White carbon black modifier activity monitoring and early warning method based on Internet of Things

Abstract

The invention relates to the field of white carbon black modifier production, and particularly discloses an activity monitoring and early warning method of a white carbon black modifier based on Internet of Things, which comprises the following steps: acquiring reaction temperature, catalyst dosage and equipment operation parameters in real time through an Internet of Things sensor, and constructing a parameter and efficiency mapping table to form a multi-dimensional process database; encoding the temperature, the stirring rate and the raw material ratio into chromosome genes by utilizing a genetic algorithm, and constructing a multi-target model by taking the bifunctional grafting rate as an optimization target and combining the equipment load rate and the energy consumption constraint; a high-adaptability parameter set is generated by adopting iterative optimization, and a three-level response mechanism of local fine tuning, population resetting and historical optimal parameter loading is realized in combination with a hierarchical early warning system. According to the method, the problems that the synthesis conditions of the bifunctional silane coupling agent cannot be accurately controlled and the continuous production data response is delayed in the traditional process are solved, and the activity dynamic optimization and green and efficient production of the white carbon black modifier are realized.

Description

Technical Field

[0001] The present invention relates to the field of production of silica modifiers, and particularly to a method for active monitoring and early warning of silica modifiers based on the Internet of Things. Background Art

[0002] The application of traditional silane coupling agents in the silica / rubber composite system has long faced dual limitations in structural design and production processes. On the one hand, the molecular architectures of conventional mono-functional or simple bi-functional silane coupling agents are difficult to balance the synergistic requirements of filler dispersion and interfacial bonding: the design of a single functional group easily leads to uneven dispersion of silica or insufficient chemical bonding strength with the rubber matrix, while the synthesis processes of traditional bi-functional silanes often cannot accurately control the spatial distribution and reaction activity of functional groups, resulting in problems such as filler agglomeration or interfacial debonding during the mixing process. On the other hand, existing production processes generally rely on high-energy-consuming batch reactions and organic solvent systems, which not only have low production efficiency but also are accompanied by a large amount of volatile organic compound emissions, making it difficult to meet the requirements of green chemical engineering and large-scale continuous production.

[0003] In the prior art, the synthesis of silane coupling agents mostly adopts stepwise modification or simple condensation routes, which are difficult to achieve the efficient directional assembly of mercapto and alkoxy bifunctional groups, and the reaction conditions are harsh and prone to side reactions. At the same time, due to the lack of means for raw material dispersion homogenization and reaction mass transfer enhancement in traditional production devices, the batch stability of products is poor and the impurity residue rate is high. In addition, the solvent-dependent process and batch operation mode further exacerbate the energy consumption and environmental protection pressure, restricting its industrial application in the field of high-end rubber products. These problems urgently need to be solved through innovative design of molecular structures and systematic optimization of production technologies. Summary of the Invention

[0004] Aiming at the deficiencies of the prior art, the present invention provides a method for active monitoring and early warning of silica modifiers based on the Internet of Things, which is used to solve the problems that the prior art cannot improve the dispersion and interfacial bonding force of silica in rubber through controllable synthesis of bifunctional silane coupling agent (Si747), and at the same time achieve the efficiency of green and high-efficiency continuous production data processing.

[0005] To solve the above technical problems, the specific technical solution of the present invention is as follows: The present invention provides a method for active monitoring and early warning of silica modifiers based on the Internet of Things, including: Step S101, constructing a multi-dimensional process parameter database, and real-time collecting the reaction kettle pressure, catalyst dispersion degree and material residence time in the Internet of Things sensor network through an edge computing gateway, and synchronizing the process parameters to a central database; Step S102: Define a genetic algorithm fitness function based on the historical production data in the multi-dimensional process parameter database. Encode the reaction temperature, stirring rate, and raw material molar ratio as the chromosome gene sequence, and use the bifunctional group grafting rate as the main optimization target. Combine the constraint conditions of production energy consumption and equipment load rate to construct a multi-objective optimization model. Step S103: Dynamically adjust the search boundary of the genetic algorithm according to the raw material purity fluctuation and equipment aging coefficient collected in real time, and verify the feasibility of the adjusted parameter combination through the digital twin. Step S104: Perform iterative optimization of the genetic algorithm in the digital twin. Generate an optimized parameter set through operations such as initializing the population, crossover and mutation, fitness evaluation, and survival of the fittest, and push the optimized parameter set to the MES system to trigger parameter calibration of the production equipment. Step S105: Reverse verify the accuracy of the optimized parameter set by comparing the deviation between the predicted bifunctional group grafting rate in the digital twin and the real-time monitoring data, and trigger retraining of the genetic algorithm when the deviation exceeds the preset threshold. Step S106: Coordinate and control the optimized parameter set with the hierarchical warning system, and call the parameter adjustment strategy corresponding to the warning level according to the warning level. The parameter adjustment strategy includes local parameter fine-tuning scheme, reset of the initial population of the genetic algorithm, and loading of the historical optimal parameter combination.

[0006] Furthermore, for the method for monitoring and warning the activity of the silica white modifier based on the Internet of Things according to the present invention, the construction of the multi-dimensional process parameter database includes: Establish a parameter-effectiveness mapping table in the central database, and the parameter-effectiveness mapping table associates the corresponding relationships between the temperature, catalyst dosage, and bifunctional group grafting rate in the historical production batches. Denoise and standardize the collected original sensing data through the edge computing gateway to generate a structured data set that meets the training requirements of the genetic algorithm.

[0007] Furthermore, for the method for monitoring and warning the activity of the silica white modifier based on the Internet of Things according to the present invention, the definition of the genetic algorithm fitness function includes: Based on the data blood relationship mapping model in the parameter-effectiveness mapping table, identify the coupling weight relationship between the reaction temperature and the catalyst dispersion degree. Use the bifunctional group grafting rate as the main optimization target, and use the dynamic thresholds of unit production energy consumption and equipment load rate as the constraint conditions to construct a multi-dimensional fitness evaluation model.

[0008] Further, for the method for monitoring and warning the activity of the silica modifier based on the Internet of Things according to the present invention, dynamically adjusting the search boundary of the genetic algorithm according to the raw material purity fluctuation and equipment aging coefficient collected in real time, and verifying the feasibility of the adjusted parameter combination through the digital twin includes: In response to the decrease in the intensity of the mercapto characteristic peak monitored by infrared spectroscopy, limit the temperature adjustment range to the safe interval set based on the material thermal stability experiment; When the real-time carbon emission reaches 95% of the preset threshold stored in the central database, dynamically lower the preset maximum value of the catalyst dosage; Simulate and run the parameter combination through the digital twin, and filter out the invalid parameter combinations that cause equipment overload or functional group oxidation.

[0009] Further, for the method for monitoring and warning the activity of the silica modifier based on the Internet of Things according to the present invention, performing iterative optimization of the genetic algorithm includes: Extract the parental chromosomes from the historical optimal parameter combination to generate the initial population; Within the dynamically adjusted search boundary, perform gene segment crossover on the chromosomes according to the crossover probability of 0.7 - 0.9, and perform random mutation according to the mutation probability of 0.01 - 0.05; Inject the mutated parameter combination into the digital twin for simulation operation, and calculate the grafting rate through the activity evaluation model to generate the fitness value; Retain the individuals with the top preset proportion of fitness rankings and iterate to the next generation until the optimization parameter set converges.

[0010] Further, for the method for monitoring and warning the activity of the silica modifier based on the Internet of Things according to the present invention, verifying the accuracy of the reverse verification optimization parameter set includes: Calculate the deviation rate between the digital twin predicted grafting rate and the actual infrared monitoring data in real time; When the deviation rate exceeds ±5%, extract the current production environment characteristic data and retrain the genetic algorithm model; Feed back the execution error of the equipment control parameters to the genetic algorithm to correct the equipment response coefficient model.

[0011] Further, for the method for monitoring and warning the activity of the silica modifier based on the Internet of Things according to the present invention, coordinating the control of the optimization parameter set with the hierarchical warning system, and calling the parameter adjustment strategy corresponding to the warning level according to the warning level, the parameter adjustment strategy includes local parameter fine-tuning scheme, resetting the initial population of the genetic algorithm, and loading the historical optimal parameter combination includes: When the deviation rate ≤ 5%, trigger a first-level warning, and call the genetic algorithm to generate a temperature compensation or stirring rate fine-tuning scheme; When the deviation rate is between 5% and 10%, a secondary warning is triggered, the fault source is located by combining abnormal conduction analysis, and the initial population search range of the genetic algorithm is reset; When the deviation rate > 10%, a tertiary warning is triggered, and the historical optimal parameter combination is loaded from the central database to start the resumption of production process.

[0012] Furthermore, for the method for monitoring and warning the activity of the silica white modifier based on the Internet of Things according to the present invention, the construction of the multi-dimensional fitness evaluation model further includes: According to the device aging coefficient collected in real time, based on the data blood relationship mapping model in the parameter and efficiency mapping table, the weight coefficient of the device load rate constraint condition is dynamically adjusted; Based on the real-time raw material purity data, the prediction calculation formula of the double functional group grafting rate is corrected.

[0013] Furthermore, for the method for monitoring and warning the activity of the silica white modifier based on the Internet of Things according to the present invention, the filtering of invalid parameter combinations that cause equipment overload or functional group oxidation includes: Pre-load the maximum pressure bearing value of the device and the critical value of material thermal decomposition in the digital twin; When the simulated operation parameter combination triggers the pre-loaded maximum pressure bearing value of the device or the critical value of material thermal decomposition in the digital twin, mark this parameter combination as an invalid solution and terminate the fitness evaluation.

[0014] Furthermore, for the method for monitoring and warning the activity of the silica white modifier based on the Internet of Things according to the present invention, the re-training of the genetic algorithm model includes: Extract the raw material purity, environmental temperature and humidity, and equipment operation log data during the period when the deviation exceeds the threshold; Use the extracted data as new training samples to update the parameter and efficiency mapping table; Adopt an online weight update algorithm based on the new data to dynamically adjust the priority weights of the reaction temperature and catalyst dosage in the chromosome coding rule.

[0015] Advantages of the present invention Through real-time collection of production data by the Internet of Things and dynamic optimization of the genetic algorithm, the present invention breaks through the limitations of traditional empirical parameter adjustment, realizes precise control of the double functional group grafting rate. The digital twin technology pre-verifies the parameter combination, avoids high-risk trial and error, significantly shortens the process optimization cycle, and at the same time ensures production safety and stability.

[0016] The present invention dynamically adjusts the algorithm boundaries and constraint conditions based on real-time data such as equipment aging coefficients and raw material purity fluctuations, enabling process parameters to always adapt to the current working conditions and reducing performance fluctuations caused by environmental changes. Combining carbon emission threshold control with solvent-free process design effectively reduces production energy consumption and pollutant emissions, promoting green and sustainable production.

[0017] Through the full-link closed-loop control of "data acquisition → algorithm optimization → virtual verification → execution feedback", the present invention eliminates data islands and response lags in traditional production. The hierarchical early warning system realizes multi-level emergency responses from local fine-tuning to global resumption of production, taking into account both production continuity and abnormal handling efficiency, providing intelligent technical support for the efficient and stable production of silica white modifiers. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, for those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on the drawings.

[0019] Figure 1 It is a system architecture diagram of the method for monitoring and warning the activity of silica white modifiers based on the Internet of Things provided by the embodiments of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0020] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the specific embodiments and corresponding drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention. The technical solutions provided by each embodiment of the present invention are described in detail below with reference to the drawings.

[0021] To better understand the objectives of the present invention, the present invention will be further described in detail below.

[0022] As Figure 1 shown, the present invention provides a method for monitoring and warning the activity of silica white modifiers based on the Internet of Things, including: Step S101, constructing a multi-dimensional process parameter database, and real-time collecting the reactor pressure, catalyst dispersion degree, and material residence time in the Internet of Things sensing network through an edge computing gateway, and synchronizing the process parameters to a central database; Step S102: Define a genetic algorithm fitness function based on the historical production data in the multi-dimensional process parameter database. Encode the reaction temperature, stirring rate, and raw material molar ratio as the chromosome gene sequence, and use the bifunctional group grafting rate as the main optimization objective. Combine the constraint conditions of production energy consumption and equipment load rate to construct a multi-objective optimization model. Step S103: Dynamically adjust the search boundary of the genetic algorithm according to the raw material purity fluctuation and equipment aging coefficient collected in real time, and verify the feasibility of the adjusted parameter combination through the digital twin. Step S104: Perform iterative optimization of the genetic algorithm in the digital twin. Generate an optimized parameter set through operations such as initializing the population, crossover and mutation, fitness evaluation, and survival of the fittest, and push the optimized parameter set to the MES system to trigger parameter calibration of the production equipment. Step S105: Reverse-verify the accuracy of the optimized parameter set by comparing the deviation between the predicted bifunctional group grafting rate in the digital twin and the real-time monitoring data, and trigger retraining of the genetic algorithm when the deviation exceeds the preset threshold. Step S106: Coordinate and control the optimized parameter set with the hierarchical early warning system, and call the parameter adjustment strategy corresponding to the early warning level according to the early warning level. The parameter adjustment strategy includes local parameter fine-tuning schemes, resetting the initial population of the genetic algorithm, and loading the historical optimal parameter combination.

[0023] The purpose of this technical solution is to solve the problem of precise control of the bifunctional group grafting rate (the core index reflecting the coupling agent efficiency) in the production process of silica white modifiers through the integration of Internet of Things, genetic algorithm, and digital twin technology, and achieve intelligent early warning of abnormal working conditions and parameter self-adjustment. The following analyzes its technical logic by module: Data collection (hardware layer → data layer); Multi-dimensional process parameter database: Integrate the data of the entire production process, including key parameters such as reactor pressure, catalyst dispersion (reflecting catalytic efficiency), and material residence time (affecting reaction sufficiency).

[0024] Edge computing gateway: Deployed at the device end, it collects Internet of Things sensor data in real time (such as pressure transmitters, torque sensors, etc.) to avoid data lag caused by network latency.

[0025] Central database: Store historical production data and real-time data, providing a data basis for subsequent algorithm optimization.

[0026] Technical value: Eliminate data islands in traditional production and ensure the real-time and integrity of parameter collection.

[0027] Construction of multi-objective optimization model (algorithm layer); Genetic algorithm fitness function: Encode production parameters (temperature, stirring rate, raw material molar ratio) as "chromosome genes" and optimize parameter combinations by simulating the biological evolution process.

[0028] Main optimization objective: Maximize the grafting rate of bifunctional groups (which determines the interfacial bonding strength between silica and rubber).

[0029] Constraints: Production energy consumption (such as heating power), equipment load rate (such as the life limit of the stirring motor).

[0030] Multi-objective optimization model: Balance the contradiction between improving the grafting rate and cost / equipment loss through weight allocation.

[0031] Technical value: Break through the limitations of traditional empirical parameter tuning and achieve global optimal solution search under multiple constraints.

[0032] Dynamic parameter adjustment and verification (dynamic optimization layer); Real-time perturbation response: According to fluctuations in raw material purity (such as excessive MPS impurities) and equipment aging coefficient (such as the decline in the heat transfer efficiency of the reaction kettle), dynamically adjust the search boundary of the genetic algorithm (such as the allowable temperature range is reduced from ±5°C to ±2°C).

[0033] Digital twin verification: Simulate the production effect after parameter adjustment in the virtual model, predict whether it will cause equipment overload or reaction out of control, and avoid the risk of direct trial and error.

[0034] Technical value: Enable the optimization algorithm to have "adaptive ability" and adapt to the dynamic changes of the production environment.

[0035] Iterative optimization and equipment calibration (execution layer); Genetic algorithm iteration process: Initial population: Generate diverse parameter combinations (such as different temperature-stirring rate pairings).

[0036] Crossover and mutation: Explore better solutions through random exchange and fine-tuning of parameter combinations.

[0037] Fitness evaluation: Score based on the grafting rate and energy consumption data predicted by the digital twin.

[0038] Survival of the fittest: Retain high-score parameter combinations and eliminate inefficient solutions.

[0039] MES system linkage: Push the optimized parameter set (such as the best temperature setting value) to the manufacturing execution system to automatically calibrate the equipment (such as adjusting the heating module of the reaction kettle).

[0040] Technical value: Achieve a closed-loop control of "algorithm optimization → equipment execution" and reduce manual intervention delay.

[0041] Reverse verification and algorithm retraining (feedback layer); Deviation monitoring: Compare the deviation between the digital twin predicted grafting rate and the actual sensor data to evaluate the accuracy of parameter optimization.

[0042] Threshold trigger mechanism: When the deviation exceeds the preset value (such as ±3%), it is determined that the model is inaccurate (possibly due to unmodeled factors such as catalyst activity decay), and the genetic algorithm is triggered to retrain (update the historical data sample library).

[0043] Technical value: Ensure that the algorithm continuously adapts to production process changes and avoid model degradation after long-term operation.

[0044] Hierarchical early warning and collaborative control (decision-making layer); Definition of early warning levels: Level 1 early warning (slight deviation): Trigger local parameter fine-tuning (such as ±2°C temperature compensation).

[0045] Level 2 early warning (moderate deviation): Reset the initial population of the genetic algorithm and re-search for the optimal solution.

[0046] Level 3 early warning (severe anomaly): Load the historical optimal parameter combination to force stable production.

[0047] Strategy coordination: Call the corresponding strategy according to the early warning level to balance production stability and optimization efficiency.

[0048] Technical value: Build a multi-level emergency response mechanism to ensure production continuity and safety.

[0049] The present invention combines data-driven optimization (genetic algorithm) and virtual-real interaction verification (digital twin) to solve the pain points of traditional production relying on fixed process parameters and being unable to dynamically adapt to changes in raw materials / equipment status. Its core innovation lies in: Dynamic boundary adjustment: Enable the optimization algorithm to respond to real-time disturbances; Closed-loop feedback mechanism: Continuously improve control accuracy through the "prediction - execution - verification" cycle; Hierarchical decision-making system: Achieve intelligent upgrading from parameter fine-tuning to system-level intervention. Finally, achieve stable improvement of the bifunctional group grafting rate (process target) and active prevention and control of production risks (management and control target).

[0050] Specifically, for the method for monitoring and warning the activity of silica white modifier based on the Internet of Things described in the present invention, the construction of the multi-dimensional process parameter database includes: Establish a parameter - efficiency mapping table in the central database, and the parameter - efficiency mapping table associates the corresponding relationships between temperature, catalyst dosage, and bifunctional group grafting rate in historical production batches; The original sensing data collected is denoised and standardized by the edge computing gateway to generate a structured data set that meets the training requirements of the genetic algorithm.

[0051] In the present invention, a parameter - efficacy mapping table is constructed. The mapping table stores in a structured data form the corresponding relationships reflecting temperature, catalyst dosage, stirring rate, and bifunctional group grafting rate, and generates parameter optimization association rules through a machine - learning model to form a multi - dimensional process database, which can intuitively reflect the influence law of different parameter combinations on the modifier performance. For example, when the catalyst dosage in a certain batch of production deviates from the historical optimal range, the system can quickly predict the change trend of the grafting rate by querying the mapping table, providing data support for the subsequent optimization algorithm. This design breaks through the limitation of traditional parameter adjustment relying on manual experience and realizes the digital association between process parameters and efficacy indicators.

[0052] Aiming at the complexity and noise interference problems of Internet of Things sensing data, the present invention pre - processes the original data through an edge computing gateway. First, a denoising algorithm is used to eliminate abnormal fluctuations in the sensor signal (such as instantaneous jumps caused by electromagnetic interference), and then through standardization processing, the dimensions and numerical ranges of parameters such as temperature and catalyst dosage are unified to generate a structured data set suitable for genetic algorithm training. This hierarchical processing mechanism not only ensures data quality but also avoids excessive consumption of cloud computing resources, enabling the genetic algorithm to efficiently mine process parameter combinations with high grafting rate and low energy consumption, laying a reliable data foundation for activity monitoring and dynamic warning.

[0053] Specifically, for the activity monitoring and warning method of the silica modifier based on the Internet of Things, the defined genetic algorithm fitness function includes: Based on the data lineage mapping model in the parameter - efficacy mapping table, identify the coupling weight relationship between reaction temperature and catalyst dispersion degree; Take the bifunctional group grafting rate as the main optimization target, and take the dynamic thresholds of unit production energy consumption and equipment load rate as constraint conditions to construct a multi - dimensional fitness evaluation model.

[0054] The present invention analyzes the dynamic correlation relationship between reaction temperature and catalyst dispersion degree through a data lineage mapping model, and clarifies the synergistic influence mechanism of the two on the bifunctional group grafting rate. This model is based on historical production data, tracking the causal dependence and interaction between parameters. For example, high temperature may lead to uneven catalyst dispersion, thus reducing the grafting efficiency. By quantifying the coupling weight (such as the contribution degree of temperature fluctuation to catalyst dispersion degree), the genetic algorithm can more accurately screen parameter combinations, avoiding the problem of local optimal solutions caused by single - parameter optimization and laying a foundation for global multi - objective optimization.

[0055] In the design of the fitness function, the present invention takes the maximization of the grafting rate of bifunctional groups as the core optimization goal, and at the same time introduces the dynamic thresholds of energy consumption per unit output and equipment load rate as constraint conditions to construct a multi-dimensional evaluation model. This design takes into account both process efficiency and production sustainability: on the one hand, it ensures that the activity of the coupling agent meets the standard, and on the other hand, it prevents equipment overload or energy waste through dynamic threshold control (such as the equipment load rate not exceeding the safety upper limit). The algorithm automatically balances the main goal and constraint conditions during the iteration process, generates an optimized parameter set that not only meets the grafting rate requirements but also conforms to the actual production constraints, and realizes the collaborative optimization of high efficiency and safety.

[0056] Specifically, for the method for monitoring and warning the activity of the silica modifier based on the Internet of Things according to the present invention, dynamically adjusting the search boundary of the genetic algorithm according to the raw material purity fluctuation and equipment aging coefficient collected in real time, and verifying the feasibility of the adjusted parameter combination through a digital twin body includes: In response to the decrease in the intensity of the mercapto characteristic peak monitored by infrared spectroscopy, limit the temperature adjustment range to a safe interval set based on the material thermal stability experiment; When the real-time carbon emission reaches 95% of the preset threshold stored in the central database, dynamically lower the preset maximum value of the catalyst dosage; Simulate the operation of the parameter combination through the digital twin body, and filter out the invalid parameter combinations that cause equipment overload or functional group oxidation. The digital twin body is constructed based on the physical model of the reaction kettle and historical production data, simulates the reaction temperature and pressure distribution through finite element analysis, and combines machine learning to predict the grafting rate of functional groups to verify the feasibility of the parameter combination.

[0057] When the infrared spectroscopy monitors that the decrease in the intensity of the mercapto characteristic peak > 15%, the system limits the temperature adjustment range to the preset safe interval (90~105°C); at the same time, when the carbon emission sensor detects that the real-time emission reaches 95% of the preset threshold (such as 100 kg / batch), dynamically lower the maximum value of the catalyst dosage to 1.2 wt% (originally 1.5 wt%).

[0058] The present invention realizes the intelligent adjustment of the search boundary of the genetic algorithm through real-time monitoring and dynamic constraint mechanism. When the infrared spectroscopy detects a decrease in the intensity of the mercapto characteristic peak (reflecting an increase in the risk of mercapto oxidation), the system automatically limits the temperature adjustment range to the preset safe interval (such as 90~105°C) to avoid the inactivation of functional groups caused by out-of-control temperature. At the same time, for environmental protection requirements, when the carbon emission approaches the threshold, the algorithm dynamically lowers the upper limit of the catalyst dosage to balance the reaction efficiency and environmental protection indicators. This boundary constraint mechanism based on real-time working conditions ensures that the optimization process always meets the requirements of process safety and green production.

[0059] To further enhance the reliability of parameter optimization, the present invention utilizes digital twins to virtually verify the parameter combinations generated by genetic algorithms. By simulating the reaction process, invalid parameter combinations that may lead to equipment overload (such as the stirrer motor operating at excessive power) or functional group oxidation (such as side reactions caused by high temperatures) can be quickly identified and filtered in advance. This design not only avoids the production risks brought by traditional trial-and-error methods but also significantly reduces the effective search space of genetic algorithms, making the optimization process efficient and safe, and ensuring that the final parameter set can be directly applied to the actual production line.

[0060] Specifically, for the active monitoring and warning method of silica modifier based on the Internet of Things described in the present invention, the execution of genetic algorithm iterative optimization includes: Extract parental chromosomes from the historical optimal parameter combinations to generate an initial population; Within the dynamically adjusted search boundaries, perform gene segment crossover on the chromosomes according to a crossover probability of 0.7 - 0.9 and perform random mutation according to a mutation probability of 0.01 - 0.05; Inject the mutated parameter combinations into the digital twin for simulation operation, and calculate the grafting rate through the activity evaluation model to generate fitness values; Retain the individuals with the top preset proportion of fitness rankings and iterate to the next generation until the optimized parameter set converges.

[0061] Technical analysis of genetic algorithms in the present invention; Basic principles of genetic algorithms; The genetic algorithm (Genetic Algorithm, GA) is a global optimization algorithm based on the principles of biological evolution, and its core mechanisms include: Coding mechanism: Convert the parameters to be optimized into chromosome gene sequences (binary / real number coding); Population evolution: Iteratively generate new parameter combinations through selection, crossover, and mutation operations; Survival of the fittest: Select high-quality individuals based on the fitness function and eliminate inefficient solutions; Convergence criterion: Terminate the optimization when the population fitness tends to be stable or reaches the preset number of iterations.

[0062] Specific technical integration in the present invention; Parameter coding and population initialization; Chromosome construction: Encode process parameters such as reaction temperature (80 - 120 °C), stirring rate (600 - 1,200 rpm), and raw material molar ratio (MPS:TEMS = 1:1.1 - 1.3) into chromosome gene sequences. For example: [temperature gene | stirring rate gene | molar ratio gene] → [100 °C, 850 rpm, 1:1.2] Initial population generation: Extract the parental chromosomes from the historical optimal parameter combinations in the central database to form an initial population (scale: 50 - 100 groups) with high fitness characteristics.

[0063] Design of the target fitness function; The fitness function is defined as:

[0064] Where: G: The grafting rate of bifunctional groups (the main optimization target, weight α = 0.6); E: Energy consumption per unit output (constraint condition, weight β = 0.25); L: Equipment load rate (constraint condition, weight γ = 0.15); Dynamically adjust the weight coefficients through the data lineage mapping model. For example, when the equipment aging coefficient > 0.8, increase γ to 0.2 to strengthen equipment protection.

[0065] State boundary constraints and evolutionary operations: Cross - over and mutation optimization: The cross - over probability Pc = 0.7 - 0.9, and the two - point cross - over method is used to exchange gene segments; The mutation probability Pm = 0.01 - 0.05, and the parameter values are randomly adjusted within the dynamic search boundary.

[0066] Example: When the intensity of the mercapto characteristic peak monitored by infrared spectroscopy decreases by > 15%, limit the mutation range of the temperature gene to [90℃, 105℃].

[0067] Digital twin pre - verification: Inject the mutated parameter combinations into the digital twin for virtual production verification, and filter out the invalid solutions that trigger the following conditions: Reactor pressure > 2.5MPa (equipment pressure - bearing threshold) Material thermal decomposition temperature > 130℃ (thermal stability critical value) Iterative optimization and feedback correction; Elite retention strategy: Retain the top 20% of individuals with the highest fitness in each generation and directly transfer them to the next generation; Convergence determination: Terminate the iteration when the fluctuation of the optimal fitness in 5 consecutive generations < 1%; Model re - training: When the deviation between the predicted grafting rate of the digital twin and the actual value > 5%: Extract the current production environment data to update the parameter and performance mapping table Adopt incremental learning to adjust the chromosome coding rule (such as adding a gene locus for catalyst dispersion degree) Genetic algorithm response actions: Level - 1 warning, start local fine - tuning (temperature compensation ±2℃).

[0068] Secondary warning, reset the initial population and narrow the search range.

[0069] Tertiary warning, load the historical optimal parameter combination to reset the production line.

[0070] By deeply coupling the genetic algorithm with Internet of Things sensing data and digital twin verification, the present invention realizes the dynamic global optimization of the production process parameters of the silica modifier, breaking through the limitations of the traditional empirical parameter adjustment mode in complex production environments.

[0071] Through the iterative optimization mechanism of the genetic algorithm, the present invention realizes the intelligent screening of the production process parameters of the silica modifier. The algorithm constructs an initial population with the historical optimal parameter combination as the parental chromosome to ensure that the search starting point has the gene characteristics of a high grafting rate. Within the dynamically adjusted search boundary, through gene fragment crossover recombination and random mutation operations, diverse parameter combinations (such as temperature - catalyst dosage pairing) are generated, which not only inherit historical experience but also explore potential optimization spaces. This design significantly improves the convergence efficiency of the algorithm and avoids the blindness of traditional random searches.

[0072] During the iteration process, the mutated parameter combinations are first virtually verified in the digital twin. The effectiveness indicators such as the grafting rate of bifunctional groups are predicted through the activity evaluation model, and the fitness value (combining weights such as grafting rate and energy consumption) is calculated. The system only retains the individuals with the top fitness rankings to enter the next iteration, gradually eliminating inefficient solutions until the parameter set converges to the global optimal solution. This "simulation screening → actual application" closed-loop optimization mode not only reduces the production trial - and - error cost but also ensures the reliability of the parameter set and the process stability.

[0073] Specifically, for the activity monitoring and warning method of the silica modifier based on the Internet of Things described in the present invention, the accuracy of the reverse verification and optimization parameter set includes: Real - time calculate the deviation rate between the digital twin predicted grafting rate and the actual infrared monitoring data; When the deviation rate exceeds ±5%, extract the characteristic data of the current production environment and retrain the genetic algorithm model; Feed back the execution error of the equipment control parameters to the genetic algorithm to correct the equipment response coefficient model.

[0074] Through a real-time deviation rate monitoring mechanism, the present invention verifies the consistency between the prediction accuracy of the digital twin model and actual production. The system continuously compares the grafting rate predicted by the digital twin with the measured data of infrared spectroscopy. When the deviation rate exceeds a preset threshold (such as ±5%), it indicates that the current optimization parameter set or environmental conditions have changed (such as catalyst activity decay, raw material batch differences). At this time, the system automatically extracts the current environmental characteristic data (such as temperature and humidity, equipment aging status), triggers the retraining of the genetic algorithm model, ensures that the algorithm adapts to the latest working conditions, and avoids the problem of model failure caused by long-term operation.

[0075] To further improve the parameter control accuracy, the present invention introduces a closed-loop feedback mechanism for equipment execution errors. When there is a deviation between the actual execution parameters of the equipment (such as temperature adjustment, catalyst dosage) and the algorithm instructions (caused by mechanical delay or sensor drift), the system inputs the error data back into the genetic algorithm to dynamically correct the equipment response coefficient model (such as the mathematical relationship between heating power and heating rate). This "execution - feedback - correction" closed-loop design enables the algorithm to accurately predict the actual effect of parameter adjustment, thereby ensuring the seamless connection of the optimization parameter set from theoretical optimum to practical usability.

[0076] Specifically, for the active monitoring and early warning method of silica modifier based on the Internet of Things described in the present invention, the optimization parameter set is cooperatively controlled with a hierarchical early warning system, and parameter adjustment strategies corresponding to the early warning level are called according to the early warning level. The parameter adjustment strategies include local parameter fine-tuning schemes, resetting the initial population of the genetic algorithm, and loading historical optimal parameter combinations, including: When the deviation rate ≤ 5%, a first-level early warning is triggered, and the genetic algorithm is called to generate a temperature compensation or stirring rate fine-tuning scheme; When the deviation rate is between 5% and 10%, a second-level early warning is triggered, the fault source is located by combining abnormal conduction analysis, and the search range of the initial population of the genetic algorithm is reset; When the deviation rate > 10%, a third-level early warning is triggered, and the historical optimal parameter combination is loaded from the central database to start the resumption of production process.

[0077] First-level early warning: Triggering condition: The deviation rate between the grafting rate predicted by the digital twin and the real-time monitoring data ≤ 5%, or the equipment load rate ≤ 75%.

[0078] Response action: Call the genetic algorithm to generate a temperature compensation (±2°C) or stirring rate fine-tuning (±50 rpm).

[0079] Second-level early warning: Triggering condition: The grafting rate deviation rate is 5% - 10%, or the equipment load rate is 75% - 90%, or the carbon emission reaches the threshold of 90% - 95%.

[0080] Response action: Reset the initial population of the genetic algorithm and narrow the parameter search range (e.g., adjust the temperature range from ±5°C to ±2°C).

[0081] Level 3 warning: Trigger conditions: The grafting rate deviation rate > 10%, or the equipment load rate > 90%, or the carbon emissions ≥ the threshold value, or the pressure of key equipment (such as the reaction kettle) exceeds the limit.

[0082] Response action: Immediately stop the machine, load the historical optimal parameter combination to restart the production line, and push a maintenance alarm.

[0083] The warning escalation logic is as follows: Time accumulation mechanism: If the level 1 warning is continuously triggered 3 times within 30 minutes, it will automatically escalate to a level 2 warning.

[0084] Compound condition trigger: When both the grafting rate deviation rate > 8% and the equipment load rate > 85% are satisfied simultaneously, directly trigger a level 2 warning.

[0085] The present invention realizes differential responses for different risk levels through a hierarchical warning mechanism. When a level 1 warning is triggered (such as a small fluctuation in the grafting rate), the system calls the genetic algorithm to quickly generate a temperature compensation or stirring rate fine-tuning scheme, and restores production stability through local parameter adjustment. If it is upgraded to a level 2 warning (such as equipment abnormality or continuous parameter deviation), the system combines the abnormal conduction analysis model to locate the fault source (such as catalyst feeding blockage or sensor failure), and resets the initial population search range of the genetic algorithm, focusing on the parameter space matching the current working conditions to improve the optimization efficiency.

[0086] In the scenario of a level 3 warning (such as forced shutdown due to equipment overload), the system automatically loads the historical optimal parameter combination from the central database and quickly restarts the production process based on the verified efficient process parameters. This design forms a closed-loop controllable warning management system through a multi-level response mechanism of "dynamic fine-tuning → fault tracing → emergency recovery", which not only avoids excessive intervention in case of minor abnormalities but also ensures the production restart efficiency in case of serious failures.

[0087] Specifically, for the method for monitoring the activity and warning of the silica modifier based on the Internet of Things described in the present invention, the construction of the multi-dimensional fitness evaluation model further includes: According to the device aging coefficient collected in real time, based on the data blood relationship mapping model in the parameter and efficiency mapping table, dynamically adjust the weight coefficient of the device load rate constraint condition; Modify the prediction calculation formula of the bifunctional group grafting rate based on the real-time raw material purity data.

[0088] Through the dynamic adjustment mechanism of the equipment aging coefficient, the present invention enhances the response ability of the fitness evaluation model to the state of production equipment. As the equipment operation time increases (such as the decline of the heat transfer efficiency of the reaction kettle or the wear of the stirring paddle), the system analyzes the aging degree based on the sensor data, automatically reduces the weight coefficient of the equipment load rate constraint, and allows the algorithm to prioritize ensuring the grafting rate target rather than forcing full-load production. This dynamic weight allocation mechanism not only extends the service life of the equipment but also avoids the problem of misjudgment of process parameters caused by the attenuation of mechanical properties.

[0089] Regarding the influence of raw material purity fluctuations on the activity of the coupling agent, the present invention embeds real-time raw material purity data (such as the impurity content of MPS) into the grafting rate prediction model and dynamically corrects the correction factor of the calculation formula. For example, when the raw material purity is lower than the standard value, the model automatically reduces the expected value of the theoretical grafting rate and guides the genetic algorithm to prioritize optimizing adjustable parameters such as the catalyst ratio. This design realizes the real-time adaptation of the process model to the raw material characteristics, ensures that the parameter optimization scheme is always based on the actual production conditions, and avoids the systematic deviation between the theoretical value and the actual efficiency.

[0090] Specifically, in the method for monitoring and warning the activity of the silica white modifier based on the Internet of Things according to the present invention, the invalid parameter combinations that cause equipment overload or functional group oxidation by filtration include: preloading the maximum pressure-bearing value of the equipment and the critical value of material thermal decomposition in the digital twin; When the simulated operation parameter combination triggers the preloaded maximum pressure-bearing value of the equipment or the critical value of material thermal decomposition in the digital twin, mark this parameter combination as an invalid solution and terminate the fitness evaluation.

[0091] The present invention constructs a safety boundary constraint in the virtual environment by preloading the maximum pressure-bearing value of the equipment (such as the pressure resistance threshold of the reaction kettle) and the critical value of material thermal decomposition (such as the thermal stability limit of the silane coupling agent) in the digital twin. When the simulated operation parameter combination (such as high temperature and high pressure setting) reaches these physical limits, the system automatically marks this combination as an invalid solution and directly terminates the fitness evaluation process. This mechanism avoids extreme parameter combinations that may cause equipment damage or material decomposition from the source, ensuring that the genetic algorithm only optimizes and screens safe and feasible solutions.

[0092] By dynamically filtering invalid parameter combinations, the present invention greatly improves the algorithm optimization efficiency and production safety. For example, if a parameter combination triggers an equipment pressure warning due to excessive pressure setting, the system can quickly eliminate this solution without waiting for actual production verification, avoiding the waste of algorithm resources on high-risk solutions. At the same time, combined with real-time physical property data (such as the thermal stability difference of raw material batches), the critical value of thermal decomposition is dynamically updated, making the filtering logic always synchronized with the current production conditions, ensuring the accuracy and reliability of virtual verification.

[0093] Specifically, for the method for active monitoring and early warning of the silica modifier based on the Internet of Things according to the present invention, the retrained genetic algorithm model includes: Extract the raw material purity, environmental temperature and humidity, and equipment operation log data during the period when the deviation exceeds the threshold; Use the extracted data as new training samples to update the parameter and performance mapping table; Adopt an online weight update algorithm based on the new data to dynamically adjust the priority weights of the reaction temperature and catalyst dosage in the chromosome coding rule.

[0094] Through the model update mechanism driven by dynamic data, the present invention realizes the continuous optimization of the genetic algorithm. When the deviation rate exceeds the threshold, the system automatically extracts the raw material purity, environmental temperature and humidity, and equipment operation status data during the abnormal period to form new training samples reflecting the current working conditions. By synchronously updating these data to the parameter and performance mapping table, the model can capture the real-time impact of raw material fluctuations or environmental disturbances on performance, enabling the genetic algorithm to adapt to the dynamic changes of the production process and avoiding optimization deviations caused by data aging.

[0095] To improve the algorithm iteration efficiency, the present invention optimizes the chromosome coding rule in an incremental learning manner. Based on the new data features (such as the type of raw material impurities or the trend of equipment performance decay), the coding weights and priorities of parameters such as temperature and catalyst dosage in the chromosome are dynamically adjusted. This optimization enables the genetic algorithm to quickly focus on the key parameter combinations under the current working conditions, avoiding the resource consumption of repeated training with full-scale data, and ensuring that the model iteration process is always synchronized with the actual production requirements.

[0096] Specific embodiments of the present invention: Embodiment 1: Continuous production process of bifunctional silane coupling agent Si747 based on the Internet of Things This embodiment will realize the green and efficient production of Si747 in combination with the Internet of Things monitoring system, and specifically includes the following steps: Reaction formula example:

[0097] Raw material pretreatment: Purification of mercapto silane monomer: Using MPS (mercaptopropyltrimethoxysilane) as the raw material, the impurity content is monitored in real time through a molecular sieve adsorption tower connected to the Internet of Things (threshold setting ≤ 0.5%). When the sensor detects that the impurity exceeds the standard, the system automatically triggers a secondary purification process.

[0098] Dehydration of olefin siloxane: TEMS (triethylenetrimethylcyclotrisiloxane) enters the dehydration reaction kettle, and through the linkage of a humidity sensor and a nitrogen purging device, the water content is controlled ≤ 50 ppm to avoid side reactions.

[0099] Intelligent regulation of the condensation reaction: Material ratio and feeding: Through the IoT weighing module of the atomization feeding system, automatic feeding is carried out according to the molar ratio of MPS to TEMS of 1:1.2, with an error range of ±0.05%.

[0100] Dynamic optimization of reaction conditions: In a jacketed reactor under nitrogen protection, the temperature sensor and the edge computing module work together to adjust the heating power in real time, so that the reaction temperature is stabilized at 95±2°C (process requirement: 80~120°C); at the same time, through the linkage of the pressure sensor and the stirring motor, the stirring rate is maintained at 800±50 rpm to ensure the uniformity of the reaction.

[0101] Precise control of catalyst: Benzoyl peroxide catalyst is added at a ratio of 1.0 wt% through a metering injection pump. When the online detected reaction conversion rate is lower than 95%, the system automatically supplements the catalyst (maximum additional amount ≤0.5%) to avoid over-crosslinking.

[0102] Continuous purification: Automation of vacuum distillation: The operation of the distillation column is controlled by a temperature-pressure coupling model. The IoT system dynamically adjusts the vacuum degree (-0.08~-0.1 MPa) according to the real-time data of gas chromatography (GC) to separate unreacted monomers and by-products.

[0103] Activated carbon adsorption and filtration: The differential pressure sensor of the microfiltration membrane filter is linked with the activated carbon adsorption tower. When the pressure before the membrane exceeds 0.3 MPa, the automatic backwashing program is triggered to ensure the product chromaticity (APHA value ≤50) and the solid residue content ≤0.1%.

[0104] Closed-loop quality control warning: Online detection of functional groups: The absorption peak intensity changes of characteristic functional groups in the reaction system are monitored in real time through online infrared spectroscopy (IR). Combining the dynamic integral ratio of the peak areas of reactants and products, the real-time conversion rate is calculated; when the system determines that the conversion rate reaches the preset threshold (≥95%), the termination reaction instruction is automatically triggered to ensure the product synthesis efficiency and quality stability.

[0105] Early warning of abnormal purity: The GC-MS data is uploaded to the cloud analysis platform. If the detected purity of the main component is <98%, the system automatically isolates the current batch and traces the cause of the abnormality, and synchronously adjusts the subsequent production parameters.

[0106] Integration of green continuous production equipment: A continuous production line composed of a spiral stirring reactor and an annular grinding disc grinding unit is adopted. The following functions are realized through the IoT central controller: Solvent-free process control: The reaction system operates in a fully enclosed manner, and the VOC emission sensor monitors it in real time to <10 ppm. <0,

[0107] Device status self-check: The vibration sensor and temperature module monitor the operating status of the grinding unit. When the bearing temperature exceeds 80 °C or the vibration amplitude > 50 μm, the device will automatically stop and send a maintenance alarm.

[0108] Effect of the embodiment: Through the Internet of Things technology, this embodiment connects the entire processes of raw material pretreatment, reaction regulation, and purification detection. It shortens the time-consuming of a single batch in traditional batch production from 12 hours to 6 hours, improves the batch stability of the product (RSD ≤ 1.5%) by 40%, and reduces the VOC emissions by 90%, achieving efficient and environmentally friendly continuous production of bifunctional silane coupling agents.

[0109] This invention uses the Internet of Things technology to collect the activity data of silica modifiers in the synthesis and production processes in real time. Combining with a dynamic feedback mechanism, it solves the problem that it is difficult to accurately control the synthesis conditions of bifunctional silane coupling agents (Si747) by traditional methods. By using a sensor network to monitor key parameters such as reaction temperature, pressure, and material ratio in real time, the system can dynamically optimize the synthesis process to ensure the stability and functionality of the Si747 molecular structure. This precise control significantly improves the dispersion uniformity of silica in rubber and enhances the interfacial bonding force between it and the rubber matrix, thus overcoming the defect of insufficient coupling agent efficiency caused by parameter fluctuations in traditional processes.

[0110] Aiming at the problem of low efficiency in processing continuous production data, this invention uses the Internet of Things platform to integrate and analyze the data of the entire production process in real time. Through edge computing and cloud collaboration, it quickly identifies abnormal working conditions and triggers an early warning mechanism. The system automatically adjusts the production rhythm and process parameters by comparing historical data with real-time data, reduces the delay of human intervention, and ensures the efficient and stable operation of the synthesis process. At the same time, by optimizing resource utilization and reducing ineffective energy consumption, this method achieves the goal of green production, reduces waste emissions, and improves the overall environmental protection benefits.

[0111] Compared with the limitations of existing technologies that rely on offline detection and empirical adjustment, this invention constructs a closed-loop control system from synthesis to production through the full-link data connection and intelligent decision-making of the Internet of Things. This technology not only solves the problems of data islands and response lags in continuous production by traditional methods, but also ensures the uniformity of modifier performance through real-time activity monitoring and early warning, providing reliable technical support for the efficient application of silica in rubber, and finally achieving the goals of green, efficient, and controllable industrial production.

Claims

1. An active monitoring and warning method for silica modifiers based on the Internet of Things, characterized in that, Including: Construct a multi-dimensional process parameter database, and collect the reactor pressure, catalyst dispersion, and material residence time in the Internet of Things sensing network in real time through an edge computing gateway, and synchronize the process parameters to the central database; Based on the historical production data in the multi-dimensional process parameter database, define a genetic algorithm fitness function, where the reaction temperature, stirring rate, and raw material molar ratio are encoded as chromosome gene sequences, and the bifunctional group grafting rate is used as the main optimization target, and a multi-objective optimization model is constructed in combination with the constraint conditions of production energy consumption and equipment load rate; According to the raw material purity fluctuation and equipment aging coefficient collected in real time, dynamically adjust the search boundary of the genetic algorithm, and verify the feasibility of the adjusted parameter combination through a digital twin; Execute genetic algorithm iterative optimization in the digital twin, generate an optimized parameter set through operations such as initializing the population, crossover and mutation, fitness evaluation, and survival of the fittest, and push the optimized parameter set to the MES system to trigger parameter calibration of production equipment; Verify the accuracy of the optimized parameter set in reverse by comparing the deviation between the predicted bifunctional group grafting rate in the digital twin and the real-time monitoring data, and trigger retraining of the genetic algorithm when the deviation exceeds a preset threshold; Cooperatively control the optimized parameter set with a hierarchical early warning system, and call the parameter adjustment strategy corresponding to the early warning level according to the early warning level. The parameter adjustment strategy includes local parameter fine-tuning schemes, resetting the initial population of the genetic algorithm, and loading the historical optimal parameter combination.

2. The active monitoring and early warning method of the silica modifier based on the Internet of Things according to claim 1, wherein, The construction of the multi-dimensional process parameter database includes: Establish a parameter and efficiency mapping table in the central database, and the parameter and efficiency mapping table associates the corresponding relationship between the temperature, catalyst dosage, and bifunctional group grafting rate in the historical production batches; Denoise and standardize the collected original sensing data through the edge computing gateway to generate a structured data set that meets the training requirements of the genetic algorithm.

3. The active monitoring and warning method of the silica modifier based on the Internet of Things according to claim 1, characterized in that The definition of the genetic algorithm fitness function includes: Based on the data blood relationship mapping model in the parameter and efficiency mapping table, identify the coupling weight relationship between the reaction temperature and catalyst dispersion; Use the bifunctional group grafting rate as the main optimization target, and use the dynamic thresholds of unit production energy consumption and equipment load rate as constraint conditions to construct a multi-dimensional fitness evaluation model.

4. The method for monitoring and warning the activity of the silica modifier based on the Internet of Things according to claim 1, characterized in that According to the raw material purity fluctuation and equipment aging coefficient collected in real time, dynamically adjust the search boundary of the genetic algorithm, and verify the feasibility of the adjusted parameter combination through a digital twin, including: In response to the decrease in the intensity of the mercapto characteristic peak monitored by infrared spectroscopy, limit the temperature adjustment range to a safe interval set based on the material thermal stability experiment; When the real-time carbon emissions reach 95% of the preset threshold stored in the central database, dynamically lower the preset maximum value of the catalyst dosage; Simulate the operation of the parameter combination through the digital twin, and filter out invalid parameter combinations that cause equipment overload or functional group oxidation.

5. The active monitoring and warning method of the silica modifier based on the Internet of Things according to claim 1, characterized in that The execution of the genetic algorithm iterative optimization includes: Extract parental chromosomes from the historical optimal parameter combination to generate an initial population; Within the dynamically adjusted search boundary, gene segment crossover is performed on the chromosomes according to a crossover probability of 0.7 - 0.9, and random mutation is performed according to a mutation probability of 0.01 - 0.05; Inject the mutated parameter combination into the digital twin for simulation operation, and calculate the grafting rate through the activity evaluation model to generate a fitness value; Retain the individuals with the top preset proportion of fitness rankings and iterate to the next generation until the optimization parameter set converges.

6. The method for monitoring and warning the activity of the silica modifier based on the Internet of Things according to claim 1, wherein, The reverse verification of the accuracy of the optimization parameter set by comparing the deviation between the predicted bifunctional group grafting rate in the digital twin and the real-time monitoring data, and triggering the retraining of the genetic algorithm when the deviation exceeds the preset threshold includes: Calculate the deviation rate between the digital twin predicted grafting rate and the actual infrared monitoring data in real time; When the deviation rate exceeds ±5%, extract the current production environment characteristic data and retrain the genetic algorithm model; Feed back the device control parameter execution error to the genetic algorithm to correct the device response coefficient model.

7. The active monitoring and warning method of the silica modifier based on the Internet of Things according to claim 1, characterized in that The collaborative control of the optimization parameter set and the hierarchical early warning system, and calling the parameter adjustment strategy corresponding to the early warning level according to the early warning level. The parameter adjustment strategy includes local parameter fine-tuning scheme, resetting the initial population of the genetic algorithm, and loading the historical optimal parameter combination includes: When the deviation rate ≤ 5%, trigger a first-level early warning, and call the genetic algorithm to generate a temperature compensation or stirring rate fine-tuning scheme; When the deviation rate is between 5% and 10%, trigger a second-level early warning, combine abnormal conduction analysis to locate the fault source, and reset the initial population search range of the genetic algorithm; When the deviation rate > 10%, trigger a third-level early warning, and load the historical optimal parameter combination from the central database to start the resumption of production process.

8. The active monitoring and warning method of the silica modifier based on the Internet of Things according to claim 3, characterized in that, The construction of the multi-dimensional fitness evaluation model also includes: According to the device aging coefficient collected in real time, based on the data blood relationship mapping model in the parameter and efficiency mapping table, dynamically adjust the weight coefficient of the device load rate constraint condition; Modify the prediction calculation formula of the bifunctional group grafting rate based on the real-time raw material purity data.

9. The method for monitoring and warning the activity of the silica modifier based on the Internet of Things according to claim 4, wherein, The filtering of invalid parameter combinations that cause equipment overload or functional group oxidation includes: preloading the maximum bearing value of the device and the critical value of material thermal decomposition in the digital twin; When the simulation operation parameter combination triggers the preloaded maximum bearing value of the device or the critical value of material thermal decomposition in the digital twin, mark this parameter combination as an invalid solution and terminate the fitness evaluation.

10. The active monitoring and early warning method of the silica modifier based on the Internet of Things according to claim 6, characterized in that Also includes: Extract the raw material purity, environmental temperature and humidity, and equipment operation log data during the period when the deviation exceeds the threshold; Use the extracted data as new training samples to update the parameter and efficiency mapping table; Adopt an online weight update algorithm based on the new data to dynamically adjust the priority weights of the reaction temperature and catalyst dosage in the chromosome encoding rule.

Images