Intelligent monitoring and regulation method and device for polymerization reaction based on spectral analysis

Abstract

The application discloses a method and device for intelligent monitoring and regulation of polymerization reaction based on spectral analysis, and relates to the technical field of intelligent monitoring.The method comprises the following steps: constructing a particle size-track contrast feature library based on a prior polymerization reaction knowledge base, combining a track feature vector prior threshold of different particle size intervals, decoupling a mixed spectral signal of a current polymerization reaction, and outputting spectral signals and track feature vectors of different particle size intervals; comparing the track feature vectors with the prior threshold, performing abnormal deviation analysis, and triggering a corresponding directional regulation strategy according to an analysis result.The application solves the technical problems that the existing polymerization reaction monitoring method has poor real-time performance, low precision, and cannot effectively cope with abnormal fluctuations in the reaction process, achieves intelligent monitoring and signal decoupling based on spectral analysis, realizes real-time and accurate regulation of the polymerization reaction process, and thus improves the product quality and production efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent monitoring, and particularly relates to a method and device for intelligent monitoring and regulation of polymerization reaction based on spectral analysis. BACKGROUND

[0002] In the process of polymerization reaction, accurate monitoring and regulation of reaction conditions are crucial for controlling product quality and improving production efficiency. Traditional monitoring methods mainly rely on periodic sampling and laboratory analysis, which have problems such as slow response speed and large error. In recent years, online monitoring technology based on spectral analysis has been gradually applied in polymerization reaction. By obtaining spectral signals in real time during the reaction process, the changes in the reaction system can be dynamically reflected. However, the spectral signal contains mixed information of multiple components, making it difficult to accurately identify the characteristics of particles in different size intervals, and still unable to achieve accurate control of the polymerization reaction process. SUMMARY

[0003] The present application provides a method and device for intelligent monitoring and regulation of polymerization reaction based on spectral analysis, which solves the technical problems of poor real-time performance, low precision and inability to effectively respond to abnormal fluctuations in the reaction process of existing polymerization reaction monitoring methods.

[0004] In a first aspect, the present application provides a method for intelligent monitoring and regulation of polymerization reaction based on spectral analysis, which comprises: constructing a particle size-trace contrast feature library based on a prior polymerization reaction knowledge base, the particle size-trace contrast feature library comprising prior threshold values of trace feature vectors corresponding to a first particle size interval and a second particle size interval in the polymerization reaction; obtaining a mixed spectral signal of the current polymerization reaction, decomposing the mixed spectral signal by a signal decoupler, outputting a first spectral signal based on the first particle size interval and a second spectral signal based on the second particle size interval, and identifying first and second trace feature vectors of the first and second particle size intervals; calling the prior threshold values of the trace feature vectors corresponding to the particle size-trace contrast feature library to perform abnormal deviation analysis on the first and second trace feature vectors, and triggering a directional regulation strategy according to the abnormal deviation analysis result.

[0005] In a second aspect of the present application, an intelligent monitoring and regulation device for polymerization reaction based on spectral analysis is provided, which comprises: a feature library construction module configured to construct a particle size-trace contrast feature library based on a prior polymerization reaction knowledge base, the particle size-trace contrast feature library comprising a trace feature vector prior threshold corresponding to a first particle size interval and a second particle size interval in the polymerization reaction; a trace feature identification module configured to obtain a mixed spectral signal of a current polymerization reaction, decompose the mixed spectral signal through a signal decoupler, output a first spectral signal based on the first particle size interval and a second spectral signal based on the second particle size interval, and identify a first trace feature vector and a second trace feature vector of the first particle size interval and the second particle size interval; and an abnormal deviation analysis module configured to perform abnormal deviation analysis on the first trace feature vector and the second trace feature vector by calling the trace feature vector prior threshold corresponding to the particle size-trace contrast feature library, and trigger a directional regulation strategy according to an abnormal deviation analysis result.

[0006] The one or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0007] The intelligent monitoring and regulation method and device for polymerization reaction based on spectral analysis provided in the present application are related to the technical field of intelligent monitoring, which realizes precise monitoring and regulation of the polymerization reaction process by constructing a particle size-trace contrast feature library, extracting spectral signals and trace feature vectors of different particle size intervals from a mixed spectral signal, and comparing and analyzing the spectral signals and the trace feature vectors with prior thresholds to identify abnormal deviation of the trace features and trigger corresponding directional regulation strategies, thereby solving the technical problems of poor real-time performance, low precision, and inability to effectively respond to abnormal fluctuations in the reaction process of the existing polymerization reaction monitoring methods, realizing intelligent monitoring and signal decoupling based on spectral analysis, and realizing real-time and precise regulation of the polymerization reaction process, thereby achieving the technical effects of improving product quality and production efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0009] Figure 1 A flowchart of the intelligent monitoring and regulation method for polymerization reaction based on spectral analysis provided in the embodiments of the present application is shown in the figure.

[0010] Figure 2 A structure diagram of the intelligent monitoring and regulation device for polymerization reaction based on spectral analysis provided in the embodiments of the present application is shown in the figure.

[0011] Explanation of reference signs: feature library construction module 11, trajectory feature identification module 12, abnormal deviation analysis module 13. DETAILED DESCRIPTION

[0012] The application provides a polymerization reaction intelligent monitoring and regulation method and device based on spectrum analysis, and aims to solve the technical problems that the existing polymerization reaction monitoring method has poor real-time performance, low precision, and cannot effectively respond to abnormal fluctuations in the reaction process.

[0013] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative work fall within the protection scope of the application.

[0014] It should be noted that the terms "first", "second" and the like in the specification and the above drawings of the application are used to distinguish similar objects, and do not necessarily indicate a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the application described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, device, product or server including a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or modules not clearly listed or inherent to these processes, methods, products or devices.

[0015] Embodiment one, as shown in the application, a polymerization reaction intelligent monitoring and regulation method based on spectrum analysis is provided, which comprises: Figure 1

[0016] P10: constructing a particle size-trjectory contrast feature library based on a prior polymerization reaction knowledge base, wherein the particle size-trjectory contrast feature library comprises a prior threshold value of a trajectory feature vector corresponding to a first particle size interval and a second particle size interval in the polymerization reaction.

[0017] Specifically, constructing the particle size-trjectory contrast feature library is a basic link to realize the polymerization reaction intelligent monitoring and regulation. The construction of the feature library is based on the prior polymerization reaction knowledge base. The prior polymerization reaction knowledge base refers to a knowledge system about the relationship between various parameters in the polymerization reaction process and the reaction results accumulated through a large amount of experimental data, theoretical research and historical production experience. These knowledge provides a theoretical support and data basis for constructing the feature library.

[0018] ​Firstly, the construction of the particle size-trace feature library relies on the in-depth analysis of particle size and trace in the polymerization process. Particle size generally refers to the size of particles in the reaction, which directly affects the speed of the polymerization reaction, the distribution of reactants, and the quality of the product. The trace feature refers to the movement path or change rule of particles within a certain time in the reaction, which can reflect the dynamic characteristics of particle behavior in the reaction process.

[0019] In the polymerization reaction, the particle size of particles has an important influence on the progress of the reaction and the performance of the final product. Therefore, the feature library divides the particles in the polymerization reaction into different intervals according to the particle size, such as the first particle size interval and the second particle size interval. This division allows the feature library to set prior threshold values for the trace feature vectors for particles in different particle size intervals, thereby achieving accurate monitoring and control of different particle sizes in the polymerization reaction. Specifically, the first particle size interval and the second particle size interval refer to particles in different particle size ranges, and the behavior of these particles in the polymerization reaction may differ significantly. For example, large particles may affect heat conduction due to their larger volume, while small particles may have a faster reaction rate. In order to accurately control and adjust the reaction process, the trace features of particles of different sizes need to be analyzed separately, and the corresponding trace feature vectors need to be established.

[0020] In the process of constructing the feature library, a large amount of polymerization reaction experimental data needs to be analyzed and processed. These data include but are not limited to parameters such as temperature, pressure, reaction time, and raw material ratio of the polymerization reaction, as well as the particle size distribution and trace features of particles under these parameters. Through statistical analysis and model fitting of these data, the normal range of the trace feature vector of particles in different particle size intervals, i.e. the trace feature vector prior threshold value, can be determined. The trace feature vector prior threshold value is used to describe the normal range of the particle trace. For example, within a certain particle size interval, the trace feature of the particle should remain within a certain fluctuation range, and once the trace feature of the particle deviates from this range, it can be considered as abnormal behavior. These thresholds provide important basis for subsequent spectral signal analysis, which can help the system to judge whether the trace of the particle deviates from the normal range in the actual reaction process, thereby providing data support for the triggering of the control strategy.

[0021] Further, in step P10 of the embodiments of the present application:

[0022] The particle size-trajectory contrast feature library comprises a first trajectory feature vector prior threshold and a second trajectory feature vector prior threshold corresponding to a first particle size interval and a second particle size interval respectively in the polymerization reaction; the first trajectory feature vector prior threshold comprises a trajectory feature vector prior threshold of a polymerization particle with a particle size in the first particle size interval, and the second trajectory feature vector prior threshold comprises a trajectory feature vector prior threshold of a polymerization particle with a particle size in the second particle size interval; wherein the first particle size interval and the second particle size interval are non-overlapping and the second particle size interval is larger than the first particle size interval.

[0023] It should be understood that the particle size-trajectory contrast feature library comprises two main parts: the first trajectory feature vector prior threshold and the second trajectory feature vector prior threshold. The first trajectory feature vector prior threshold is specifically for polymerization particles with a particle size in the first particle size interval, which are usually active in the initial stage of the polymerization reaction, and their trajectory feature vectors can reflect the dynamic behavior and distribution state of small particle size particles. By setting the first trajectory feature vector prior threshold, the trajectory of particles in the first particle size interval can be accurately monitored to determine whether they deviate from the normal range. For example, in the initial stage of the polymerization reaction, the trajectory feature vector of small particle size particles may exhibit a high diffusion rate and a relatively uniform spatial distribution. By comparing and analyzing with the prior threshold, abnormal behavior of small particle size particles, such as too fast diffusion rate or uneven distribution, can be found in time, thereby providing a basis for subsequent control strategies.

[0024] The second trajectory feature vector prior threshold is for polymerization particles with a particle size in the second particle size interval. The particles in the second particle size interval have a larger particle size than those in the first particle size interval, and these larger particle size particles are more critical in the later stage of the polymerization reaction, and their trajectory feature vectors can reflect the mature stage of the polymerization reaction and the final state of the product. By setting the second trajectory feature vector prior threshold, the trajectory of particles in the second particle size interval can be monitored to ensure that the final product of the polymerization reaction meets the expected quality standards. For example, in the later stage of the polymerization reaction, the trajectory feature vector of large particle size particles may exhibit a low diffusion rate and a relatively concentrated spatial distribution. By comparing and analyzing with the prior threshold, it can be determined whether the trajectory of large particle size particles is normal, thereby ensuring that the final product of the polymerization reaction has an ideal molecular weight distribution and physical properties.

[0025] In constructing the particle size-trace contrast feature library, the first particle size interval and the second particle size interval are explicitly divided into two non-overlapping parts, and the second particle size interval is larger than the first particle size interval. This setting can ensure that each particle can be clearly attributed to a certain particle size interval, thereby avoiding the confusion of particle trace features between different intervals. In the polymerization reaction, the particle size of the particle usually increases gradually with time, therefore, by dividing the particle size interval into two non-overlapping parts, the polymerization behavior at different stages can be more clearly distinguished. The non-overlapping particle size interval ensures that the particles in each particle size interval have a unique trace feature vector, avoiding confusion in monitoring and control due to overlapping particle size intervals, and improving the accuracy of monitoring and the pertinence of control. The size relationship of the particle size interval reflects the natural law of the polymerization reaction, so that the feature library can better adapt to the dynamic changes of the polymerization reaction, and provide more reliable basis for intelligent monitoring and control.

[0026] In practical applications, the construction of the particle size-trace contrast feature library requires analysis and processing of a large amount of polymerization reaction experimental data. These data include parameters such as temperature, pressure, reaction time, and raw material ratio of the polymerization reaction, as well as the particle size distribution and trace feature of the particles under these parameters. Through statistical analysis and model fitting of these data, the normal range of the trace feature vector of particles in different particle size intervals, i.e. the prior threshold, can be determined.

[0027] Further, in step P10 of the embodiments of the present application:

[0028] The particle size-trace contrast feature library further includes a third trace feature vector prior threshold based on a third particle size interval in the polymerization reaction, and the third trace feature vector prior threshold includes a trace feature vector prior threshold of a polymerization reaction particle with a particle size in the third particle size interval; wherein the third particle size interval and the second particle size interval are non-overlapping and the third particle size interval is larger than the second particle size interval.

[0029] Optionally, the particle size-trace contrast feature library not only includes the trace feature vector prior threshold for the first particle size interval and the second particle size interval, but also includes the trace feature vector prior threshold for the third particle size interval. The third particle size interval is mainly used to describe larger particles in the polymerization reaction, which usually exhibit different behaviors and trace features during the reaction process. In order to more accurately monitor the reaction process, the particle size-trace contrast feature library associates the trace feature vector of the particles in the third particle size interval with the prior threshold, thereby providing a more comprehensive basis for subsequent abnormal deviation analysis.

[0030] Specifically, the third particle size interval is non-overlapping with the second particle size interval, and the particle size of the third particle size interval is larger than that of the second particle size interval, meaning that these particles are generally larger than the active growth particles and have different reaction dynamics. This division of particle size intervals ensures the independence of particle trajectory characteristics, allowing the behavior of each particle size interval to be analyzed and optimized separately. For particles in the third particle size interval, they are usually formed due to aggregation or other abnormal phenomena during the reaction process, and these particles can cause instability of the polymerization reaction, thereby affecting the quality of the final product. Therefore, it is particularly important to accurately identify and monitor the behavior of particles in this interval.

[0031] In constructing the particle size-trjectory contrast feature library, for particles in the third particle size interval, the system extracts the motion trajectory characteristics of particles in this particle size range from historical data, such as particle velocity, acceleration, displacement, etc., and determines the corresponding trajectory characteristic vector prior threshold. These prior thresholds are obtained by analyzing historical samples and are used to define the normal fluctuation range of particle trajectories in this particle size range. Once an abnormal deviation of particle trajectory occurs during the reaction process, the system can use these prior thresholds for quick judgment and feedback, triggering the reaction control strategy in a timely manner.

[0032] In practical applications, by monitoring the spectral data of the polymerization reaction system in real time and comparing the monitoring results with the prior thresholds in the particle size-trjectory contrast feature library, abnormal situations during the reaction process, such as abnormal particle aggregation, can be discovered in a timely manner. Once an abnormality is found, the system will automatically trigger the corresponding directional control strategy, such as adding a small amount of polymerization inhibitor, forcibly cooling the reaction system, or increasing the stirring intensity, to correct the deviation and ensure the normal progress of the polymerization reaction.

[0033] Further, in step P10 of the embodiments of the present application:

[0034] The first particle size interval corresponds to the particle size range of nucleation particles in the polymerization reaction, the second particle size interval corresponds to the particle size range of active growth particles, and the third particle size interval corresponds to the particle size range of abnormal aggregation particles.

[0035] Specifically, the division of particle size intervals is used to accurately monitor the behavior of particles during the polymerization reaction to ensure the stability of the reaction and the quality of the product. This step defines three particle size intervals, corresponding to different particle types and stages in the polymerization reaction.

[0036] Firstly, the first particle size interval corresponds to the particle size range of nucleation particles in the polymerization reaction. Nucleation particles are usually the initial particles formed at the beginning of the reaction, with small particle sizes and rapid generation in the early stage of the reaction. These particles have small particle sizes, usually below 50 nanometers. By monitoring the behavior of particles in this particle size interval, the start-up process and nucleation stage of the polymerization reaction can be tracked in real time, providing basic data for the subsequent reaction. If there is deviation or abnormality in the nucleation stage, the reaction conditions can be adjusted in time to ensure the smooth progress of the reaction.

[0037] The second particle size interval corresponds to the particle size range of active growing particles in the polymerization reaction. These particles usually appear in the middle stage of the reaction, with larger particle sizes, generally between 50 and 150 nanometers. Particles in this particle size interval indicate that particles in the reaction are growing through polymerization. The growth rate of particles in this stage directly affects the efficiency of the reaction and the particle size distribution of the final product. Real-time monitoring of the behavior of particles in this particle size interval not only helps to understand the reaction rate, but also provides early warning for possible problems in the reaction process. For example, if the particle growth is too fast or too slow, it may lead to uneven reaction or quality problems of the product, which can be adjusted by adjusting parameters such as temperature, feed rate, etc.

[0038] Finally, the third particle size interval corresponds to the particle size range of abnormal aggregation particles. These particles are usually generated due to abnormal aggregation phenomena in the polymerization reaction process, with particle sizes usually larger than 150 nanometers. The appearance of abnormal aggregation particles indicates that there may be some abnormal factors in the reaction process, such as too high reaction temperature, inappropriate reactant concentration, or external disturbance. The aggregation of these particles may affect the uniformity and quality of the final product, so particles in this particle size interval need special attention. By monitoring this particle size interval, deviations in the reaction process can be detected in time, and appropriate control measures can be taken, such as adjusting reaction conditions or adding appropriate dispersants, to prevent abnormal aggregation phenomena in the reaction process.

[0039] By accurately dividing and real-time monitoring the three particle size intervals, the dynamics of various particles in the polymerization reaction process can be effectively captured, ensuring that the reaction can proceed within the normal range. Once the particle behavior deviates from the preset trajectory, the control strategy can be triggered according to the set threshold to maintain the stability of the reaction.

[0040] Further, based on the prior polymerization reaction knowledge base, a particle size-trace comparison feature library is constructed, and the step P10 of the embodiment of the present application further includes:

[0041] P11: determining polymerization SOP history samples including polymerization reaction type, polymerization reaction formula, and polymerization process parameters; P12: extracting mixed spectrum signal samples and trajectory feature verification data samples of the polymerization SOP history samples, establishing time point association of the mixed spectrum signal samples and the trajectory feature verification data samples, and obtaining a prior polymerization knowledge base; P13: dividing the prior polymerization knowledge base to obtain a trajectory feature vector prior threshold corresponding to a first particle size interval, a trajectory feature vector prior threshold corresponding to a second particle size interval, and a trajectory feature vector prior threshold corresponding to a third particle size interval, and constructing a particle size-trjectory contrast feature library.

[0042] It should be understood that the process of constructing the particle size-trjectory contrast feature library involves extracting key information from the standard operating procedure (SOP) history samples of the polymerization reaction, and using these information to establish a database containing different particle size interval particle trajectory feature vector prior thresholds.

[0043] Firstly, it is necessary to collect and determine the standard operating procedure (SOP) history samples of the polymerization reaction. These samples should include the type, formula, and process parameters of the polymerization reaction. The type of polymerization reaction may include free radical polymerization, ionic polymerization, etc., each of which may produce different particle size distribution and trajectory characteristics. Formula information includes monomer species, initiator, emulsifier, stabilizer, etc., which directly affect the kinetics of polymerization reaction and the particle size distribution of the product. Process parameters such as temperature, pressure, stirring rate, reaction time, etc. will also affect the progress of polymerization reaction and the characteristics of the final product.

[0044] Next, mixed spectrum signal samples and trajectory feature verification data samples are extracted from these determined polymerization SOP history samples. Mixed spectrum signal samples are spectral data collected at different time points during the polymerization reaction, which reflect the particle size distribution and change of particles. Trajectory feature verification data samples are particle trajectory feature data obtained by experiment or other verification methods, used to verify the accuracy of spectral signal analysis results. After extracting these data, the next task is to associate the spectral signal samples and the trajectory feature verification data samples at time points, that is, to synchronize the two kinds of data through time alignment, to ensure that each spectral signal sample and its corresponding trajectory feature verification data sample are matched in time, thereby establishing a prior polymerization knowledge base.

[0045] Then, the established prior polymerization reaction knowledge base is used to further divide the trajectory feature vector prior threshold corresponding to different particle size intervals. These particle size intervals include a first particle size interval, a second particle size interval, and a third particle size interval. The first particle size interval corresponds to the particle size range of nucleation particles, which is usually less than 50 nanometers. The second particle size interval corresponds to the particle size range of active growth particles, which is usually between 50 and 150 nanometers. The third particle size interval corresponds to the particle size range of abnormal aggregation particles, which is usually greater than 150 nanometers. Then, using the particle trajectory data in the historical samples, combined with the characteristics of particles in different particle size intervals, the normal fluctuation range of the trajectory feature vector in each particle size interval is determined. These fluctuation ranges are the trajectory feature vector prior threshold, which reflects the typical behavior of particles in each particle size interval under normal polymerization reaction conditions.

[0046] Then, according to the divided particle size intervals and their corresponding trajectory feature vector prior thresholds, a particle size-trjectory contrast feature library is constructed, which provides a basis for subsequent real-time reaction monitoring and control. In the actual application of polymerization reaction, when the real-time data of spectral signals and particle trajectories deviate from the prior thresholds in the library, the system will automatically identify the abnormality and trigger the corresponding control measures, thereby ensuring the stability of the polymerization reaction process and improving the quality of the final product.

[0047] P20: Obtain a mixed spectral signal of a current polymerization reaction, decompose the mixed spectral signal through a signal decoupler, output a first spectral signal based on the first particle size interval and a second spectral signal based on the second particle size interval, and identify a first trajectory feature vector and a second trajectory feature vector of the first particle size interval and the second particle size interval.

[0048] Further, in step P20 of the embodiments of the present application:

[0049] The signal decoupler is obtained by deep learning network training on the prior polymerization reaction knowledge base. The mixed spectral signal sample of the prior polymerization reaction knowledge base is the input sample, the decoupling output sample of the signal decoupler is obtained, the trajectory feature vector of the decoupling output sample is extracted, the trajectory feature verification data sample is used as the verification label to calculate the loss of the extracted trajectory feature vector output sample, and the loss data is obtained. The decoupling parameters of the signal decoupler are fed back and optimized according to the loss data, and the learning completed signal decoupler is obtained.

[0050] Optionally, the mixed spectrum signal of the current polymerization reaction is obtained and decomposed by the signal decoupler to extract the spectrum signal based on different particle size intervals and the corresponding particle trajectory feature vector. Specifically, the signal decoupler plays a key role in this process, which can effectively decompose the information in the mixed spectrum signal, and then realize the extraction and analysis of the particle trajectory in different particle size intervals.

[0051] The training process of the signal decoupler is based on the prior polymerization reaction knowledge base. The knowledge base contains historical mixed spectrum signal samples and is trained by a deep learning network. Specifically, first, the mixed spectrum signal samples are extracted from the prior polymerization reaction knowledge base, and these signals are sent as input samples into the signal decoupler for processing. The mixed spectrum signal is the data collected by the spectrum detector in real time during the polymerization reaction, which contains the spectral information of different components in the reaction system. However, since these signals are the result of multiple particle size interval particles, they need to be split into their own particle size interval spectrum signals through decoupling technology.

[0052] In the signal decoupling process, the signal decoupler will decompose the mixed spectrum signal, and output the spectrum signals corresponding to the first particle size interval and the second particle size interval, respectively. The first particle size interval usually corresponds to the nucleated particles, and the second particle size interval corresponds to the active growing particles. Through the decoupling of these signals, the behavior of different particle sizes in the reaction can be more accurately identified, providing a data basis for subsequent analysis. In the training process, the mixed spectrum signal samples in the prior polymerization reaction knowledge base are used as input samples, and the decoupling output samples obtained by the decoupler are used as the output of the model.

[0053] In order to verify the accuracy of the decoupling output samples, it is necessary to extract the trajectory feature vectors from these samples. This step involves identifying the trajectory feature vectors related to a specific particle size interval from the decoupled spectrum signal. These feature vectors reflect the behavior characteristics of particles in the polymerization reaction process, such as particle size growth rate, distribution uniformity, etc.

[0054] In order to optimize the performance of the decoupler, the model needs to be trained and feedback optimized. In the training process, the trajectory feature verification data samples are used as verification labels to help evaluate the accuracy of the decoupling output samples. Specifically, by comparing the trajectory feature vectors extracted from the decoupling output samples with the actual trajectory features in the verification data samples, the deviation between the two is calculated, i.e. the loss data. These loss data reflect the errors that may exist in the processing of the decoupler, and then provide the basis for subsequent optimization.

[0055] Then, according to the loss data, the decoupling parameters of the signal decoupler are fed back and optimized. Through the back propagation algorithm, the parameters of the decoupler are adjusted to minimize the loss data, so that it can more accurately extract the spectral signals and trajectory feature vectors of different particle size intervals from the mixed spectral signals. After several optimization iterations, the decoupling effect of the signal decoupler will be improved until the model can stably output high-quality decoupled signals and trajectory features.

[0056] Finally, the trained signal decoupler can receive mixed spectral signals in real time during the actual polymerization reaction, and according to the rules learned by the deep learning network, decouple the signals into spectral signals of different particle size intervals and corresponding particle trajectory features, output the first spectral signal based on the first particle size interval and the second spectral signal based on the second particle size interval. At the same time, the first trajectory feature vector and the second trajectory feature vector corresponding to these spectral signals are identified. These feature vectors will be used for subsequent abnormal deviation analysis and control strategy triggering, providing support for intelligent monitoring and control of polymerization reactions.

[0057] P30: The trajectory feature vectors corresponding to the particle size-trail contrast feature library are called to perform abnormal deviation analysis on the first trajectory feature vector and the second trajectory feature vector, and a directional control strategy is triggered according to the abnormal deviation analysis result. The trajectory feature vector includes the spectral concentration rate of each particle size interval, the appearance time point of the characteristic functional group spectral signal, and the change distribution of the signal intensity over time.

[0058] Further, the step P30 of the embodiments of the present application further comprises:

[0059] P31: Calculate the first trajectory feature abnormal deviation vector of the first trajectory feature vector and the first trajectory feature vector prior threshold; P32: Calculate the second trajectory feature abnormal deviation vector of the second trajectory feature vector and the second trajectory feature vector prior threshold; P33: Compare the first trajectory feature abnormal deviation vector and the second trajectory feature abnormal deviation vector with the preset tolerance threshold, and return the abnormal deviation analysis result.

[0060] It should be understood that the trajectory feature vectors in the particle size-trail contrast feature library are compared with the corresponding trajectory feature vector prior threshold to identify the abnormal deviation of the particle trajectory in the reaction process. This analysis process involves comparing the real-time monitored trajectory feature vector with the pre-set prior threshold to identify any possible abnormal deviation, and accordingly triggering the corresponding control strategy.

[0061] The composition of the trajectory feature vector includes the spectral concentration change rate of the particle trajectory in each particle size interval, the appearance time point of the characteristic functional group spectral signal, and the distribution of signal intensity change over time. These features describe the dynamic characteristics of the behavior of the particles during the polymerization reaction. The spectral concentration change rate reflects the speed of the concentration change of the components in the reaction over time, the appearance time point of the characteristic functional group spectral signal provides information about the generation or disappearance of specific chemical groups during the reaction, and the change of signal intensity over time helps to evaluate the progress of the reaction and the reaction rate of the particles.

[0062] Next, the first trajectory feature abnormal deviation vector of the first trajectory feature vector and the first trajectory feature vector prior threshold is calculated. By comparing the particle trajectory feature vector of the first particle size interval obtained in the current reaction process with the prior set threshold, the deviation degree of the trajectory feature is found. The greater the deviation, the more inconsistent the behavior of the particle trajectory with the normal range of historical data, which may indicate that an abnormal phenomenon occurs in the reaction process. For example, the concentration change rate of the particle exceeds the normal fluctuation range, which may indicate that the reaction rate is abnormal or that some abnormal reaction path occurs.

[0063] Similarly, the second trajectory feature abnormal deviation vector of the second trajectory feature vector and the second trajectory feature vector prior threshold is calculated. This step processes the particle trajectory features of the second particle size interval to detect the deviation of the active growing particle trajectory in the reaction process. If the deviation exceeds the normal range, it may have an adverse effect on the reaction result, indicating that some changes in reaction conditions or materials have caused instability.

[0064] Next, the first trajectory feature abnormal deviation vector and the second trajectory feature abnormal deviation vector are compared with the preset tolerance threshold. The tolerance threshold is set according to the characteristics and tolerance of the reaction in actual operation, which specifies the maximum deviation range allowed by the particle trajectory features. When the deviation vector exceeds the tolerance threshold, the system considers that an abnormality occurs in the reaction, and generates an abnormal deviation analysis result.

[0065] Based on the calculated abnormal deviation analysis result, the corresponding directional regulation strategy is triggered according to the specific deviation. For example, when the particle trajectory deviates greatly, the reaction temperature, feed rate or catalyst usage may need to be adjusted to ensure that the reaction process returns to the normal trajectory range and avoid unstable reaction paths or quality problems of the final product.

[0066] Further, according to the abnormal deviation analysis result, the directional regulation strategy is triggered, and the step P30 of the embodiment of the present application further includes:

[0067] P34: Constructing a directional regulation strategy template, which includes a first regulation strategy template based on nucleation particles, a second regulation strategy template based on active growth particles, and a third regulation strategy template based on abnormal aggregation particles; P35: Analyzing the abnormal deviation analysis result to select a first regulation strategy template or a second regulation strategy template or a third regulation strategy template for the particle size interval, and triggering a directional regulation strategy.

[0068] Specifically, after obtaining the abnormal deviation analysis result, a directional regulation strategy can be triggered based on the abnormal deviation of the particle trajectory to ensure that the polymerization reaction can proceed smoothly and improve the stability of the reaction and the quality of the final product.

[0069] First, a directional regulation strategy template needs to be constructed. These templates can be constructed based on the analysis of the polymerization reaction process and historical data, aiming to provide preset solutions for different stages of the polymerization reaction and possible problems. Specifically, there are three different regulation strategies, respectively for nucleation particles, active growth particles and abnormal aggregation particles. Each particle size interval has different reaction characteristics, so the regulation strategy corresponding to each particle size interval is also different. For example, for nucleation particles, the regulation strategy mainly focuses on the initial stage of the reaction, including initiator compensation, local temperature regulation and initial stirring intensity adjustment; for active growth particles, the regulation strategy focuses on the control of particle growth during the reaction, such as adjusting the feed rate, chain transfer agent addition and main reaction zone temperature adjustment; and for abnormal aggregation particles, measures need to be taken to prevent particle aggregation, including polymerization inhibitor addition, reaction system forced cooling and increased stirring intensity, etc.

[0070] Next, analyze the abnormal deviation analysis result to determine the deviated particle size interval, and select the corresponding regulation strategy template according to the interval. For example, when the particle trajectory characteristics of the nucleation particle size interval deviate, the nucleation particle regulation strategy template can be selected. Specific regulation measures include generating initiator compensation instructions, local temperature regulation instructions and initial stirring intensity adjustment instructions. The initiator compensation instruction is to ensure that the concentration of the initiator in the initial stage of the reaction is appropriate, avoiding instability in the nucleation process; the local temperature regulation instruction is used to control the temperature in the initial stage of the reaction to avoid too fast or too slow reaction; the initial stirring intensity adjustment instruction helps to optimize the particle distribution in the initial stage of the reaction, preventing uneven nucleation phenomenon.

[0071] When the particle trajectory characteristics of the active growth particle size interval deviate, the active growth particle control strategy template can be selected. At this time, the generated control measures include feed rate adjustment instructions, chain transfer agent dosing instructions, and main reaction zone temperature return instructions. The feed rate adjustment instructions help control the supply rate of reactants, ensuring the uniformity of particle growth; the chain transfer agent dosing instructions help adjust the length of the polymerization chain, thereby controlling the molecular weight distribution; and the main reaction zone temperature return instructions can adjust the reaction temperature, keeping the temperature of the reaction zone within a suitable range and preventing excessive temperature fluctuations from affecting particle growth.

[0072] When the particle trajectory characteristics of the abnormal aggregation particle size interval deviate, the abnormal aggregation particle control strategy template can be selected. In order to prevent particle aggregation, the system will trigger instructions including trace dosing of inhibitors, forced cooling of the reaction system, and increased stirring. The trace dosing of inhibitors helps reduce the aggregation phenomenon between particles, the forced cooling instructions inhibit further aggregation of particles by reducing the temperature of the reaction system, and the increased stirring instructions increase the stirring intensity of the reaction system to prevent excessive aggregation or settling of particles.

[0073] Through the above control process, the progress of the polymerization reaction can be monitored in real time, and control measures can be taken in time when abnormalities are detected. Each control strategy is tailored according to the reaction stage and trajectory characteristics of the particles, so as to accurately control the reaction process, reduce fluctuations in the reaction, and effectively improve the quality of the product.

[0074] In summary, the embodiments of the present application have at least the following technical effects:

[0075] The present application can monitor the particle behavior in the polymerization reaction process in real time and accurately by decoupling the spectral signal and constructing the particle size-trace comparison feature library, overcoming the slow response and low precision problems of traditional monitoring methods. By comparing the trajectory characteristic vector with the prior threshold, the abnormal deviation of the particle trajectory can be effectively identified, and the directional control strategy can be triggered in time to ensure the stability of the reaction process. Through intelligent control measures, the reaction conditions are optimized to avoid fluctuations or deviations in the reaction, thereby improving the stability of the polymerization reaction and the quality of the final product. In addition, the automated monitoring and control system reduces manual intervention, improves production efficiency, and reduces human error.

[0076] The technical effects of intelligent monitoring and signal decoupling based on spectral analysis are achieved, realizing real-time and accurate control of the polymerization reaction process, thereby improving product quality and production efficiency.

[0077] Embodiment Two, based on the same inventive concept as the intelligent monitoring and control method for polymerization reaction based on spectral analysis in the preceding embodiments, like Figure 2As shown, the present application provides an intelligent monitoring and regulation device for polymerization reaction based on spectral analysis. The device and method embodiments in the present application are based on the same inventive concept. The device comprises:

[0078] The feature library construction module 11 is configured to construct a particle size-trace contrast feature library based on a prior polymerization reaction knowledge base, wherein the particle size-trace contrast feature library comprises a first trace feature vector prior threshold corresponding to a first particle size interval and a second trace feature vector prior threshold corresponding to a second particle size interval in a polymerization reaction.

[0079] The trace feature recognition module 12 is configured to obtain a mixed spectral signal of a current polymerization reaction, decompose the mixed spectral signal through a signal decoupler, output a first spectral signal based on the first particle size interval and a second spectral signal based on the second particle size interval, and recognize a first trace feature vector and a second trace feature vector of the first particle size interval and the second particle size interval.

[0080] The abnormal deviation analysis module 13 is configured to perform abnormal deviation analysis on the first trace feature vector and the second trace feature vector by calling the trace feature vector prior threshold corresponding to the particle size-trace contrast feature library, and trigger a directional regulation strategy according to the abnormal deviation analysis result.

[0081] Further, in the feature library construction module 11:

[0082] The particle size-trace contrast feature library comprises a first trace feature vector prior threshold corresponding to a first particle size interval and a second trace feature vector prior threshold corresponding to a second particle size interval in a polymerization reaction; the first trace feature vector prior threshold comprises a trace feature vector prior threshold of a polymerization particle with a particle size in the first particle size interval, and the second trace feature vector prior threshold comprises a trace feature vector prior threshold of a polymerization particle with a particle size in the second particle size interval; wherein the first particle size interval and the second particle size interval are non-overlapping and the second particle size interval is larger than the first particle size interval.

[0083] Further, in the feature library construction module 11:

[0084] The particle size-trace contrast feature library further comprises a third trace feature vector prior threshold corresponding to a third particle size interval in a polymerization reaction, and the third trace feature vector prior threshold comprises a trace feature vector prior threshold of a polymerization particle with a particle size in the third particle size interval; wherein the third particle size interval and the second particle size interval are non-overlapping and the third particle size interval is larger than the second particle size interval.

[0085] Further, in the feature library construction module 11:

[0086] The first particle size interval corresponds to the particle size range of the nucleation particles in the polymerization reaction, the second particle size interval corresponds to the particle size range of the active growth particles, and the third particle size interval corresponds to the particle size range of the abnormal aggregation particles.

[0087] Further, the feature library construction module 11 is further used to execute the following steps:

[0088] Determine the polymerization reaction SOP history sample, which includes the polymerization reaction type, the polymerization reaction formula and the polymerization reaction process parameters; extract the mixed spectrum signal sample and the trajectory feature verification data sample of the polymerization reaction SOP history sample, establish the time point association of the mixed spectrum signal sample and the trajectory feature verification data sample, and obtain the prior polymerization reaction knowledge base; divide the prior polymerization reaction knowledge base to obtain the trajectory feature vector prior threshold value corresponding to the first particle size interval, the trajectory feature vector prior threshold value corresponding to the second particle size interval and the trajectory feature vector prior threshold value corresponding to the third particle size interval, and construct the particle size-trajectory contrast feature library.

[0089] Further, the trajectory feature recognition module 12 is further used to execute the following steps:

[0090] The mixed spectrum signal is decomposed by a signal decoupler, and the signal decoupler is obtained by deep learning network training on the prior polymerization reaction knowledge base; wherein the mixed spectrum signal sample of the prior polymerization reaction knowledge base is an input sample, the decoupling output sample of the signal decoupler is obtained, the trajectory feature vector is extracted from the decoupling output sample, the trajectory feature vector output sample is calculated by using the trajectory feature verification data sample as a verification label, and loss data is obtained; the decoupling parameters of the signal decoupler are fed back and optimized according to the loss data, and the learning completed signal decoupler is obtained.

[0091] Further, the abnormal deviation analysis module 13 is further used to execute the following steps:

[0092] Calculate the first trajectory feature abnormal deviation vector of the first trajectory feature vector and the first trajectory feature vector prior threshold value; calculate the second trajectory feature abnormal deviation vector of the second trajectory feature vector and the second trajectory feature vector prior threshold value; compare the first trajectory feature abnormal deviation vector and the second trajectory feature abnormal deviation vector with the preset tolerance threshold, and return the abnormal deviation analysis result.

[0093] Further, in the abnormal deviation analysis module 13:

[0094] The trajectory feature vector includes the spectrum concentration rate of change corresponding to each particle size interval, the appearance time point of the characteristic functional group spectrum signal, and the change distribution of the signal intensity with time.

[0095] Further, the abnormal deviation analysis module 13 is further configured to perform the following steps:

[0096] constructing a directional regulation strategy template, the directional regulation strategy template comprising a first regulation strategy template based on nucleation particles, a second regulation strategy template based on active growth particles and a third regulation strategy template based on heterogeneous aggregation particles; selecting the first regulation strategy template or the second regulation strategy template or the third regulation strategy template triggered by the directional regulation strategy according to the abnormal deviation analysis result.

[0097] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. And the above describes a specific embodiment of the present application. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or can be advantageous.

[0098] The above only describes the preferred embodiments of the present application, and does not limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0099] The specification and drawings are merely exemplary of the present application, and any and all modifications, variations, combinations or equivalents that fall within the scope of the present application are considered to be covered by the present application. Obviously, those skilled in the art can make various modifications and changes to the present application without departing from the scope of the present application. Thus, if these modifications and changes of the present application belong to the scope of the present application and its equivalent technology, the present application is intended to include these modifications and changes.

Claims

1. A method for intelligent monitoring and control of polymerization reactions based on spectral analysis, characterized in that, The method includes: A particle size-trajectory comparison feature library is constructed based on a prior knowledge base of polymerization reaction. The particle size-trajectory comparison feature library includes prior thresholds for trajectory feature vectors corresponding to the first and second particle size intervals in the polymerization reaction. The mixed spectral signal of the current polymerization reaction is acquired, and the mixed spectral signal is decomposed by a signal decoupler to output a first spectral signal based on the first particle size range and a second spectral signal based on the second particle size range. The first trajectory feature vector and the second trajectory feature vector of the first particle size range and the second particle size range are identified. The prior threshold values ​​of the trajectory feature vectors corresponding to the particle size-trajectory comparison feature library are called to perform anomaly deviation analysis on the first trajectory feature vector and the second trajectory feature vector, and the directional control strategy is triggered based on the anomaly deviation analysis results.

2. The method as described in claim 1, characterized in that, The particle size-trajectory comparison feature library includes a first trajectory feature vector prior threshold and a second trajectory feature vector prior threshold based on the first particle size interval and the second particle size interval, respectively, in the polymerization reaction. The first trajectory feature vector prior threshold includes the trajectory feature vector prior threshold of polymerization reaction particles whose particle size is within the first particle size range, and the second trajectory feature vector prior threshold includes the trajectory feature vector prior threshold of polymerization reaction particles whose particle size is within the second particle size range. Wherein, the first particle size range and the second particle size range do not overlap, and the second particle size range is larger than the first particle size range.

3. The method as described in claim 2, characterized in that, The particle size-trajectory comparison feature library also includes a prior threshold for the third trajectory feature vector based on the third particle size interval in the polymerization reaction. The prior threshold for the third trajectory feature vector includes a prior threshold for the trajectory feature vector of the polymerization reaction particles whose particle size is within the third particle size interval. Wherein, the third particle size range and the second particle size range do not overlap, and the third particle size range is larger than the second particle size range.

4. The method as described in claim 3, characterized in that, The first particle size range corresponds to the particle size range of nucleating particles in the polymerization reaction, the second particle size range corresponds to the particle size range of active growth particles, and the third particle size range corresponds to the particle size range of anomalously aggregated particles.

5. The method as described in claim 3, characterized in that, A particle size-trajectory comparison feature library is constructed based on a priori polymerization reaction knowledge base. The methods include: Determine historical samples of SOPs for polymerization reactions, including polymerization reaction type, polymerization reaction formulation, and polymerization reaction process parameters; Extract the mixed spectral signal samples and trajectory feature verification data samples of the historical samples of the polymerization reaction SOP, establish a time point association between the mixed spectral signal samples and the trajectory feature verification data samples, and obtain a priori polymerization reaction knowledge base; The prior aggregation reaction knowledge base is divided to obtain the prior threshold values ​​of trajectory feature vectors corresponding to the first particle size interval, the prior threshold values ​​of trajectory feature vectors corresponding to the second particle size interval, and the prior threshold values ​​of trajectory feature vectors corresponding to the third particle size interval, and a particle size-trajectory comparison feature library is constructed.

6. The method as described in claim 5, characterized in that, The mixed spectral signal is decomposed by a signal decoupler, which is obtained by training a deep learning network on the prior knowledge base of polymerization reactions. The mixed spectral signal samples of the prior polymerization reaction knowledge base are used as input samples. The decoupling output samples of the signal decoupler are obtained. The trajectory feature vectors of the decoupling output samples are extracted. The trajectory feature verification data samples are used as verification labels to calculate the loss of the extracted trajectory feature vector output samples to obtain loss data. The decoupling parameters of the signal decoupler are optimized and learned by feedback based on the loss data to obtain a signal decoupler that has been fully learned.

7. The method as described in claim 2, characterized in that, The method involves calling the prior threshold values ​​of the trajectory feature vectors corresponding to the particle size-trajectory comparison feature library to perform anomaly deviation analysis on the first trajectory feature vector and the second trajectory feature vector, and includes: Calculate the first trajectory feature anomaly deviation vector between the first trajectory feature vector and the prior threshold of the first trajectory feature vector; Calculate the second trajectory feature anomaly deviation vector between the second trajectory feature vector and the prior threshold of the second trajectory feature vector; The first trajectory feature abnormal deviation vector and the second trajectory feature abnormal deviation vector are compared with a preset tolerance threshold, and the abnormal deviation analysis result is returned.

8. The method as described in claim 7, characterized in that, The trajectory feature vector includes the rate of change of spectral concentration for each particle size range, the time point of appearance of the spectral signal of the feature functional group, and the distribution of signal intensity over time.

9. The method as described in claim 4, characterized in that, Based on the results of abnormal deviation analysis, targeted control strategies are triggered. Methods include: A targeted regulation strategy template is constructed, which includes a first regulation strategy template based on nucleating particles, a second regulation strategy template based on active growth particles, and a third regulation strategy template based on heteroaggregated particles. Analyze the particle size range to which the abnormal deviation analysis results belong, and select the first, second, or third control strategy template to trigger a targeted control strategy.

10. A smart monitoring and control device for polymerization reactions based on spectral analysis, characterized in that, The device includes: The feature library construction module is used to construct a particle size-trajectory comparison feature library based on a prior knowledge base of polymerization reaction. The particle size-trajectory comparison feature library includes prior thresholds for trajectory feature vectors corresponding to the first particle size interval and the second particle size interval in the polymerization reaction. The trajectory feature recognition module is used to acquire the mixed spectral signal of the current polymerization reaction, decompose the mixed spectral signal through a signal decoupler, output a first spectral signal based on the first particle size range and a second spectral signal based on the second particle size range, and identify the first trajectory feature vector and the second trajectory feature vector of the first particle size range and the second particle size range. The abnormal deviation analysis module is used to call the prior threshold of the trajectory feature vector corresponding to the particle size-trajectory comparison feature library to perform abnormal deviation analysis on the first trajectory feature vector and the second trajectory feature vector, and trigger a targeted control strategy based on the abnormal deviation analysis results.

Images