Preparation method of starch-based concrete additives by multi-component synergistic grafting and digital control

Abstract

This invention relates to the field of green high-performance concrete admixtures, and discloses a method for preparing starch-based concrete admixtures using multi-component synergistic grafting and digital control, comprising the following steps: S1, pre-establishing a soft sensor model for online analysis of monomer concentrations and a kinetic model capable of predicting dynamic changes in the reaction process; S2, during the reaction process, online acquisition of process information such as spectra, and real-time analysis of the current concentrations of various monomers using the soft sensor model; S3, using the real-time concentrations as initial conditions, online solving of a control strategy aimed at optimizing the overall performance of the final product using the kinetic model; S4, based on the optimized control strategy, dynamic closed-loop control of the reaction process is implemented by adjusting operating variables such as dripping rate and temperature. This invention, through soft sensor and model predictive control technology, achieves online optimization of the reaction process for end-application performance, ensuring precise control of product quality and batch stability.

Description

Technical Field

[0001] This invention relates to the field of green high-performance concrete admixtures technology, specifically a method for preparing starch-based concrete additives using multi-component synergistic grafting and digital control. Background Technology

[0002] This invention relates to the interdisciplinary field of polymer chemistry and civil engineering, specifically to a method for preparing starch-based concrete admixtures through multi-component synergistic grafting and digital control. Starch, as a widely available, inexpensive, and biodegradable natural polymer, has become an important direction for developing green, high-performance concrete admixtures through chemical modification to impart specific functions. By selectively grafting multiple functional monomers onto the starch macromolecular backbone, admixtures with comprehensive properties such as excellent water dispersibility, long-lasting slump retention, and strong anti-mud properties can be prepared. However, how to precisely control this complex liquid-phase free radical copolymerization process to stably obtain the molecular structure and final properties that meet specific engineering requirements is a key technical challenge that urgently needs to be solved in this field.

[0003] In existing technologies, the preparation of this type of starch-based adjuvant is typically carried out in a batch or semi-batch polymerization reactor equipped with a stirrer and a jacketed temperature control system. A typical process flow is as follows: first, natural starch is gelatinized in an aqueous medium to form a homogeneous starch matrix solution; then, at a set reaction temperature, the initiator system (preferably a redox initiator system, such as ammonium persulfate-sodium bisulfite) is started, and aqueous solutions of one or more functional monomers are added dropwise according to a predetermined procedure. Throughout the reaction, operators typically control the dropping rate of each material and the temperature profile of the reactor according to a pre-defined process formulation. To monitor the reaction progress, offline analysis is commonly used, i.e., samples are periodically extracted from the reactor and sent to a laboratory for analysis of key indicators such as the residual concentration of monomers or the grafting rate of the product using methods such as high-performance liquid chromatography (HPLC), chemical titration, or Fourier transform infrared spectroscopy (FTIR).

[0004] While existing technologies can produce starch-based additives using the methods described above, they still have some shortcomings in terms of process control precision and product performance stability. Current control strategies primarily rely on offline detection, leading to significant control lag. The root cause is that the entire process, from reactor sampling, sample pretreatment, instrument analysis to data interpretation, is time-consuming. By the time the analysis results are fed back to the control system, the actual operating conditions within the reactor have already deviated from the state at the time of sampling. This renders any control measures based on these results ineffective in real-time, failing to truly guide the dynamic reaction process precisely, and making it difficult to cope with sudden disturbances such as batch-to-batch differences in raw materials or fluctuations in ambient temperature.

[0005] Furthermore, traditional processes establish a weak link between process control and final product performance. Control systems often target easily measurable process variables such as monomer conversion rate or the concentration of a certain intermediate product. However, these variables have complex nonlinear relationships with end-product performance indicators such as the water dispersibility and slump retention of additives in concrete applications. Due to the lack of a mathematical model that directly links the two, optimization of the production process often relies on extensive experimental trial and error, making it difficult to achieve "customized" production for specific engineering needs. This "open-loop" control logic and over-reliance on operator experience ultimately makes it difficult to guarantee the performance consistency between product batches. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a method for preparing starch-based concrete additives using multi-element synergistic grafting and digital control. This method solves the problems in existing technologies, which rely on offline detection and preset open-loop control strategies, resulting in difficulties in precisely controlling the starch-based graft copolymerization process, unstable batch performance of products, and the inability to directly correlate with end-application performance.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for preparing starch-based concrete additives using multi-component synergistic grafting and digital control, comprising the following steps:

[0008] A soft measurement model capable of analyzing the concentrations of various monomers based on online process information of the reaction system, and a reaction process kinetic model capable of predicting the dynamic changes in the graft copolymerization process are established in advance.

[0009] During the graft copolymerization reaction, the process information of the reaction system is obtained online using a reaction process kinetic model, and the current concentration of various monomers in the reaction system is obtained in real time using the aforementioned soft measurement model.

[0010] Based on the current concentrations of various monomers obtained from this real-time analysis as initial conditions, and using the aforementioned reaction process kinetic model, an online optimization solution is obtained to find a control strategy that can optimize the preset comprehensive performance index of the final product.

[0011] The graft copolymerization process is then regulated based on the control strategy derived from the optimization solution.

[0012] The steps of pre-establishing the soft measurement model include:

[0013] First, a series of grafting reaction experiments covering different process conditions were conducted, and online spectral data of the reaction system and samples for offline analysis were collected simultaneously during the experiments.

[0014] Secondly, the absolute concentrations of various monomers in the sample were determined using offline analysis methods;

[0015] Finally, partial least squares regression analysis was performed on the spectral data and the corresponding monomer concentrations to establish the mapping relationship between spectral information and actual concentrations. The mathematical model is as follows:

[0016] ;

[0017] In the formula, This is the true monomer concentration matrix obtained through offline analysis; This is a matrix of online spectral data acquired synchronously. This is the regression coefficient matrix of the PLS model obtained through regression training; Let be the residual matrix of the model.

[0018] Preferably, firstly, the present invention requires the prior establishment of a "soft measurement model" capable of characterizing the intrinsic correlation of the reaction system. The process of establishing this model is as follows: in multiple reference experiments, "online process information" (such as infrared spectroscopy, Raman spectroscopy, etc.) is continuously acquired through sensors such as online spectral probes inserted into the reaction vessel; at the same time, physical samples are taken at different time points, and standard chemical analysis methods such as high-performance liquid chromatography (HPLC) are used to accurately determine the "concentration of multiple monomers" in the sample.

[0019] Finally, using chemometric algorithms such as partial least squares, a large amount of online process information was correlated and regressed with the corresponding true values ​​of monomer concentrations to construct and calibrate the soft measurement model.

[0020] In the actual graft copolymerization process, the system continuously acquires process information of the reaction system online and transmits it as input to the aforementioned soft sensing model in real time. The model performs high-speed calculations based on the calibrated mathematical relationships, thereby obtaining the current concentration of various monomers in the reaction system in real time.

[0021] By using soft measurement technology, the time required to obtain monomer concentration data is shortened, enabling the control system to "see clearly" the actual chemical composition changes inside the reactor in real time and accurately at a time scale that matches the dynamic changes of the reaction.

[0022] The steps of pre-establishing the kinetic model of the reaction process include:

[0023] By experimental determination and parameter separation method, a kinetic equation describing the rate of change of grafting degree is constructed, and its core equation is:

[0024] ;

[0025] In the formula: To determine the grafting degree, samples were taken during the reaction using the ninhydrin method. Pre-exponential factors; The initiator concentration was determined using an ammonium persulfate-sodium bisulfite redox system. The order of the initiator reaction; Total monomer concentration; The order of the monomer reaction; It is the apparent activation energy; This refers to starch concentration. Starch concentration index;

[0026] The parameters in the equation are determined by the following steps:

[0027] Temperature gradient experiment: fixed The reaction remains unchanged and proceeds within the temperature range of 50–80°C, via obtained by linear regression and ;

[0028] Initiator concentration experiment: fixed temperature, Unchanged, Changed ,pass Determining the slope =0.5;

[0029] Monomer concentration experiment: fixed temperature, Unchanged, Changed ,pass Determining the slope =1.0;

[0030] Starch concentration experiment: fixed temperature, Unchanged, Changed ,pass Determining the slope =−0.2;

[0031] Substitute the obtained parameters into the equation to predict the grafting degree change curve of the experimental group that did not participate in parameter fitting, verify that the prediction error is less than 5%, and couple the verified kinetic equation with the material balance equation and the energy balance equation to form a complete reaction process kinetic model.

[0032] Preferably, before initiating optimization, this invention also requires the pre-establishment of a reaction process kinetic model capable of predicting the dynamic changes in the graft copolymerization reaction. During the reaction, the system inputs the "current concentrations of multiple monomers" obtained in real-time from S2 as initial conditions into the kinetic model. These current concentrations include, but are not limited to: the current concentration of the first functional monomer (e.g., itaconic anhydride with a carboxyl group), the current concentration of the second functional monomer (e.g., glycidyl acrylate with an epoxy group), and the current concentration of the third functional monomer (e.g., zwitterionic monomer with a sulfonic acid group). Subsequently, the Model Predictive Control (MPC) algorithm initiates the "online optimization solution" process. Specifically, with the optimization objective of "optimizing the preset comprehensive performance indicators of the final product," and under the premise of satisfying various process constraints (e.g., safe temperature and pressure ranges), through repeated deduction of the kinetic model, it calculates how to operate (e.g., adjusting the dripping rate of each material and the reaction temperature) in the future to achieve the best comprehensive performance of the final product (e.g., water dispersibility, slump retention), thereby overcoming uncertainties such as raw material fluctuations and environmental changes, and improving the performance consistency and stability between product batches.

[0033] The steps of acquiring process information online and analyzing it in real time to obtain the current concentration include:

[0034] The mid-infrared full spectrum data of the reaction system is acquired in real time using an online Fourier transform infrared probe, which serves as the online process information.

[0035] The real-time acquired spectral data is substituted into the pre-established soft measurement model to instantly calculate the current concentration of various monomers in the reaction system.

[0036] In the steps of the online optimization solution control strategy, the preset final product comprehensive performance index is an evaluation function. This function is established based on the performance of the additive in concrete applications, and its specific form is as follows:

[0037] ;

[0038] In the formula, The overall performance index to be maximized; For water dispersibility performance indicators; The hydrolysis kinetic constant of the slow-release unit; The characteristic evaluation time for ensuring collapse performance; For mud resistance performance indicators; , , These are the weighting factors for each performance metric.

[0039] The steps for online optimization of the control strategy are implemented through a model predictive control algorithm, which cyclically executes the complete optimization process within the rolling time domain.

[0040] First, based on the current concentrations of the various monomers as initial conditions, the reaction process kinetic model is used to predict the evolution trajectory of the reaction system state in the future prediction time domain under different alternative control strategies.

[0041] Then, an open-loop optimization problem with the goal of optimizing the overall performance index of the final product is solved to obtain the optimal control sequence covering multiple future steps;

[0042] Finally, only the first control action in the optimal control sequence is executed, and the optimization process is repeated at the next control time.

[0043] The steps for solving the open-loop optimization problem include:

[0044] Given the constraints of the process model, initial state, control input, and state variables, find the optimal control sequence that maximizes the following objective function:

[0045] ;

[0046] In the formula, This is the current control moment; To predict the length of the time domain; For at any time Predicted values ​​for the end state in the prediction time domain; For at any time Control inputs for planning.

[0047] The steps of regulating the reaction process according to the control strategy include:

[0048] Based on the optimized control strategy, at least one or more process variables selected from the following are adjusted in real time:

[0049] The dripping flow rate of each monomer solution, the dripping flow rate of the initiator solution, and the heating or cooling power of the reactor jacket.

[0050] The method further includes a reaction endpoint determination step, which is based on the results of the online optimization solution, specifically:

[0051] When the predicted future gain of the final product's overall performance index is lower than a preset threshold, the reaction is determined to have reached its optimal endpoint and the reaction is terminated.

[0052] The preset threshold is defined as follows:

[0053] Within three consecutive control cycles, the predicted Increment All are less than or equal to 0.5% of the current Value, where, Predict the increment of the comprehensive performance index between adjacent control times; For at any time Predicting the next moment The overall performance index; For a moment The current overall performance index.

[0054] This invention provides a method for preparing starch-based concrete additives using multi-component synergistic grafting and digital control. It offers the following advantages:

[0055] 1. This invention overcomes the limitations of traditional offline sampling and analysis by combining online spectral sensing technology with a soft-sensor model. It enables real-time analysis of various monomer concentration changes, providing dynamic data support with high spatiotemporal resolution for the reaction process. Combined with the multi-step predictive capabilities of the reaction kinetic model, the control strategy can proactively compensate for interference factors such as material consumption and temperature fluctuations, improving the control accuracy of grafting degree and batch consistency.

[0056] 2. This invention reverse-maps the terminal performance indicators of concrete additives (water dispersibility, slump retention, and mud resistance) to the control targets of the reaction process, achieving a closed-loop linkage between process parameters and product performance through a comprehensive performance index function. This design allows the production process to dynamically adjust and optimize weights according to different engineering needs, flexibly adapting to diverse concrete mix design scenarios.

[0057] 3. This invention is based on the Model Predictive Control (MPC) framework, utilizing the predictive ability of a dynamic model to forecast reaction trends. It adjusts operating parameters in advance before external disturbances (such as raw material fluctuations or environmental temperature changes) significantly affect the grafting degree. This feedforward-feedback composite control mechanism effectively suppresses overshoot and oscillation phenomena commonly found in traditional PID control. Attached Figure Description

[0058] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0059] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0060] Please see the appendix Figure 1This invention provides a method for preparing starch-based concrete additives using multi-component synergistic grafting and digital control, comprising the following steps:

[0061] S1. Establish a soft measurement model in advance that can analyze the concentration of various monomers based on the online process information of the reaction system, and a reaction process kinetic model that can predict the dynamic changes of the graft copolymerization process.

[0062] S1 constructs a mapping relationship between spectral information and monomer concentration through experimental design and multivariate correction algorithms.

[0063] First, in a laboratory-scale reaction apparatus, graft copolymerization experiments covering different temperatures, material ratios, and reaction times were designed to ensure that the process conditions covered the operational range that might occur in actual production. During the reaction, mid-infrared full-spectrum data of the reaction system were acquired in real time using an online Fourier transform infrared (FTIR) spectrometer to form a spectral data matrix. Each row corresponds to an experimental sample, and each column corresponds to the absorbance value at a specific wavenumber.

[0064] Simultaneously, at the same time point as the spectral acquisition, the reaction solution was sampled, and the concentrations of the three functional monomers (itaconic anhydride, glycidyl acrylate, and sulfobetaine) in the sample were accurately determined using offline analysis methods such as high-performance liquid chromatography (HPLC), forming a concentration data matrix. Partial least squares regression (PLS) algorithm is used to model the spectral and concentration data. Its mathematical expression is as follows:

[0065] ;

[0066] In the formula, This is the true monomer concentration matrix obtained through offline analysis; This is a matrix of online spectral data acquired synchronously. This is the regression coefficient matrix of the PLS model obtained through regression training; Let be the residual matrix of the model.

[0067] Based on the principles of chemical reaction engineering and experimental data fitting, a state-space model describing the dynamic behavior of graft copolymerization is established. The core of the kinetic model is the grafting degree change rate equation, which takes the following form:

[0068] ;

[0069] In the formula: To determine the grafting degree, samples were taken during the reaction using the ninhydrin method. Pre-exponential factors; The initiator concentration was determined using an ammonium persulfate-sodium bisulfite redox system. The order of the initiator reaction; Total monomer concentration; The order of the monomer reaction; It is the apparent activation energy; This refers to starch concentration. This is the starch concentration index.

[0070] A temperature gradient experiment was conducted in a constant-temperature reactor, with the reaction proceeding within the range of 50–80°C while keeping other conditions constant. obtained by linear regression and ;

[0071] By changing the monomer concentration in the experiment, using Determining the slope =0.5;

[0072] By changing the monomer concentration in the experiment, using Determining the slope =1.0;

[0073] By changing the starch concentration, using Determining the slope =−0.2;

[0074] Substituting the calibrated parameters into the kinetic equation, the grafting degree change curve of the independent experimental group was predicted, and the prediction error was verified to be less than 5%. The verified grafting degree kinetic equation was then coupled with the following auxiliary equation to form a complete kinetic model of the reaction process:

[0075] Material balance equations: describe the dynamic consumption and replenishment of each monomer and initiator in the reaction system;

[0076] Energy balance equation: reflects the effect of reaction heat and external heating / cooling power on system temperature;

[0077] State-space expression: Integrating the above equations, the system state vector is used. and control input vector The formal description of the system dynamics.

[0078] Material balance equations and energy balance equations are existing technologies and will not be elaborated further in this paper.

[0079] S2. During the graft copolymerization reaction, the process information of the reaction system is obtained online using the reaction process kinetic model, and the current concentration of various monomers in the reaction system is obtained in real time using the soft measurement model established above.

[0080] In this embodiment, the technical solution of acquiring process information online and analyzing monomer concentration in real time is achieved through the synergistic effect of online spectral sensing technology and soft measurement model. The specific implementation process is as follows:

[0081] During the graft copolymerization reaction, an online Fourier transform infrared (FTIR) spectral probe integrated into the reactor continuously acquires mid-infrared full-spectrum data of the reaction system at a preset sampling frequency. The spectral data covers the absorption bands of characteristic functional groups (e.g., characteristic absorption peaks of carbonyl, epoxy, and sulfonic acid groups), forming a real-time spectral vector. The spectral acquisition system transmits data to the control unit via optical fiber, ensuring real-time performance and interference resistance.

[0082] The pre-trained partial least squares (PLS) soft measurement model is embedded into the real-time control system. For each sampling time... The system performs the following operations:

[0083] Data preprocessing: processing the real-time acquired spectral vectors Standardization is performed to eliminate baseline drift and noise interference, specifically including Savitzky-Golay smoothing filtering and multivariate scattering correction (MSC).

[0084] Concentration analysis: The preprocessed spectral data is input into the PLS model, and the estimated concentration of each monomer is calculated in real time through matrix operations.

[0085] ;

[0086] In the formula, The regression coefficient matrix obtained during the model training phase has dimensions that match the number of spectral variables and the types of monomers to be tested. This refers to the spectral vector acquired in real time.

[0087] Output: Concentration vector obtained from analysis It includes the real-time concentration values ​​of each monomer, which serve as input parameters for subsequent optimization control.

[0088] To ensure the reliability of soft measurement results, the system is equipped with the following verification mechanism:

[0089] Spectral quality assessment: Real-time calculation of the signal-to-noise ratio (SNR) and absorbance dynamic range of the acquired spectrum; when the SNR is lower than the threshold, an alarm is triggered and resampling is initiated.

[0090] Concentration rationality verification: The current concentration estimate is compared with the historical data of the previous period. If a sudden change occurs (such as a single change rate exceeding 10%), redundant calculation or manual intervention process is initiated.

[0091] Real-time concentration data obtained through analysis These will be used as initial conditions input into the reaction process kinetic model. Specifically, at the beginning of each optimization cycle of model predictive control (MPC), the system substitutes the concentrations of each monomer into the state vector. And update the following key parameters:

[0092] Initiator consumption: Calculate the current effective initiator concentration based on the initiator concentration decay kinetic equation;

[0093] Response progress indicator: The cumulative grafting degree is updated in the form of an integral to provide a basis for determining the endpoint.

[0094] All analytical results and raw spectral data are stored in a distributed time-series database, and the concentration change curves, spectral quality indicators and system alarm status are dynamically displayed through a human-machine interface (HMI) to provide process monitoring support for operators.

[0095] S3. Based on the current concentrations of various monomers obtained from the real-time analysis as initial conditions, and using the aforementioned reaction process kinetic model, an online optimization solution is obtained to find a control strategy that optimizes the preset comprehensive performance index of the final product.

[0096] In this embodiment, the technical solution for online optimization of the control strategy is achieved through the collaborative optimization of the model predictive control (MPC) algorithm and the reaction process kinetic model. The specific implementation process is as follows:

[0097] Based on comprehensive performance indicators An optimization objective function oriented towards the application performance of concrete additives was constructed. This function integrates quantitative indicators of water dispersibility, slump retention, and mud resistance, and its mathematical form is:

[0098] ;

[0099] In the formula, The overall performance index to be maximized; For water dispersibility performance indicators; The hydrolysis kinetic constant of the slow-release unit; The characteristic evaluation time for ensuring collapse performance; For mud resistance performance indicators; , , These are the weighting factors for each performance metric.

[0100] A rolling time-domain optimization strategy is adopted in each control cycle. Perform the following operations:

[0101] State initialization: The real-time monomer concentration obtained from step S2 is initialized. as the initial state vector And load the reaction process kinetic model;

[0102] Predicting in the future time domain Within, different control input sequences Evolution trajectory of the lower system state In the formula, To be in discrete time steps The control input vector; To predict the length of the time domain; In order to control the moment For future moments State prediction value; This is the current control moment; For at any time Control inputs for planning;

[0103] Open-loop optimization: Solve the following constrained optimization problem:

[0104] ;

[0105] In the formula, This is the current control moment; To predict the length of the time domain; For at any time Predicted values ​​for the end state in the prediction time domain; For at any time Control inputs for planning;

[0106] Control execution: Select the first control action from the optimized control sequence. This is then converted into actual execution commands (such as adjusting the metering pump speed or adjusting the jacket temperature setpoint).

[0107] The optimization algorithm is also implemented through the following techniques:

[0108] Discretization: The continuous-time dynamic model is converted into a discrete-time state-space model, and numerical integration is performed using the fourth-order Runge-Kutta method. The time step is synchronized with the sampling period of the control system.

[0109] Gradient calculation: The gradient of the objective function with respect to the control input is calculated using the adjoint equation method or the finite difference method, which accelerates the optimization convergence;

[0110] Real-time solver: Uses interior-point method or sequential quadratic programming (SQP) algorithm to solve constrained nonlinear optimization problems, and embeds pre-compiled optimization libraries (such as IPOPT) to ensure computational efficiency.

[0111] The aforementioned fourth-order Runge-Kutta method, adjoint equation method or finite difference method, interior point method or sequential quadratic programming algorithm and pre-compiled optimization library are all existing technologies, and will not be elaborated further in this paper.

[0112] To accommodate different concrete engineering needs (such as prioritizing high water reduction rates or long slump retention times), a dynamic configuration interface for weighting factors is set up. Operators can adjust these factors in real time via a human-machine interface (HMI). , , The system will automatically update the objective function and trigger re-optimization based on the given value.

[0113] S4. The process of graft copolymerization is regulated according to the control strategy obtained from the optimization solution.

[0114] In this embodiment, the technical solution for dynamically controlling the reaction process according to the optimization strategy is achieved through the synergistic effect of a closed-loop control architecture and multiple actuators. The specific implementation process is as follows:

[0115] The optimal control sequence output by the Model Predictive Control (MPC) algorithm. Converted into executable operation instructions. Control vector. Includes the following key operational variables:

[0116] Flow rate setpoints for each monomer solution metering pump ;

[0117] Flow rate setpoint of initiator solution metering pump ;

[0118] The set value of the reactor jacket temperature controller .

[0119] Setpoint commands are sent to the corresponding actuators via industrial fieldbuses (such as PROFIBUS or Modbus) to ensure the real-time and reliable transmission of commands.

[0120] Flow control subsystem: Each metering pump is equipped with a high-precision stepper motor and flow sensor, and uses a proportional-integral-derivative (PID) control algorithm to adjust the speed in real time, so that the actual flow rate is controlled. Quickly track set value The flow feedback signal is filtered by Kalman filtering to eliminate pulsation noise and ensure control stability.

[0121] Temperature control subsystem: Jacket temperature controller according to set value With respect to the actual temperature of the reaction system To compensate for deviations, dynamically adjust the heater power and cooling water valve opening.

[0122] Preferably, a feedforward-feedback composite control strategy is adopted, which uses a reaction heat model to predict changes in heat load and compensate for temperature fluctuations in advance.

[0123] After the control command is executed, the control effect is verified through the following mechanism:

[0124] Flow deviation monitoring: Real-time calculation of the relative deviation between the actual flow rate of each material and the set value. When the deviation continues to exceed 5%, an alarm is triggered and the flow calibration program is started.

[0125] Temperature gradient monitoring: The uniformity of temperature distribution inside the reaction system is detected by a distributed temperature sensor array. If the maximum temperature difference exceeds 2℃, the stirring rate or jacket zone temperature control strategy is automatically adjusted.

[0126] After each control action is executed, the spectral data of the reaction system are reacquired. The monomer concentration estimate is updated using a soft sensor model. The updated concentration data is fed back to the MPC optimizer, forming a closed-loop control loop of "measurement-estimation-optimization-execution" to achieve adaptive adjustment under dynamic operating conditions.

[0127] The system will automatically switch to safe operating mode when it detects the following abnormal conditions:

[0128] Material shortage warning: When the liquid level sensor in the storage tank detects that the storage level of any raw material is lower than the safety threshold, it triggers a pause in feeding and notifies the tank to replenish the material.

[0129] Runaway reaction warning: When the rate of temperature rise exceeds the safety limit or the pressure sensor fluctuates abnormally, the emergency cooling and pressure relief procedure will be activated.

[0130] Model mismatch detection: When the prediction residual of the soft measurement model continues to increase, the model is automatically updated online or switched to the backup control strategy.

[0131] All control commands, actuator status, process variables, and alarm events are stored in the industrial database in the form of timestamps, supporting multi-dimensional retrieval and analysis by batch, time range, or anomaly type, providing a data foundation for process optimization and fault diagnosis.

[0132] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for preparing starch-based concrete additives using multi-element synergistic grafting and digital control, characterized in that, Includes the following steps: S1. Establish a soft measurement model in advance that can analyze the concentration of various monomers based on the online process information of the reaction system, and a reaction process kinetic model that can predict the dynamic changes of the graft copolymerization process. The steps of pre-establishing the kinetic model of the reaction process include: Through experimental determination and parameter separation, a grafting degree change rate equation describing the grafting degree change rate is constructed, in the form: ； In the formula: To determine the grafting degree, samples were taken during the reaction using the ninhydrin method. Pre-exponential factors; The initiator concentration was determined using an ammonium persulfate-sodium bisulfite redox system. The order of the initiator reaction; Total monomer concentration; The order of the monomer reaction; It is the apparent activation energy; This refers to starch concentration; This refers to the starch concentration index; The parameters in the grafting degree change rate equation are determined by the following steps: Temperature gradient experiment: fixed The reaction remains unchanged and proceeds within the temperature range of 50–80°C, via obtained by linear regression and ; Initiator concentration experiment: fixed temperature, Unchanged, Changed ,pass Determining the slope =0.5; Monomer concentration experiment: fixed temperature, Unchanged, Changed ,pass Determining the slope =1.0; Starch concentration experiment: fixed temperature, Unchanged, Changed ,pass Determining the slope =−0.2; Substitute the obtained parameters into the grafting degree change rate equation to predict the grafting degree change curve of the experimental group that did not participate in parameter fitting, verify that the prediction error is less than 5%, and couple the verified grafting degree change rate equation with the material balance equation and the energy balance equation to form a complete reaction process kinetic model. S2. During the graft copolymerization reaction, the process information of the reaction system is obtained online using the reaction process kinetic model, and the current concentration of various monomers in the reaction system is obtained in real time using the soft measurement model established above. The steps of acquiring process information online and analyzing it in real time to obtain the current concentration include: The mid-infrared full spectrum data of the reaction system is acquired in real time using an online Fourier transform infrared probe, which serves as the online process information. The real-time acquired spectral data is substituted into the pre-established soft measurement model to instantly calculate the current concentration of various monomers in the reaction system; In the steps of the online optimization solution to the control strategy, the preset comprehensive performance index of the final product is an evaluation function. This function is based on the performance of the additive in concrete applications, and its specific form is as follows: ； In the formula, The overall performance index to be maximized; For water dispersibility performance indicators; The hydrolysis kinetic constant of the slow-release unit; The characteristic evaluation time for ensuring collapse performance; For mud resistance performance indicators; , , These are the weighting factors for each performance metric; S3. Based on the current concentrations of various monomers obtained from the real-time analysis as initial conditions, and using the aforementioned reaction process kinetic model, an online optimization solution is obtained to find a control strategy that optimizes the preset comprehensive performance index of the final product. S4. The graft copolymerization process is regulated according to the control strategy obtained from the optimization solution.

2. The method for preparing starch-based concrete additives using multi-element synergistic grafting and digital control according to claim 1, characterized in that, The steps of pre-establishing the soft measurement model include: First, a series of grafting reaction experiments covering different process conditions were conducted, and online spectral data of the reaction system and samples for offline analysis were collected simultaneously during the experiments. Secondly, the absolute concentrations of various monomers in the sample were determined using offline analysis methods; Finally, partial least squares regression analysis was performed on the spectral data and the corresponding monomer concentrations to establish the mapping relationship between spectral information and actual concentrations. The mathematical model is as follows: ； In the formula, This is the true monomer concentration matrix obtained through offline analysis; This is a matrix of online spectral data acquired synchronously. This is the regression coefficient matrix of the PLS model obtained through regression training; Let be the residual matrix of the model.

3. The method for preparing starch-based concrete additives using multi-element synergistic grafting and digital control according to claim 1, characterized in that, The steps for online optimization of the control strategy are implemented through a model predictive control algorithm, which cyclically executes the complete optimization process within the rolling time domain. First, based on the current concentrations of the various monomers as initial conditions, the reaction process kinetic model is used to predict the evolution trajectory of the reaction system state in the future prediction time domain under different alternative control strategies. Then, an open-loop optimization problem with the goal of optimizing the overall performance index of the final product is solved to obtain the optimal control sequence covering multiple future steps; Finally, only the first control action in the optimal control sequence is executed, and the optimization process is repeated at the next control time.

4. The method for preparing starch-based concrete additives with multi-element synergistic grafting and digital control according to claim 3, characterized in that, The steps for solving the open-loop optimization problem include: Given the constraints of the process model, initial state, control input, and state variables, find the optimal control sequence that maximizes the following objective function: ； In the formula, This is the current control moment; To predict the length of the time domain; For at any time Predicted values ​​for the end state in the prediction time domain; For at any time Control inputs for planning.

5. The method for preparing starch-based concrete additives using multi-element synergistic grafting and digital control according to claim 4, characterized in that, The steps of regulating the reaction process according to the control strategy include: Based on the optimized control strategy, at least one or more process variables selected from the following are adjusted in real time: The dripping flow rate of each monomer solution, the dripping flow rate of the initiator solution, and the heating or cooling power of the reactor jacket.

6. The method for preparing starch-based concrete additives using multi-element synergistic grafting and digital control according to claim 5, characterized in that, The method further includes a reaction endpoint determination step, which is based on the results of the online optimization solution, specifically: When the predicted future gain of the final product's overall performance index is lower than a preset threshold, the reaction is determined to have reached its optimal endpoint and the reaction is terminated. The preset threshold is defined as follows: Within three consecutive control cycles, the predicted Increment All are less than or equal to 0.5% of the current Value, where, Predict the increment of the comprehensive performance index between adjacent control times; For at any time Predicting the next moment The overall performance index; For a moment The current overall performance index.

Images