Method and system for precise control of initiator addition rate in OLED material synthesis

Abstract

This application provides a method and system for precise control of the initiator addition rate in OLED material synthesis, relating to the field of initiator technology. The method includes: real-time monitoring of the OLED material synthesis process to obtain monomer concentration and degree of polymerization sequences; collecting reactor temperature, pressure, and stirring rate data to obtain a temperature-pressure-stirring rate sequence; performing state-time sequence memory transfer interaction to generate a synthesis reaction scenario representation vector; identifying the initiator addition scenario from the synthesis reaction scenario representation vector to obtain the optimal initiator addition rate; and generating control commands for a high-precision metering pump based on the optimal initiator addition rate. This application solves the technical problem of poor accuracy in controlling the initiator addition rate in OLED material synthesis in existing technologies, achieving the technical effect of improving the accuracy of initiator addition rate control in OLED material synthesis.

Description

Technical Field

[0001] This application relates to the field of initiator technology, and in particular to a method and system for precisely controlling the initiator addition rate in the synthesis of OLED materials. Background Technology

[0002] With the rapid development of OLED material systems in the display and lighting fields, the requirements for material purity, polymerization uniformity and reaction process controllability are constantly increasing. High-precision and stable synthesis process control technology has become a key factor affecting material performance and product consistency.

[0003] Currently, existing OLED material synthesis control technologies mostly employ fixed formulation parameters, empirical control models, or semi-empirical curves for reaction operations. For example, in the initiator addition stage, addition is often based on manual experience or preset quantitative schemes, but lacks the ability to deeply analyze and adaptively adjust to real-time reaction scenarios. Although some processes introduce monitoring equipment such as online mass spectrometers and spectrometers to obtain data such as monomer concentration and degree of polymerization, this data is often only used as a process monitoring tool and is not integrated into a closed loop with initiator addition rate control. Traditional methods also lack the ability to map multi-scale concentration changes, reaction rate changes, and kinetic scenarios, making it difficult for the actual addition rate to follow the real-world requirements of the reaction system.

[0004] In summary, existing technologies suffer from several technical problems. These problems include the lack of a high-dimensional characterization mechanism for historical reaction scenarios, the lack of a prediction model for the optimal initiator addition rate based on polymerization kinetics, and the lack of closed-loop control capability to map the prediction results to the high-precision metering pump execution end. As a result, the initiator addition process cannot be dynamically matched with the real-time needs of the reaction system, which further affects the stability of the polymerization reaction, the uniformity of the reaction rate, and the final structural performance and product consistency of OLED materials. Summary of the Invention

[0005] The purpose of this application is to provide a method and system for precise control of the initiator addition rate in OLED material synthesis, in order to solve the technical problems in the prior art, which are caused by the lack of a high-dimensional characterization mechanism for historical reaction scenarios, the lack of an optimal initiator addition rate prediction model based on polymerization kinetics, and the lack of closed-loop control capability to map the prediction results to the execution end of a high-precision metering pump. As a result, the initiator addition process cannot be dynamically matched with the real-time needs of the reaction system, which further affects the stability of the polymerization reaction, the uniformity of the reaction rate, and the final structural performance and product consistency of the OLED material.

[0006] In view of the above problems, this application provides a method and system for precise control of the initiator addition rate in OLED material synthesis.

[0007] Firstly, this application provides a method for precisely controlling the initiator addition rate in OLED material synthesis, which is achieved through a precise control system for the initiator addition rate in OLED material synthesis. This method includes: real-time monitoring of the OLED material synthesis process using an online mass spectrometer and a UV-Vis spectrometer to obtain monomer concentration and degree of polymerization sequences; simultaneous acquisition of reactor temperature, pressure, and stirring rate using multiple sensors to obtain a temperature-pressure-stirring rate sequence; performing state-time sequence memory transfer interaction on the monomer concentration sequence, degree of polymerization sequence, and temperature-pressure-stirring rate sequence to generate a synthesis reaction scenario characterization vector; calling a polymerization kinetics prediction model to identify the initiator addition scenario on the synthesis reaction scenario characterization vector to obtain the optimal initiator addition rate; and generating control commands for a high-precision metering pump based on the optimal initiator addition rate.

[0008] Preferably, the method for precisely controlling the initiator addition rate in the synthesis of OLED materials further includes: acquiring the basic synthesis process chain and basic material information of OLED materials; using the basic synthesis process chain and basic material information as indexes, performing big data mining of process data to obtain a set of mining monomer concentration sequences, a set of mining degree of polymerization sequences, and a set of mining temperature-pressure-stirring rate sequences; traversing the set of mining monomer concentration sequences, the set of mining degree of polymerization sequences, and the set of mining temperature-pressure-stirring rate sequences to identify intra-sequence fluctuations, determining the set of mining monomer fluctuation bandwidths, the set of mining degree of polymerization fluctuation bandwidths, and the set of mining temperature-pressure-stirring rate fluctuation bandwidths; performing mean shift filtering on the set of mining monomer fluctuation bandwidths, the set of mining degree of polymerization fluctuation bandwidths, and the set of mining temperature-pressure-stirring rate fluctuation bandwidths to determine the representative fluctuation bandwidths of mining monomers, the representative fluctuation bandwidths of mining degree of polymerization, and the representative fluctuation bandwidths of mining temperature-pressure-stirring rate; and distributing the minimum value among the representative fluctuation bandwidths of mining monomers, the representative fluctuation bandwidths of mining degree of polymerization, and the representative fluctuation bandwidths of mining temperature-pressure-stirring rate as the target monitoring bandwidth to an online mass spectrometer, an ultraviolet-visible spectrometer, and a multi-source sensor.

[0009] Preferably, the method for precisely controlling the initiator addition rate in the synthesis of OLED materials further includes: identifying the nearest neighbor concentration decrease rate and polymerization degree increase rate of the monomer concentration sequence and polymerization degree sequence to obtain a monomer concentration decrease rate sequence and a polymerization degree increase rate sequence; performing state-time memory transfer interaction based on the monomer concentration decrease rate sequence to obtain a monomer concentration decrease rate feature; performing state-time memory transfer interaction based on the polymerization degree increase rate sequence and the temperature-pressure-stirring rate sequence to obtain a polymerization degree increase rate feature and a temperature-pressure-stirring rate feature; and performing feature vector encoding on the monomer concentration decrease rate feature, polymerization degree increase rate feature, and temperature-pressure-stirring rate feature to obtain a synthesis reaction scenario characterization vector.

[0010] Preferably, the method for precisely controlling the initiator addition rate in the synthesis of OLED materials further includes: extracting the monomer concentration decrease rate sequence and parsing it according to a preset multi-scale state-time sequence feature analyzer to obtain a multi-scale monomer concentration decrease rate feature set; serializing the multi-scale monomer concentration decrease rate feature set according to the order of analysis scale from small to large to obtain a multi-scale monomer concentration decrease rate feature sequence; and performing state-time sequence memory transfer interaction on the multi-scale monomer concentration decrease rate feature sequence to obtain monomer concentration decrease rate features.

[0011] Preferably, the method for precisely controlling the initiator addition rate in the synthesis of OLED materials further includes: extracting the first and second multi-scale monomer concentration decrease rate features from the multi-scale monomer concentration decrease rate feature sequence; performing similarity recognition on the first and second multi-scale monomer concentration decrease rate features and normalizing the recognition results to obtain a first state temporal memory matrix; using the first state temporal memory matrix to perform memory transfer interaction on the second multi-scale monomer concentration decrease rate to obtain a first interactive monomer concentration decrease rate feature; performing state temporal memory transfer interaction on the third multi-scale monomer concentration decrease rate feature in the multi-scale monomer concentration decrease rate feature sequence based on the first interactive monomer concentration decrease rate feature, and so on, to obtain the monomer concentration decrease rate feature.

[0012] Preferably, the method for precisely controlling the initiator addition rate in OLED material synthesis further includes: acquiring multiple sample state temporal memory matrices, multiple sample multi-scale monomer concentration decrease rates, and corresponding multiple sample interactive monomer concentration decrease rate features as training data; using the training data to perform supervised training on a framework constructed based on a feedforward neural network until training convergence, thereby obtaining a memory transfer interactor; and identifying the first state temporal memory matrix and the second multi-scale monomer concentration decrease rate based on the memory transfer interactor to obtain the first interactive monomer concentration decrease rate feature.

[0013] Preferably, the method for precisely controlling the initiator addition rate in OLED material synthesis further includes: acquiring historical OLED material synthesis records, and extracting a set of historically compliant OLED material synthesis records that meet quality standards from these records; identifying initiator addition scenarios and initiator addition rates on the set of historically compliant OLED material synthesis records to obtain a set of historically compliant reaction scenario representation vectors and a set of historical initiator addition rates; performing homogeneous aggregation on the set of historically compliant reaction scenario representation vectors, and mapping and aggregating the set of historical initiator addition rates based on the aggregation results to obtain multiple sets of aggregation historically compliant reaction scenario representation vectors and multiple sets of aggregation historical initiator addition rates; performing representative screening on the multiple sets of aggregation historically compliant reaction scenario representation vectors to determine multiple screened historically compliant reaction scenario representation vectors; performing representative screening on the multiple sets of aggregation historical initiator addition rates to determine multiple screened historical initiator addition rates; and constructing a polymerization kinetics prediction model based on mapping and correlation between the multiple screened historically compliant reaction scenario representation vectors and the multiple screened historical initiator addition rates.

[0014] Preferably, the method for precisely controlling the initiator addition rate in OLED material synthesis further includes: extracting a first set of polymerization historical compliance reaction scenario representation vectors from the plurality of polymerization historical compliance reaction scenario representation vector sets; extracting the historical compliance reaction scenario representation vector with the highest similarity to other historical compliance reaction scenario representation vectors in the first set of polymerization historical compliance reaction scenario representation vectors as an initial screening center; iterating on the initial screening center in the first set of polymerization historical compliance reaction scenario representation vectors according to a preset representative screening step size to obtain an iterative screening center, wherein the iterative screening center is the historical compliance reaction scenario representation vector in the first set of polymerization historical compliance reaction scenario representation vectors that is closest to the initial screening center in terms of Euclidean distance and preset representative screening step size; determining whether the iterative screening center can replace the initial screening center; if so, continuing to iterate on the iterative screening center according to the preset representative screening step size until the maximum number of iterations is met to obtain the first screening historical compliance reaction scenario representation vector; if not, stopping the iteration and using the initial screening center as the first screening historical compliance reaction scenario representation vector; and adding the first screening historical compliance reaction scenario representation vector to the plurality of screening historical compliance reaction scenario representation vectors.

[0015] Preferably, the method for precisely controlling the initiator addition rate in the synthesis of OLED materials further includes: counting the number of historical compliant reaction scenario representation vectors in the first set of historical compliant reaction scenario representation vectors whose Euclidean distance to the iterative screening center is within a preset representative screening step size, to obtain the iterative screening coefficient; counting the number of historical compliant reaction scenario representation vectors in the first set of historical compliant reaction scenario representation vectors whose Euclidean distance to the initial screening center is within a preset representative screening step size, to obtain the initial screening coefficient; when the iterative screening coefficient is greater than or equal to the initial screening coefficient, substitution is allowed; when the iterative screening coefficient is less than the initial screening coefficient, substitution is not allowed.

[0016] Secondly, this application also provides a precise control system for the initiator addition rate in OLED material synthesis, used to execute the precise control method for the initiator addition rate in OLED material synthesis as described in the first aspect, including: a polymerization degree sequence acquisition module, used to monitor the OLED material synthesis process in real time using an online mass spectrometer and an ultraviolet-visible spectrometer to obtain a monomer concentration sequence and a polymerization degree sequence; a temperature-pressure-stirring rate sequence acquisition module, used to simultaneously collect the reactor temperature, pressure, and stirring rate using multi-source sensors to obtain a temperature-pressure-stirring rate sequence; a synthesis reaction scenario characterization vector generation module, used to perform state-time sequence memory transfer interaction on the monomer concentration sequence, polymerization degree sequence, and temperature-pressure-stirring rate sequence to generate a synthesis reaction scenario characterization vector; and a control command generation module, used to call a polymerization kinetics prediction model to identify the initiator addition scenario on the synthesis reaction scenario characterization vector, obtain the optimal initiator addition rate, and generate control commands for a high-precision metering pump based on the optimal initiator addition rate.

[0017] The technical solution provided in this application has at least the following technical effects or advantages: by achieving the technical goal of accurate analysis of polymerization kinetics based on multi-source data fusion modeling and adaptive control of the initiator addition process, the technical effect of significantly improving the timing accuracy, flow accuracy and overall polymerization quality consistency of initiator addition is achieved while ensuring the stability of the reaction system.

[0018] The above description is merely an overview of the technical solution of this application. To enable a clearer understanding of the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating the method for precisely controlling the initiator addition rate in the synthesis of OLED materials according to this application.

[0021] Figure 2 This is a schematic diagram of the structure of the precise control system for the initiator addition rate in the synthesis of OLED materials in this application.

[0022] Figure labeling: Module 1 for obtaining degree of polymerization sequence, Module 2 for obtaining temperature-pressure-stirring rate sequence, Module 3 for generating characterization vector of synthesis reaction scenario, and Module 4 for generating control command. Detailed Implementation

[0023] This application provides a method and system for precise control of the initiator addition rate in OLED material synthesis. It addresses the technical problems in existing technologies where the lack of a high-dimensional characterization mechanism for historical reaction scenarios, the lack of an optimal initiator addition rate prediction model based on polymerization kinetics, and the lack of closed-loop control capabilities to map the prediction results to a high-precision metering pump actuator prevents the initiator addition process from dynamically matching the real-time needs of the reaction system. This, in turn, affects the stability of the polymerization reaction, the uniformity of the reaction rate, and the final structural performance and product consistency of the OLED material. The application achieves the technical goal of precise analysis of polymerization kinetics based on multi-source data fusion modeling and adaptive control of the initiator addition process. This results in a significant improvement in the timing accuracy, flow rate precision, and overall polymerization quality consistency of the initiator addition process while ensuring the stability of the reaction system.

[0024] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0025] Example 1, please refer to the appendix. Figure 1 This application provides a method for precisely controlling the initiator addition rate in OLED material synthesis, which is applied to a precise control system for the initiator addition rate in OLED material synthesis. Specifically, it includes the following steps:

[0026] S1: The synthesis process of OLED materials was monitored in real time using an online mass spectrometer and a UV-Vis spectrometer to obtain monomer concentration sequences and degree of polymerization sequences.

[0027] Specifically, real-time monitoring of the OLED material synthesis process using online mass spectrometry and ultraviolet-visible spectroscopy refers to continuously acquiring dynamic information on the composition and optical properties of substances within the reaction system during the synthesis reaction, using online detection devices configured outside the reaction apparatus or at pipeline nodes. The online mass spectrometer is used to analyze the mass-charge ratio of various monomer molecules and their associated fragment ions in the reaction system, generating time-series data reflecting monomer consumption rates and transformation trends. The ultraviolet-visible spectroscopy is used to analyze the light absorption characteristics of species with different degrees of polymerization generated during the reaction, generating time-series data characterizing the degree of polymerization chain growth and distribution changes. The online mass spectrometer and ultraviolet-visible spectroscopy work together to acquire and output measurement results in real time, generating monomer concentration sequences and degree of polymerization sequences respectively. This provides continuous and reliable basic data support for subsequent time-series analysis and precise control of the OLED material synthesis reaction state.

[0028] S2: The temperature, pressure and stirring rate of the reactor are collected simultaneously by multiple source sensors to obtain the temperature-pressure-stirring rate sequence.

[0029] Specifically, by simultaneously acquiring temperature, pressure, and stirring rate data from a reactor using multiple sensors to obtain a temperature-pressure-stirring rate sequence, this refers to the parallel acquisition of the thermodynamic and mechanical vibration states of the reaction system during the synthesis of OLED materials using various types of real-time monitoring devices placed at different monitoring locations within the reactor. Temperature sensors continuously detect the heat distribution and temperature field changes within the reactor to reflect the exothermic, endothermic, and heat transfer behaviors during the reaction. Pressure sensors record real-time pressure changes in the gas and liquid phases within the reactor to characterize system pressure fluctuations caused by monomer conversion, solvent evaporation, and polymerization. Stirring rate sensors monitor the rotational speed of the stirring mechanism to correspond to the mixing efficiency, material dispersion, and the impact of disturbances in local reaction kinetics. These three types of sensors output continuous data on the same time axis, and their respective time series are synchronously integrated according to timestamps, forming a temperature-pressure-stirring rate sequence that reflects the dynamic changes of temperature, pressure, and stirring rate over time in the reaction system. This provides multi-dimensional process data for characterizing the multi-physics evolution of the reaction system and supporting subsequent state modeling.

[0030] S3: Perform state-time memory transfer interaction on monomer concentration sequence, degree of polymerization sequence and temperature-pressure-stirring rate sequence to generate synthetic reaction scenario characterization vector.

[0031] Furthermore, this application also includes: identifying the nearest neighbor concentration decrease rate and polymerization increase rate of the monomer concentration sequence and the degree of polymerization sequence to obtain a monomer concentration decrease rate sequence and a degree of polymerization increase rate sequence; performing state-time memory transfer interaction based on the monomer concentration decrease rate sequence to obtain a monomer concentration decrease rate feature; performing state-time memory transfer interaction based on the degree of polymerization increase rate sequence and the temperature-pressure-stirring rate sequence to obtain a degree of polymerization increase rate feature and a temperature-pressure-stirring rate feature; and performing feature vector encoding on the monomer concentration decrease rate feature, degree of polymerization increase rate feature, and temperature-pressure-stirring rate feature to obtain a synthesis reaction scenario characterization vector.

[0032] Furthermore, this application also includes: extracting the monomer concentration decrease rate sequence and parsing it according to a preset multi-scale state-time series feature analyzer to obtain a multi-scale monomer concentration decrease rate feature set; serializing the multi-scale monomer concentration decrease rate feature set according to the order of analysis scale from small to large to obtain a multi-scale monomer concentration decrease rate feature sequence; and performing state-time series memory transfer interaction on the multi-scale monomer concentration decrease rate feature sequence to obtain monomer concentration decrease rate features.

[0033] Furthermore, this application also includes: extracting the first and second multi-scale monomer concentration decrease rate features from the multi-scale monomer concentration decrease rate feature sequence; performing similarity recognition on the first and second multi-scale monomer concentration decrease rate features and normalizing the recognition results to obtain a first state temporal memory matrix; using the first state temporal memory matrix to perform memory transfer interaction on the second multi-scale monomer concentration decrease rate to obtain a first interactive monomer concentration decrease rate feature; performing state temporal memory transfer interaction on the third multi-scale monomer concentration decrease rate feature in the multi-scale monomer concentration decrease rate feature sequence based on the first interactive monomer concentration decrease rate feature, and so on, to obtain the monomer concentration decrease rate feature.

[0034] Furthermore, this application also includes: acquiring multiple sample state temporal memory matrices, multiple sample multi-scale monomer concentration decrease rates, and corresponding multiple sample interactive monomer concentration decrease rate features as training data; using the training data to perform supervised training on a framework constructed based on a feedforward neural network until training converges, thereby obtaining a memory transfer interactor; and identifying the first state temporal memory matrix and the second multi-scale monomer concentration decrease rate based on the memory transfer interactor to obtain the first interactive monomer concentration decrease rate feature.

[0035] Specifically, identifying the nearest neighbor concentration decrease rate and polymerization increase rate for monomer concentration sequences and degree of polymerization sequences involves taking the data sequences of monomer concentration and degree of polymerization changes over time obtained from real-time monitoring, using the numerical difference between adjacent time points as local change quantities, calculating the magnitude and rate of decrease of monomer concentration between adjacent sampling points to form a monomer concentration decrease rate sequence; and calculating the magnitude and rate of increase of degree of polymerization between adjacent sampling points to form a degree of polymerization increase rate sequence. This ensures that both types of sequences can reflect the dynamic trends of monomer consumption and polymer chain growth during the polymerization reaction.

[0036] Extracting the monomer concentration decline rate sequence and parsing it according to the preset multi-scale state time series feature analyzer means that when performing feature processing on the sequence data of monomer concentration decline rate changing over time, the preset multi-scale time series feature analysis model is invoked. This model performs multi-scale decomposition and analysis on the same decline rate sequence by setting different time window lengths, frequency domain resolutions, or feature extraction scales, in order to extract various features such as local trends, global changes, short-term fluctuations, and long-term evolution at different scales, thereby forming a multi-scale monomer concentration decline rate feature set for subsequent modeling.

[0037] Subsequently, the multi-scale monomer concentration decrease rate feature set is serialized according to the order of analysis scale from small to large. This means that the features extracted from the set at different scales are sorted according to the scale size, with small-scale features taking priority and large-scale features following. The sorted features are then arranged in order according to time or scale index, thereby generating a structured multi-scale monomer concentration decrease rate feature sequence, so that the features have a strict scale association order when input.

[0038] Extracting the first and second multi-scale monomer concentration decrease rate features from the multi-scale monomer concentration decrease rate feature sequence refers to selecting two feature vectors located at the first and second positions in a multi-scale feature sequence that has been organized in scale order. These vectors are used as the initial scale features and their adjacent scale features, respectively, to establish the initial input for cross-scale feature association. This allows subsequent state interaction calculations to use the most detailed scale change information as the modeling basis.

[0039] Next, similarity identification of the first and second multi-scale monomer concentration decrease rate features is performed, and the identification results are normalized. This means that similarity measurement calculation is performed on two decrease rate feature vectors with the same source but different scales. The degree of closeness between the two in numerical patterns is identified by vector dot product, distance measurement or specific similarity function, and the similarity value is normalized in interval to construct a standardized cross-scale correlation metric of the same type. Based on this, a first-state temporal memory matrix is ​​formed. This memory matrix is ​​filled into a preset structured matrix by normalized similarity values ​​and is used as the weight basis for cross-scale information transfer in subsequent steps.

[0040] In the step of obtaining multiple sample state temporal memory matrices, multiple sample multi-scale monomer concentration decrease rates, and corresponding multiple sample interaction monomer concentration decrease rate features as training data, the state temporal memory matrix is ​​used to characterize the state evolution characteristics of samples at different reaction stages, the multi-scale monomer concentration decrease rate is used to reflect the monomer depletion dynamics of samples at different time scales, and the interaction monomer concentration decrease rate features are used to describe the coupling influence between different state variables in the time dimension, thereby forming a complete training dataset that can cover the multi-dimensional temporal behavior of samples.

[0041] Meanwhile, the framework based on the feedforward neural network is supervised and trained using training data until the training converges, so that the weights inside the network can accurately fit the mapping relationship between the sample state memory information and the multi-scale concentration decrease rate. The feedforward neural network is used to establish a nonlinear mapping between input features and target features. Training convergence means that the loss function reaches a stable state within a preset threshold, thereby obtaining a memory transfer interactor that can perform temporal feature interaction inference.

[0042] Subsequently, based on the memory transfer interactor, the first-state temporal memory matrix and the second-dimensional multi-scale monomer concentration decrease rate are identified. By inputting the two types of temporal input features into the memory transfer interactor, the first interactive monomer concentration decrease rate feature representing the coupling relationship between the two types of features is output, thereby achieving a deep characterization of the monomer consumption dynamics in the reaction system.

[0043] Subsequently, based on the first interactive monomer concentration decrease rate feature, the third multi-scale monomer concentration decrease rate feature in the multi-scale feature sequence is subjected to state temporal memory transfer interaction. This means that the updated interactive feature is used as the new state memory input, and the same similarity recognition, weight application and feature transfer process is performed with the third feature, so that the information between scales diffuses sequentially. This process is repeated until the entire multi-scale sequence has completed the association interaction calculation, thereby obtaining the monomer concentration decrease rate feature that can comprehensively reflect the multi-scale decrease rate pattern.

[0044] Furthermore, the state-time memory transfer interaction based on the polymerization degree rise rate sequence and the temperature-pressure-stirring rate sequence refers to taking the polymerization degree rise rate sequence, which reflects the change in the growth rate of the polymerization chain, and the temperature, pressure, and stirring rate sequences, which reflect the changes in the physical environment, as joint inputs. Through the state-time memory transfer interaction mechanism, the temporal information across physical quantities is correlated and modeled to extract the polymerization degree rise rate feature that can characterize the development trend of polymerization kinetics and the temperature-pressure-stirring rate feature that can reflect the overall physical environment evolution behavior of the reaction system. This allows the two types of features to simultaneously describe the mutual influence between material changes and operating condition changes in the time dimension.

[0045] Furthermore, feature vector encoding of monomer concentration decrease rate, polymerization degree increase rate, and temperature-pressure-stirring rate features refers to the unified data structuring of three types of features with different physical meanings and originating from different state sequences. Through vector concatenation, linear mapping, or embedding encoding, a synthetic reaction scenario characterization vector is generated to comprehensively characterize the chemical and physical evolution state of the reaction system, so that the vector can serve as the overall input for polymerization reaction state identification and control strategy inference.

[0046] S4: Call the polymerization kinetics prediction model to identify the initiator addition scenario of the synthesis reaction scenario characterization vector, obtain the optimal initiator addition rate, and generate control commands for the high-precision metering pump based on the optimal initiator addition rate.

[0047] Furthermore, this application also includes: obtaining historical OLED material synthesis records; extracting a set of historically compliant OLED material synthesis records with quality compliance from the historical OLED material synthesis records; identifying initiator addition scenarios and initiator addition rates on the set of historically compliant OLED material synthesis records to obtain a set of historically compliant reaction scenario characterization vectors and a set of historical initiator addition rates; performing homogeneous aggregation on the set of historically compliant reaction scenario characterization vectors, and mapping and aggregating the set of historical initiator addition rates according to the aggregation results to obtain multiple sets of aggregation historically compliant reaction scenario characterization vectors and multiple sets of aggregation historical initiator addition rates; performing representative screening on the multiple sets of aggregation historically compliant reaction scenario characterization vectors to determine multiple screened historically compliant reaction scenario characterization vectors; performing representative screening on the multiple sets of aggregation historical initiator addition rates to determine multiple screened historical initiator addition rates; and constructing a polymerization kinetics prediction model based on mapping and association of the multiple screened historically compliant reaction scenario characterization vectors and the multiple screened historical initiator addition rates.

[0048] Furthermore, this application also includes: extracting a first aggregated historical compliance response scenario representation vector set from the plurality of aggregated historical compliance response scenario representation vector sets; extracting the historical compliance response scenario representation vector with the highest similarity to other historical compliance response scenario representation vectors in the first aggregated historical compliance response scenario representation vector set, as an initial screening center; iterating on the initial screening center in the first aggregated historical compliance response scenario representation vector set according to a preset representative screening step size, thereby obtaining an iterative screening center, wherein the iterative screening center is the historical compliance response scenario representation vector in the first aggregated historical compliance response scenario representation vector set that is closest to the initial screening center in terms of Euclidean distance and preset representative screening step size; determining whether the iterative screening center can replace the initial screening center, and if so, continuing to iterate on the iterative screening center according to the preset representative screening step size until the maximum number of iterations is met to obtain the first screening historical compliance response scenario representation vector; if not, stopping the iteration and using the initial screening center as the first screening historical compliance response scenario representation vector; adding the first screening historical compliance response scenario representation vector to the plurality of screening historical compliance response scenario representation vectors.

[0049] Furthermore, this application also includes: counting the number of historical compliance response scenario representation vectors in the first aggregated historical compliance response scenario representation vector set whose Euclidean distance to the iterative screening center is within a preset representative screening step size, to obtain an iterative screening coefficient; counting the number of historical compliance response scenario representation vectors in the first aggregated historical compliance response scenario representation vector set whose Euclidean distance to the initial screening center is within a preset representative screening step size, to obtain an initial screening coefficient; when the iterative screening coefficient is greater than or equal to the initial screening coefficient, it can be substituted; when the iterative screening coefficient is less than the initial screening coefficient, it cannot be substituted.

[0050] Specifically, historical OLED material synthesis records are obtained. These records characterize the raw material ratios, reaction conditions, process control, and product quality data involved in previous material preparation processes. From these historical OLED material synthesis records, a set of historically compliant OLED material synthesis records is extracted. Quality compliance indicates that the corresponding products meet preset performance indicators and process stability requirements. This set of historically compliant OLED material synthesis records then serves as a reliable data foundation for subsequent model training and analysis.

[0051] Subsequently, the initiator addition scenario and initiator addition rate were identified in the historical compliant OLED material synthesis record set. By analyzing the time series operation data and reaction condition parameters in the record, the operation mode, timing of initiator addition, and the changing addition rate during the addition process were identified. The historical compliant reaction scenario characterization vector set and the historical initiator addition rate set were extracted and formed respectively to describe the reaction environment state and initiator dynamics in the actual synthesis process.

[0052] Subsequently, the set of historical compliant reaction scenario representation vectors is aggregated by category, and the set of historical initiator addition rates is mapped and aggregated based on the aggregation results. Multiple reaction scenario vectors are divided into several groups by similarity measurement, and the initiator addition rates of the corresponding groups are clustered simultaneously, so as to obtain multiple aggregated sets of historical compliant reaction scenario representation vectors and multiple aggregated sets of historical initiator addition rates, thereby realizing the corresponding classification of reaction condition features and initiator addition behavior.

[0053] In the process of extracting the first set of aggregated historical compliance reaction scenario representation vectors from multiple sets of aggregated historical compliance reaction scenario representation vectors, multiple sets of aggregated historical compliance reaction scenario representation vectors are used to represent different reaction scenario groups obtained after the same type of aggregation. The reaction scenario representation vectors in each group are used to characterize the multi-dimensional conditional features such as temperature field, concentration field, addition strategy or reaction dynamics in the corresponding OLED material synthesis process. The first set of aggregated historical compliance reaction scenario representation vectors is an arbitrarily selected group among them, used to perform representative screening operations.

[0054] Subsequently, the historical compliance response scenario representation vector with the highest similarity to other historical compliance response scenario representation vectors in the first aggregated historical compliance response scenario representation vector set is extracted as the initial screening center. The similarity is used to represent the cumulative value of the vector and the vector of the same class. The higher the similarity, the more comprehensively it can represent the key features of the group, and therefore it can be used as the starting point for screening.

[0055] Next, according to the preset representative screening step size, the initial screening center is iterated in the first aggregated historical compliance response scenario representation vector set to obtain the iterative screening center. The preset representative screening step size is used to limit the range of Euclidean distance between the new center selected in each iteration and the previous center, so as to achieve gradual and stable center migration. The iterative screening center is the response scenario representation vector in the first aggregated historical compliance response scenario representation vector set that is closest to the initial screening center in terms of Euclidean distance and the step size, and is used to guide the screening process to advance along the representative direction within the group.

[0056] The number of historical compliance response scenario representation vectors in the first aggregated set of historical compliance response scenario representation vectors whose Euclidean distance to the iterative screening center is within a preset representative screening step size is counted to obtain the iterative screening coefficient. Here, the Euclidean distance is used to measure the degree of difference between two response scenario representation vectors in the multidimensional feature space, the preset representative screening step size is used to limit the acceptable distance range, the number of vectors falling within this range is used to measure the representative coverage of the iterative screening center within the class, and the iterative screening coefficient reflects the representativeness of the new candidate center for the samples within the class.

[0057] Subsequently, the number of historical compliance response scenario representation vectors in the first aggregated historical compliance response scenario representation vector set whose Euclidean distance to the initial screening center is within the preset representative screening step size is counted to obtain the initial screening coefficient. The initial screening coefficient is used to characterize the effective coverage of the initial screening center to the group samples and is a benchmark value used to compare with the representativeness of the iterative screening center.

[0058] Based on this, when the iterative screening coefficient is greater than or equal to the initial screening coefficient, it can be replaced. This indicates that the iterative screening center can cover a number of class samples no less than the number of initial screening centers under a given distance constraint, thus demonstrating that it has at least equal or even higher representativeness. Therefore, it can be used as a new screening center to guide subsequent iterative screening. Conversely, when the iterative screening coefficient is less than the initial screening coefficient, it cannot be replaced. This indicates that the iterative screening center is not representative enough for the class group and cannot provide more effective centrality features than the initial screening center. Therefore, the replacement should be terminated and the initial screening center should be retained as the final result.

[0059] Subsequently, it is determined whether the iterative screening center can replace the initial screening center. By comparing the centrality index, similarity coverage, and whether they can more effectively represent the overall characteristics of the group, if it is determined that they can be replaced, the iterative screening center is iterated according to the preset representative screening step size until the maximum number of iterations is met, and the first screening historical compliance response scenario representation vector is obtained to ensure the stability and representativeness of the screening results.

[0060] If not, the iteration stops, and the initial screening center is used as the first screening historical compliance response scenario representation vector, thereby preventing the screening results from deviating from the core features of the group. The first screening historical compliance response scenario representation vector is added to multiple screening historical compliance response scenario representation vectors, making it one of the typical input conditions for constructing the aggregation dynamics prediction model, in order to improve the model's generalization ability to different response scenarios.

[0061] Meanwhile, representative screening was performed on multiple sets of polymerization history initiator addition rates. Based on the stability, mean characteristics and trend of the rate sequence in the group, multiple screening history initiator addition rates were determined, so as to continuously retain addition rate behavior characteristics that are highly representative for model construction.

[0062] Finally, based on the mapping and correlation of multiple historical compliant reaction scenario characterization vectors and multiple historical initiator addition rates, a polymerization kinetics prediction model is constructed by establishing the corresponding mapping relationship between different reaction scenarios and initiator rates. This model enables the model to infer the optimal or reasonable initiator addition dynamics based on the reaction scenario input, thereby achieving the prediction and control of the polymerization reaction kinetics of OLED materials.

[0063] The process of generating control commands for a high-precision metering pump based on the optimal initiator addition rate refers to determining the optimal initiator addition rate that meets the target polymerization kinetics requirements, using this rate parameter as a control input, and converting the continuous rate value into discrete execution commands that can be executed by the high-precision metering pump through a control algorithm. The high-precision metering pump is a metering execution device capable of stable, uniform, and repeatable delivery within a small flow range. It uses a stepper motor, linear driver, or servo mechanism to precisely adjust the liquid feed rate. The control commands are standardized digital signals used to drive the metering pump to change the instantaneous flow rate, cumulative flow rate, or pulse frequency, ensuring that the actual addition process strictly follows the optimal addition rate curve, thereby guaranteeing that the heat release, free radical generation, and chain growth behavior of the reaction system are within the expected control trajectory.

[0064] Furthermore, this application also includes: obtaining the basic synthesis process chain and basic material information of OLED materials; using the basic synthesis process chain and basic material information as indexes, performing big data mining of process data to obtain a set of mining monomer concentration sequences, a set of mining polymerization degree sequences, and a set of mining temperature-pressure-stirring rate sequences; traversing the set of mining monomer concentration sequences, the set of mining polymerization degree sequences, and the set of mining temperature-pressure-stirring rate sequences to identify intra-sequence fluctuations, determining a set of mining monomer fluctuation bandwidths, a set of mining polymerization degree fluctuation bandwidths, and a set of mining temperature-pressure-stirring rate fluctuation bandwidths; performing mean shift filtering on the set of mining monomer fluctuation bandwidths, the set of mining polymerization degree fluctuation bandwidths, and the set of mining temperature-pressure-stirring rate fluctuation bandwidths respectively to determine representative fluctuation bandwidths of mining monomers, representative fluctuation bandwidths of mining polymerization degree, and representative fluctuation bandwidths of mining temperature-pressure-stirring rate; using the minimum value among the representative fluctuation bandwidths of mining monomers, representative fluctuation bandwidths of mining polymerization degree, and representative fluctuation bandwidths of mining temperature-pressure-stirring rate as the target monitoring bandwidth, and distributing the target monitoring bandwidth to an online mass spectrometer, an ultraviolet-visible spectrometer, and a multi-source sensor.

[0065] Specifically, the basic synthesis process chain and material information of OLED materials are obtained. Data describing the key process sequence, operating condition range and reaction node characteristics of OLED material synthesis process are extracted from existing process knowledge bases or historical production records. At the same time, basic attribute information related to the target OLED material, such as molecular structure properties, monomer types, prepolymer characteristic parameters and typical behavior patterns under different process conditions, is extracted and used as retrieval indexes and screening criteria for subsequent data mining processes.

[0066] Subsequently, using the basic synthetic process chain and material information as indexes, big data mining of process data is performed. This involves retrieving multiple batches of process operation data consistent with the current process and material properties from a large-scale historical production batch database through index matching and conditional filtering. From this data, a set of sequences is formed, containing multiple historical monomer concentration change sequences, multiple degree of polymerization change sequences, and multiple corresponding temperature, pressure, and stirring rate change sequences. These are respectively used as the set for mining monomer concentration sequences, degree of polymerization sequences, and temperature-pressure-stirring rate sequences. The temperature, pressure, and stirring rate sequences are combined in a one-to-one correspondence with each batch to reflect the multivariate coupled changes under actual reaction conditions.

[0067] The system iterates through sets of monomer concentration sequences, degree of polymerization sequences, and temperature-pressure-stirring rate sequences to identify intra-sequence fluctuations. For each sequence in the set, fluctuation analysis is performed based on local gradient changes, differential peak and valley identification, or amplitude shift calculation. The intervals of the rate of decrease in monomer concentration are extracted in the sequence to form a set of monomer fluctuation bandwidths, the intervals of the rate of increase in degree of polymerization are extracted in the sequence to form a set of degree of polymerization fluctuation bandwidths, and the fluctuation intervals of temperature, pressure, and stirring rate in the time dimension are identified to form a set of temperature-pressure-stirring rate fluctuation bandwidths. This allows various fluctuation bandwidths to reflect the true dynamic instability characteristics of the reaction system.

[0068] Furthermore, mean shift screening is performed on the set of individual fluctuation bandwidth, the set of aggregation degree fluctuation bandwidth, and the set of temperature-pressure-stirring rate fluctuation bandwidth respectively. This means that the density estimation and cluster center migration calculation of the data distribution of each type of fluctuation bandwidth set are performed based on the mean shift algorithm. The stable and representative distribution center in the set is gradually approximated through an iterative method. Representative fluctuation bandwidth that can represent the fluctuation scale characteristics of the whole set is identified from each set. These representative fluctuation bandwidths are used as the representative fluctuation bandwidths of individual fluctuation bandwidth, aggregation degree fluctuation bandwidth, and temperature-pressure-stirring rate fluctuation bandwidth respectively, so that the representative bandwidths can be used to describe the main fluctuation scale of this type of parameter under typical process conditions.

[0069] Furthermore, the minimum value among the representative fluctuation bandwidths of monomers, polymerization degree, and temperature-pressure-stirring rate is selected as the target monitoring bandwidth. This means selecting the minimum bandwidth among the typical fluctuation scales of the three key sequences as the minimum sensitivity threshold for real-time monitoring, so that the monitoring equipment can capture subtle changes in the key parameters of the reaction system with sufficient resolution. The target monitoring bandwidth is then distributed to online mass spectrometers, UV-Vis spectrometers, and multi-source sensors, enabling various monitoring devices to adjust the sampling frequency or signal resolution configuration based on a unified minimum monitoring bandwidth, ensuring that the real-time monitoring data accurately reflects the dynamic change trend of the reaction system.

[0070] In summary, the method for precise control of initiator addition rate in OLED material synthesis provided in this application has the following technical effects: by achieving the technical goal of precise analysis of polymerization kinetics based on multi-source data fusion modeling and adaptive control of the initiator addition process, the method significantly improves the timing accuracy, flow rate accuracy and overall polymerization quality consistency of initiator addition while ensuring the stability of the reaction system.

[0071] Example 2: Based on the same inventive concept as the method for precisely controlling the initiator addition rate in OLED material synthesis described in the foregoing examples, this application also provides a control system for precisely controlling the initiator addition rate in OLED material synthesis. Please refer to the appendix. Figure 2 The system includes: a polymerization degree sequence acquisition module 1, used to monitor the OLED material synthesis process in real time using an online mass spectrometer and a UV-Vis spectrometer to obtain monomer concentration and polymerization degree sequences; a temperature-pressure-stirring rate sequence acquisition module 2, used to simultaneously collect reactor temperature, pressure, and stirring rate data from multiple sensors to obtain a temperature-pressure-stirring rate sequence; a synthesis reaction scenario characterization vector generation module 3, used to perform state-time sequence memory transfer interaction on the monomer concentration sequence, polymerization degree sequence, and temperature-pressure-stirring rate sequence to generate a synthesis reaction scenario characterization vector; and a control command generation module 4, used to call a polymerization kinetics prediction model to identify the initiator addition scenario on the synthesis reaction scenario characterization vector, obtain the optimal initiator addition rate, and generate control commands for a high-precision metering pump based on the optimal initiator addition rate.

[0072] Furthermore, the precise control system for initiator addition rate in OLED material synthesis is also used for: acquiring the basic synthesis process chain and basic material information of OLED materials; using the basic synthesis process chain and basic material information as indexes, performing big data mining of process data to obtain a set of mining monomer concentration sequences, a set of mining degree of polymerization sequences, and a set of mining temperature-pressure-stirring rate sequences; traversing the set of mining monomer concentration sequences, the set of mining degree of polymerization sequences, and the set of mining temperature-pressure-stirring rate sequences to identify intra-sequence fluctuations, determining the set of mining monomer fluctuation bandwidths, the set of mining degree of polymerization fluctuation bandwidths, and the set of mining temperature-pressure-stirring rate fluctuation bandwidths; performing mean shift filtering on the set of mining monomer fluctuation bandwidths, the set of mining degree of polymerization fluctuation bandwidths, and the set of mining temperature-pressure-stirring rate fluctuation bandwidths to determine the representative fluctuation bandwidths of mining monomers, the representative fluctuation bandwidths of mining degree of polymerization, and the representative fluctuation bandwidths of mining temperature-pressure-stirring rate; and distributing the minimum value among the representative fluctuation bandwidths of mining monomers, the representative fluctuation bandwidths of mining degree of polymerization, and the representative fluctuation bandwidths of mining temperature-pressure-stirring rate as the target monitoring bandwidth to an online mass spectrometer, an ultraviolet-visible spectrometer, and a multi-source sensor.

[0073] Furthermore, the precise control system for the initiator addition rate in the synthesis of OLED materials is also used for: identifying the nearest neighbor concentration decrease rate and polymerization increase rate of the monomer concentration sequence and polymerization degree sequence to obtain the monomer concentration decrease rate sequence and polymerization degree increase rate sequence; performing state-time memory transfer interaction based on the monomer concentration decrease rate sequence to obtain monomer concentration decrease rate characteristics; performing state-time memory transfer interaction based on the polymerization degree increase rate sequence and temperature-pressure-stirring rate sequence to obtain polymerization degree increase rate characteristics and temperature-pressure-stirring rate characteristics; and performing feature vector encoding on the monomer concentration decrease rate characteristics, polymerization degree increase rate characteristics, and temperature-pressure-stirring rate characteristics to obtain a synthesis reaction scenario characterization vector.

[0074] Furthermore, the precise control system for the initiator addition rate in the OLED material synthesis is also used for: extracting the monomer concentration decrease rate sequence and parsing it according to a preset multi-scale state-time sequence feature analyzer to obtain a multi-scale monomer concentration decrease rate feature set; serializing the multi-scale monomer concentration decrease rate feature set according to the order of analysis scale from small to large to obtain a multi-scale monomer concentration decrease rate feature sequence; and performing state-time sequence memory transfer interaction on the multi-scale monomer concentration decrease rate feature sequence to obtain monomer concentration decrease rate features.

[0075] Furthermore, the precise control system for the initiator addition rate in the OLED material synthesis is also used for: extracting the first and second multi-scale monomer concentration decrease rate features from the multi-scale monomer concentration decrease rate feature sequence; performing similarity recognition on the first and second multi-scale monomer concentration decrease rate features and normalizing the recognition results to obtain a first state temporal memory matrix; using the first state temporal memory matrix to perform memory transfer interaction on the second multi-scale monomer concentration decrease rate to obtain a first interactive monomer concentration decrease rate feature; performing state temporal memory transfer interaction on the third multi-scale monomer concentration decrease rate feature in the multi-scale monomer concentration decrease rate feature sequence based on the first interactive monomer concentration decrease rate feature, and so on, to obtain the monomer concentration decrease rate feature.

[0076] Furthermore, the precise control system for the initiator addition rate in the OLED material synthesis is also used to: acquire multiple sample state temporal memory matrices, multiple sample multi-scale monomer concentration decrease rates, and corresponding multiple sample interactive monomer concentration decrease rate features as training data; use the training data to perform supervised training on the framework constructed based on the feedforward neural network until the training converges to obtain a memory transfer interactor; and identify the first state temporal memory matrix and the second multi-scale monomer concentration decrease rate based on the memory transfer interactor to obtain the first interactive monomer concentration decrease rate feature.

[0077] Furthermore, the precise control system for initiator addition rate in OLED material synthesis is also used for: acquiring historical OLED material synthesis records, extracting a set of historically compliant OLED material synthesis records with quality compliance from these records; identifying initiator addition scenarios and initiator addition rates on the set of historically compliant OLED material synthesis records to obtain a set of historically compliant reaction scenario representation vectors and a set of historical initiator addition rates; performing similar aggregation on the set of historically compliant reaction scenario representation vectors, and mapping and aggregating the set of historical initiator addition rates based on the aggregation results to obtain multiple sets of aggregation historically compliant reaction scenario representation vectors and multiple sets of aggregation historical initiator addition rates; performing representative screening on the multiple sets of aggregation historically compliant reaction scenario representation vectors to determine multiple screened historically compliant reaction scenario representation vectors; performing representative screening on the multiple sets of aggregation historical initiator addition rates to determine multiple screened historical initiator addition rates; and constructing a polymerization kinetics prediction model based on mapping and correlation between the multiple screened historically compliant reaction scenario representation vectors and the multiple screened historical initiator addition rates.

[0078] Furthermore, the precise control system for the initiator addition rate in OLED material synthesis is also used for: extracting a first set of historical compliance reaction scenario representation vectors from the plurality of historical compliance reaction scenario representation vector sets; extracting the historical compliance reaction scenario representation vector with the highest similarity to other historical compliance reaction scenario representation vectors in the first set of historical compliance reaction scenario representation vectors as an initial screening center; iterating on the initial screening center in the first set of historical compliance reaction scenario representation vectors according to a preset representative screening step size to obtain an iterative screening center, wherein the iterative screening center is the historical compliance reaction scenario representation vector in the first set of historical compliance reaction scenario representation vectors that is closest to the initial screening center in terms of Euclidean distance and preset representative screening step size; determining whether the iterative screening center can replace the initial screening center, and if so, continuing to iterate on the iterative screening center according to the preset representative screening step size until the maximum number of iterations is met to obtain the first screening historical compliance reaction scenario representation vector; if not, stopping the iteration and using the initial screening center as the first screening historical compliance reaction scenario representation vector; and adding the first screening historical compliance reaction scenario representation vector to the plurality of screening historical compliance reaction scenario representation vectors.

[0079] Furthermore, the precise control system for the initiator addition rate in the OLED material synthesis is also used to: count the number of historical compliant reaction scenario representation vectors in the first set of historical compliant reaction scenario representation vectors whose Euclidean distance to the iterative screening center is within a preset representative screening step size, and obtain the iterative screening coefficient; count the number of historical compliant reaction scenario representation vectors in the first set of historical compliant reaction scenario representation vectors whose Euclidean distance to the initial screening center is within a preset representative screening step size, and obtain the initial screening coefficient; when the iterative screening coefficient is greater than or equal to the initial screening coefficient, substitution is allowed; when the iterative screening coefficient is less than the initial screening coefficient, substitution is not allowed.

[0080] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The precise control method and specific examples of the initiator addition rate in OLED material synthesis in the foregoing embodiment 1 are also applicable to the precise control system of the initiator addition rate in OLED material synthesis in this embodiment. Through the foregoing detailed description of the precise control method of the initiator addition rate in OLED material synthesis, those skilled in the art can clearly understand the precise control system of the initiator addition rate in OLED material synthesis in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0081] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0082] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for precisely controlling the initiator addition rate in OLED material synthesis, characterized in that, The method includes: The synthesis process of OLED materials was monitored in real time using an online mass spectrometer and a UV-Vis spectrometer to obtain monomer concentration sequences and degree of polymerization sequences. By simultaneously acquiring the temperature, pressure, and stirring rate of the reactor using multiple source sensors, a temperature-pressure-stirring rate sequence is obtained. A state-temporal memory transfer interaction is performed on monomer concentration sequence, degree of polymerization sequence, and temperature-pressure-stirring rate sequence to generate a synthetic reaction scenario characterization vector; The polymerization kinetics prediction model is called to identify the initiator addition scenario of the synthesis reaction scenario characterization vector, obtain the optimal initiator addition rate, and generate control commands for the high-precision metering pump based on the optimal initiator addition rate. Specifically, state-time memory transfer interaction is performed on the monomer concentration sequence, degree of polymerization sequence, and temperature-pressure-stirring rate sequence to generate a synthetic reaction scenario characterization vector, including: The nearest neighbor concentration decrease rate and polymerization increase rate were identified from the monomer concentration sequence and degree of polymerization sequence to obtain the monomer concentration decrease rate sequence and the degree of polymerization increase rate sequence. Based on the monomer concentration decrease rate sequence, state temporal memory transfer interaction is performed to obtain monomer concentration decrease rate characteristics. Based on the polymerization degree rise rate sequence and the temperature-pressure-stirring rate sequence, state temporal memory transfer interaction is performed to obtain polymerization degree rise rate characteristics and temperature-pressure-stirring rate characteristics; Feature vector encoding is performed on the monomer concentration decrease rate characteristic, polymerization degree increase rate characteristic, and temperature-pressure-stirring rate characteristic to obtain a synthetic reaction scenario characterization vector; Based on the monomer concentration decrease rate sequence, a state-time memory transfer interaction is performed to obtain monomer concentration decrease rate characteristics, including: The monomer concentration decrease rate sequence is extracted and analyzed using a preset multi-scale state time series feature analyzer to obtain a multi-scale monomer concentration decrease rate feature set. The multi-scale monomer concentration decrease rate feature set is serialized according to the order of analysis scale from small to large to obtain the multi-scale monomer concentration decrease rate feature sequence. A state-time memory transfer interaction is performed on the multi-scale monomer concentration decrease rate feature sequence to obtain monomer concentration decrease rate features; Perform state-time memory transfer interaction on the multi-scale monomer concentration decrease rate feature sequence to obtain monomer concentration decrease rate features, including: Extract the first and second multi-scale monomer concentration decrease rate features from the multi-scale monomer concentration decrease rate feature sequence. The first-level multi-scale monomer concentration decrease rate feature and the second-level multi-scale monomer concentration decrease rate feature are subjected to the same type feature similarity identification, and the identification results are normalized to obtain the first state temporal memory matrix. The memory transfer interaction of the second-dimensional multi-scale monomer concentration decrease rate is performed using the first-state temporal memory matrix to obtain the first interaction monomer concentration decrease rate characteristics. Based on the first interactive monomer concentration decrease rate feature, the state temporal memory transfer interaction of the third multi-scale monomer concentration decrease rate feature in the multi-scale monomer concentration decrease rate feature sequence refers to using the updated interactive feature as the new state memory input, and performing the same similarity recognition, weight application and feature transfer process as the third feature, so that the information between scales diffuses sequentially, and so on, to obtain the monomer concentration decrease rate feature.

2. The method for precisely controlling the initiator addition rate in OLED material synthesis as described in claim 1, characterized in that, Also includes: Obtain the basic synthesis process chain and basic material information for OLED materials; Using the basic synthetic process chain and material information as indexes, big data mining of process data is performed to obtain sets of monomer concentration sequences, sets of polymerization degree sequences, and sets of temperature-pressure-stirring rate sequences. Traverse the set of mining monomer concentration sequence, mining degree of polymerization sequence, and mining temperature-pressure-stirring rate sequence to identify intra-sequence fluctuations and determine the set of mining monomer fluctuation bandwidth, mining degree of polymerization fluctuation bandwidth, and mining temperature-pressure-stirring rate fluctuation bandwidth. The mean drift of the set of the single-unit fluctuation bandwidth, the set of the fluctuation bandwidth of the degree of aggregation, and the set of the fluctuation bandwidth of the temperature-pressure-stirring rate of the excavation were respectively screened by the set of the mean drift to determine the representative fluctuation bandwidth of the single-unit fluctuation bandwidth, the representative fluctuation bandwidth of the degree of aggregation, and the representative fluctuation bandwidth of the temperature-pressure-stirring rate of the excavation. The minimum value among the representative fluctuation bandwidth of the mining monomer, the representative fluctuation bandwidth of the mining degree of polymerization, and the representative fluctuation bandwidth of the mining temperature-pressure-stirring rate is taken as the target monitoring bandwidth, and the target monitoring bandwidth is distributed to the online mass spectrometer, ultraviolet-visible spectrometer, and multi-source sensor.

3. The method for precisely controlling the initiator addition rate in OLED material synthesis as described in claim 1, characterized in that, Using the first-state temporal memory matrix, a memory transfer interaction is performed on the second-dimensional multi-scale monomer concentration decrease rate to obtain the first interaction monomer concentration decrease rate characteristics, including: Multiple sample state temporal memory matrices, multiple sample multi-scale monomer concentration decrease rates, and corresponding multiple sample interaction monomer concentration decrease rate features are obtained as training data. The training data is used to supervise the training of the framework built on the feedforward neural network until the training converges, and the memory transfer interactor is obtained. Based on the memory transfer interactor, the first state temporal memory matrix and the second bit multi-scale monomer concentration decrease rate are identified to obtain the first interactive monomer concentration decrease rate feature.

4. The method for precisely controlling the initiator addition rate in OLED material synthesis as described in claim 1, characterized in that, The polymerization kinetics prediction model is used to identify initiator addition scenarios from the synthesis reaction scenario characterization vector, and the optimal initiator addition rate is obtained, including: Obtain historical OLED material synthesis records, and extract a set of historically compliant OLED material synthesis records that meet quality standards from these records. The initiator addition scenario and initiator addition rate are identified in the historical compliant OLED material synthesis record set to obtain a historical compliant reaction scenario characterization vector set and a historical initiator addition rate set. The set of historical compliance response scenario representation vectors is aggregated in the same category, and the set of historical initiator addition rates is mapped and aggregated according to the aggregation results to obtain multiple aggregated historical compliance response scenario representation vector sets and multiple aggregated historical initiator addition rate sets. Representative screening is performed on the multiple aggregated historical compliance response scenario representation vector sets to determine multiple screened historical compliance response scenario representation vectors; Representative screening was performed on multiple sets of polymerization history initiator addition rates to determine multiple screened historical initiator addition rates; A polymerization kinetics prediction model is constructed by mapping and associating multiple historical compliance reaction scenario characterization vectors with multiple historical initiator addition rates.

5. The method for precisely controlling the initiator addition rate in OLED material synthesis as described in claim 4, characterized in that, Representative screening is performed on the multiple aggregated historical compliance response scenario representation vector sets to determine multiple selected historical compliance response scenario representation vectors, including: Extract the first aggregated historical compliance response scenario representation vector set from the multiple aggregated historical compliance response scenario representation vector sets; Extract the historical compliance response scenario representation vector with the highest similarity to other historical compliance response scenario representation vectors from the first aggregated historical compliance response scenario representation vector set, and use it as the initial screening center; According to a preset representative screening step size, the initial screening center is iterated in the first aggregated historical compliance response scenario representation vector set to obtain an iterative screening center. The iterative screening center is the historical compliance response scenario representation vector in the first aggregated historical compliance response scenario representation vector set that is closest to the initial screening center in terms of Euclidean distance and preset representative screening step size. Determine whether the iterative screening center can replace the initial screening center. If so, continue to iterate the iterative screening center according to the preset representative screening step size until the maximum number of iterations is met, and obtain the first screening historical compliance response scenario representation vector. If not, stop the iteration and use the initial screening center as the first screening historical compliance response scenario representation vector; Add the first screening historical compliance response scenario representation vector into multiple screening historical compliance response scenario representation vectors.

6. The method for precisely controlling the initiator addition rate in OLED material synthesis as described in claim 5, characterized in that, Determining whether the iterative screening center can replace the initial screening center includes: The number of historical compliance response scenario representation vectors in the first aggregated set of historical compliance response scenario representation vectors whose Euclidean distance to the iterative screening center is within a preset representative screening step size is counted to obtain the iterative screening coefficient. The number of historical compliance response scenario representation vectors in the first aggregated historical compliance response scenario representation vector set whose Euclidean distance to the initial screening center is within a preset representative screening step size range is counted to obtain the initial screening coefficient. When the iterative screening coefficient is greater than or equal to the initial screening coefficient, it can be replaced; When the iterative screening coefficient is less than the initial screening coefficient, substitution is not allowed.

7. A precise control system for the initiator addition rate in OLED material synthesis, characterized in that, The steps for implementing the method for precisely controlling the initiator addition rate in the synthesis of OLED materials according to any one of claims 1 to 6 include: The polymerization degree sequence acquisition module is used to monitor the OLED material synthesis process in real time using an online mass spectrometer and a UV-Vis spectrometer to obtain monomer concentration sequences and polymerization degree sequences; The temperature-pressure-stirring rate sequence acquisition module is used to simultaneously acquire the temperature, pressure and stirring rate of the reactor through multiple source sensors to obtain the temperature-pressure-stirring rate sequence. The synthetic reaction scenario characterization vector generation module is used to perform state-time memory transfer interaction on monomer concentration sequence, degree of polymerization sequence and temperature-pressure-stirring rate sequence to generate synthetic reaction scenario characterization vector; The control command generation module is used to call the polymerization kinetics prediction model to identify the initiator addition scenario from the synthesis reaction scenario characterization vector, obtain the optimal initiator addition rate, and generate control commands for the high-precision metering pump based on the optimal initiator addition rate.

Images