spot migration risk index corresponding to each hot spot spatial label and the preset hot spot migration risk index reference threshold, the disturbance response frequency adjustment coefficient for adjusting the stirring response frequency is calculated, and the calculation expression is as follows:

[0119] ,

[0120] In the formula, R d It is a hotspot migration risk index, representing the risk intensity corresponding to a certain hotspot spatial label at the current moment. It is a comprehensive judgment value of the system regarding the possibility of hotspot migration, diffusion, or instability. R t This is the reference threshold for the hot spot migration risk index, representing a critical instability reference value obtained based on historical operating conditions and empirical modeling. It serves as the baseline used by the system to measure the risk level of hot spots. l f It is the disturbance response frequency adjustment coefficient, used to dynamically adjust the disturbance frequency and rhythm of the stirring system. It is the core driving parameter for subsequent rotation direction switching and pulse stirring strategies. As a unified disturbance frequency adjustment benchmark, it runs through the two major processes of stirring direction switching and amplitude modulation, ensuring that the response intensity is accurately matched with the hot spot risk level. R d < R t hour, l f ≈0, the disturbance response tends to be mild; when R d > R t hour, l f The temperature rises sharply, and the stirring system enters a high-frequency response state.

[0121] The purpose of this step is to dynamically adjust the response frequency of the thermal control system based on the intensity of hot spot risk, thereby improving the ability to intervene in high-risk hot spot areas in a timely manner and avoiding excessive stirring in low-risk conditions.

[0122] Obtaining the disturbance response frequency adjustment coefficient l f Then, a periodic signal is generated to control the rotation direction of the agitator, which drives the agitator to switch between forward and reverse rotation at a dynamic frequency. The generation formula is as follows:

[0123] ,

[0124] In the formula, i ( t This is the stirring direction control signal. An output value of +1 indicates that the stirring paddle rotates in the forward direction, and -1 indicates that the stirring paddle rotates in the reverse direction. It directly affects the drive control logic of the stirring motor to dynamically switch the rotation direction. By periodically switching the rotation direction, it helps to break the unidirectional flow structure and bubble chain aggregation path formed in the reaction liquid, thereby promoting bubble dispersion, improving heat transfer uniformity, and suppressing hot spots. t It is a time variable. f It is the initial phase angle, which controls the starting point of the sine wave and prevents all stirring devices from switching at the same time, which could lead to resonance or disturbance imbalance. sign It is a sign determination function;

[0125] In thermal control systems, sign determination functions sign The main function is to convert continuously changing sinusoidal signals into discrete control commands with clear execution meanings, used to control the rotation direction of the agitator. Since the sinusoidal function oscillates continuously between -1 and +1, it itself does not possess a directly executable physical meaning. The `sign` function extracts its positive and negative values, quickly converting them to +1 or -1, thus clearly indicating whether the agitator should rotate forward or backward. This method not only enables periodic switching of the agitator's rotation direction, breaking the local flow structure formed by bubble aggregation, but also avoids complex logical judgments through the simplicity of mathematical expression and computational efficiency, giving the system real-time and efficient direction control capabilities under high-frequency thermal disturbance environments. Therefore, sign Functions serve as a bridge in stirring control, transforming hot spot risk signals into physical actions, and are an indispensable basic mathematical tool for constructing intelligent disturbance strategies.

[0126] This step is used to disrupt the local bubble aggregation structure through periodic perturbation, thereby improving the reconstruction efficiency of the micro-flow of the reaction liquid and enhancing the bubble shearing and deagglomeration effect.

[0127] Combined with disturbance response frequency adjustment coefficient l f With stirring direction control signal i ( t The rotational speed of the agitator is adjusted in real time to achieve pulsating agitation. The adjustment formula is as follows:

[0128] ,

[0129] In the formula, A ( t The speed range of the agitator at the current time is used as the final control command for pulsating agitation. Through dynamic adjustment over time, it promotes the breaking up of local bubbles, improves heat transfer efficiency, and avoids heat retention. A0 This is the reference stirring speed, representing the basic operating speed of the stirring paddle under stable operation without thermal control intervention. α This is the pulsation enhancement coefficient, representing the maximum allowable relative increase ratio when the thermal control system applies pulsating disturbances. Adjusting the strength of the pulsation amplitude determines the allowable stirring speed during dynamic disturbances. A 0 The range.

[0130] The speed regulation process creates an alternating perturbation energy field, which helps to promote the local breakup and rapid diffusion of bubbles, thereby enhancing the heat release efficiency, inhibiting the continuous growth and spatial migration of hot spots, and improving the thermal stability of the entire reaction system.

[0131] The aforementioned sodium methoxide production method based on reaction heat regulation enables multi-dimensional perception, intelligent identification, and dynamic intervention of local thermal imbalances and bubble aggregation behavior during the reaction process. This effectively addresses key safety hazards in traditional processes, such as the difficulty in real-time detection of hot spots, uneven heat release, and bubbles shielding heat conduction paths. This method, through deep integration of three-dimensional thermal field modeling and bubble image analysis, achieves spatial mapping identification of multi-source data on "temperature-bubble-heat flux" for the first time, accurately locating potential hot spot regions. Furthermore, it constructs a risk index using multi-factor feature parameters to predict hot spot migration trends. Then, when the risk index exceeds a threshold, a combined mechanism of periodic directional perturbation and pulsating stirring is used to actively disrupt bubble aggregation structures, enhance heat diffusion rates, and effectively suppress the generation and expansion of hot spots. Overall, this method significantly improves the thermal control accuracy and safety stability of the sodium methoxide reaction process, providing a predictable, interventionable, and sustainable intelligent thermal management solution for the industrialization of high-risk, strongly exothermic reactions.

[0132] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0133] The foregoing has described certain exemplary embodiments of the invention by way of illustration only. Undoubtedly, in this text, the use of relational terms such as "first" and "second" is merely to distinguish one entity or operation from another, and does not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0134] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0135] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0136] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0137] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0138] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0139] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0140] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for producing sodium methoxide based on reaction heat regulation, characterized in that, Includes the following steps: The temperature of each spatial measuring point inside the reactor is continuously collected, and time-series gradient difference calculation is performed to obtain the instantaneous heat flux direction and heat flux intensity between adjacent measuring points, thereby constructing a three-dimensional dynamic heat distribution matrix. Based on the three-dimensional dynamic thermal distribution matrix, spatial regions with high temperature fluctuation frequency and discontinuous heat flux direction are identified as high-frequency thermal disturbance regions; at the same time, bubble images of the surface and interior of the reaction medium are acquired to identify regions with abnormally high local bubble concentration. The high-frequency thermal disturbance region and the bubble concentration abnormal region are spatially mapped. Regions that simultaneously exhibit severe temperature fluctuations and high-density bubble distribution characteristics in the same spatial location are extracted, marked as potential hot spot regions, and hot spot spatial labels are established for each potential hot spot region. For each hot spot spatial label, the temperature fluctuation amplitude sequence, the bubble concentration change frequency sequence per unit area, and the heat retention duration per unit volume within a preset time window are statistically analyzed and standardized to generate hot spot temperature characteristic factor, hot spot bubble density characteristic factor, and hot spot heat retention characteristic factor, respectively. The specific steps for generating the hot spot temperature characteristic factor, hot spot bubble density characteristic factor, and hot spot thermal retention characteristic factor are as follows: Within a preset time window, the temperature change sequence, bubble recognition image sequence, and heat flux distribution data of each hot spot spatial label are extracted within the three-dimensional coordinate range. The temperature fluctuation amplitude curve, the bubble concentration change frequency curve per unit area, and the heat retention time sequence per unit volume are calculated respectively. The acquired raw data sequence is numerically normalized to ensure that the parameters are comparable in the subsequent modeling process; Based on the normalization results, the root mean square value of temperature fluctuation amplitude is extracted as the hot spot temperature feature factor, the maximum frequency peak value of bubble concentration change frequency is extracted as the hot spot bubble density feature factor, and the average duration of heat retention time series is extracted as the hot spot heat retention feature factor. The three together constitute the multidimensional feature expression of the hot spot region. The specific steps for extracting the temperature change sequence, bubble recognition image sequence, and heat flux distribution data at the corresponding location within a preset time window, around the three-dimensional coordinate range of each hot spot spatial label, and calculating the temperature fluctuation amplitude curve, the bubble concentration change frequency curve per unit area, and the heat retention time sequence per unit volume are as follows: Based on the three-dimensional coordinate position recorded in each hot spot spatial label, the temperature data corresponding to the spatial region within the entire preset time window is extracted from the three-dimensional thermal distribution dynamic matrix, and a continuous temperature time series is constructed. Synchronously call the bubble recognition image sequence obtained by the image recognition system within the same spatial coordinate range, count the number of bubbles per unit area in each frame image, and calculate the bubble density change frequency curve per unit time. Based on the local heat conduction velocity and direction data in the heat flux vector model, the cumulative heat flux residence time per unit volume in the hot spot spatial region is calculated, and a continuous heat residence time series is constructed to provide an accurate data foundation for subsequent feature factor extraction and risk modeling. The generated feature factors are combined into a hot spot feature parameter vector, and the hot spot feature parameter vector is input into the hot spot migration risk modeling function established based on historical hot spot instability events to calculate the hot spot migration risk index. When the hot spot migration risk index exceeds the set safety threshold, the direction of the stirring paddle is periodically switched between forward and reverse rotation based on the current spatial location and migration trend of the hot spot, and intermittent pulsating stirring operation is implemented to break the local bubble aggregation state and promote the uniform dispersion and rapid breakup of bubbles.

2. The method for producing sodium methoxide based on reaction heat regulation according to claim 1, characterized in that, The steps for constructing a three-dimensional dynamic matrix of heat distribution covering the entire reaction space are as follows: Before the reaction begins, a fixed sampling period is preset, and multiple sets of thermocouple measuring points are evenly distributed in different spatial locations inside the reactor. The real-time temperature values ​​of each measuring point are obtained through a high-frequency synchronous acquisition system in each sampling period to establish a complete time series temperature dataset. Based on continuously acquired time-series temperature data, frame-by-frame temporal gradient difference calculation is performed to analyze the temperature change trend between adjacent time nodes. For each measuring point, an instantaneous heat flux direction vector and heat flux intensity vector model are constructed in its neighboring direction to reflect local heat transfer behavior. Spatial location matching and data fusion are performed on the heat flux vector models of all measuring points. The heat flow information of discrete measuring points is continuously reconstructed through a three-dimensional interpolation algorithm, thereby forming a three-dimensional dynamic heat distribution matrix that can reflect the heat distribution and change law inside the reactor in real time.

3. The method for producing sodium methoxide based on reaction heat regulation according to claim 2, characterized in that, The steps for identifying high-frequency thermal disturbance regions are as follows: Based on the continuous temperature change data of each spatial grid in the three-dimensional thermal distribution dynamic matrix, the temperature fluctuation frequency curve of each grid within a preset time window is extracted. By performing Fourier spectrum analysis on the temperature fluctuation frequency of each grid, spatial regions with a high frequency component ratio higher than a set threshold are extracted, and the degree of variation of the heat flux direction vector in the current region is calculated. The spatial regions that simultaneously satisfy the condition that the temperature fluctuation frequency is higher than the set value and the heat flux direction vector is highly discontinuous in space are identified and marked as high-frequency thermal disturbance regions.

4. The method for producing sodium methoxide based on reaction heat regulation according to claim 3, characterized in that, The steps to identify areas with abnormal bubble concentration are as follows: During the reaction, a sequence of dynamic images of bubbles on the surface and inside the reaction medium is continuously acquired, and grayscale normalization and edge enhancement preprocessing are performed on each frame of the image. The contour boundaries of each bubble are identified by an image edge reconstruction algorithm, and the bubble area, position and morphological features are extracted to generate a bubble distribution feature map for each frame of the image. Based on morphological clustering algorithms, bubble features in continuous images are clustered and analyzed to identify spatial regions where the bubble density is significantly higher than that of the surrounding area per unit time and per unit area. These regions are identified as areas with abnormally high bubble concentration and are used for spatial correlation and fusion analysis with high-frequency thermal disturbance regions.

5. The method for producing sodium methoxide based on reaction heat regulation according to claim 4, characterized in that, The steps for constructing potential hot spot regions and establishing hot spot spatial labels are as follows: Based on the high-frequency thermal disturbance regions identified in the three-dimensional thermal distribution dynamic matrix, their spatial location information in the three-dimensional coordinate system is extracted, and the spatial coordinate set of the abnormal bubble concentration regions obtained from the bubble image clustering analysis is acquired simultaneously. One-to-one spatial matching was performed on the two types of spatial data. Overlapping voxel analysis was used to determine whether there were both drastic temperature fluctuations and high-density bubble distribution characteristics in the same three-dimensional coordinate range. Overlapping regions that met both conditions were then selected. Each overlapping spatial region that meets the conditions is defined as a potential hot spot region, and a unique hot spot spatial label is assigned to it. The label contains a spatial coordinate index, an identification timestamp, and a corresponding perturbation intensity factor.

6. The method for producing sodium methoxide based on reaction heat regulation according to claim 5, characterized in that, The steps for calculating the hot spot migration risk index are as follows: The hot spot temperature feature factor, hot spot bubble density feature factor, and hot spot thermal retention feature factor corresponding to each hot spot spatial label are assembled into a hot spot feature parameter vector in a predetermined order. The hot spot feature parameter vector is input into the hot spot migration risk modeling function constructed based on historical hot spot instability events. The pre-trained empirical weight coefficients are called in the hot spot migration risk modeling function to assign different importance levels to the three feature factors to reflect the differentiated influence of each factor on the hot spot migration trend. A multi-index fusion algorithm is used to nonlinearly combine the weighted feature factors to output a unique hot spot migration risk index.

7. The method for producing sodium methoxide based on reaction heat regulation according to claim 6, characterized in that, When the hot spot migration risk index exceeds the set safety threshold, based on the current spatial location and migration trend of the hot spot, the specific steps for periodically switching the forward and reverse rotation directions of the agitator and implementing intermittent pulsating stirring operation are as follows: Based on the hot spot migration risk index corresponding to each hot spot spatial label and the preset hot spot migration risk index reference threshold, the disturbance response frequency adjustment coefficient for adjusting the stirring response frequency is calculated, and the calculation expression is as follows: ， In the formula, R d It is a hot spot migration risk index. R t This is the reference threshold for the hot spot migration risk index. λ f It is the disturbance response frequency adjustment coefficient; Obtaining the disturbance response frequency adjustment coefficient λ f Then, a periodic signal is generated to control the rotation direction of the agitator, which drives the agitator to switch between forward and reverse rotation at a dynamic frequency. The generation formula is as follows: ， In the formula, θ ( t () is the stirring direction control signal. t It is a time variable. φ It is the initial phase angle. sign It is a sign determination function; Combined with disturbance response frequency adjustment coefficient λ f With stirring direction control signal θ ( t The rotational speed of the agitator is adjusted in real time to achieve pulsating agitation. The adjustment formula is as follows: ， In the formula, A ( t () represents the rotational speed range of the agitator at the current time. A 0 This is the reference stirring speed. α It is the pulsation enhancement coefficient.

Images