A catalyst preparation process control system and method

Abstract

The application relates to the technical field of catalyst preparation process control, and discloses a catalyst preparation process control system and method, the method comprising the following steps: acquiring characteristic spectrum signals collected by each spectrum detection unit in real time, wherein the characteristic spectrum signals are used for reflecting titanium species chemical states; determining a reaction progress index of a current reaction system based on a comparison result of the characteristic spectrum signals and a plurality of preset multi-stage reference spectrum characteristics; under the conditions of satisfying a hydrolysis chelation rate matching constraint and a precipitation generation constraint, performing multivariate collaborative calculation on a deviation of the reaction progress index and a target progress to generate an operation variable group; when an abnormal scattering characteristic appears in the characteristic spectrum signals, switching to a protection linkage calculation mode to generate a protection linkage operation instruction; and synchronously regulating feeding rates, reaction temperatures and system pH values of micro-channel reaction units; the application can reduce batch-to-batch catalyst activity differences and improve the stability of a catalyst preparation process and product qualification rates.

Description

Technical Field

[0001] This application relates to the field of catalyst preparation process control technology, and more specifically to a catalyst preparation process control system and method. Background Technology

[0002] Currently, in the polyester industry, titanium-based catalysts are gradually replacing traditional antimony-based catalysts due to their high activity and environmental friendliness, becoming the mainstream catalyst for the production of high-quality polyester chips, differentiated fibers, and engineering plastics-grade polyester materials. Titanium-based catalysts are typically prepared from titanate esters as precursors through three sequential processes: hydrolysis, chelation, and curing. Titanate ester precursors are extremely sensitive to moisture; excessive local water concentration can trigger uncontrolled hydrolysis and polycondensation reactions, generating deactivated titanium dioxide precipitates, directly leading to the loss of the catalyst's active components. The hydrolysis stage requires precise control of the contact method and mixing scale between the titanium source and trace amounts of water. The chelation stage requires the timely introduction of chelating ligands before the hydrolysis products undergo deep polycondensation to stabilize the titanium-oxygen oligomer structure. The curing stage requires promoting structural rearrangement of the chelation products under isothermal conditions. The process parameters of the three stages are interrelated and mutually restrictive; deviations in parameters of any stage will propagate and amplify along the process flow, ultimately manifesting as fluctuations in catalyst activity and batch-to-batch inconsistency, affecting the stable industrial production of titanium-based catalysts.

[0003] The existing technology has the following problems: most titanium catalyst preparation devices are batch reactors, which have limited macroscopic scale of stirring and mixing, making it difficult to achieve uniform contact between the titanium source and trace amounts of water at the molecular level, resulting in local areas with excessively high water concentrations within the reactor; in process control, the operating parameters such as temperature, feed rate, and stirring speed of each stage of hydrolysis, chelation, and aging are adjusted separately by independent single-loop controllers, with no information exchange between the loops and a lack of coordinated control, leading to significant differences in catalyst activity and affecting product yield; to solve at least one of the above problems, this application proposes a catalyst preparation process control system and method. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the purpose of this application is to provide a control system and method for catalyst preparation processes, which can effectively solve the problems in the background technology. The specific technical solution of this application is as follows: A method for controlling a catalyst preparation process is applied to a control system comprising at least three tandem microchannel reaction units and multiple spectral detection units, wherein the multiple spectral detection units are respectively disposed between stages and at the final stage outlet of each microchannel reaction unit, comprising: The characteristic spectral signals acquired in real time by each spectral detection unit are obtained, whereby the characteristic spectral signals are used to reflect the chemical state of the titanium species; Based on the comparison results between the characteristic spectral signal and the preset multi-stage reference spectral features, the reaction process index of the current reaction system is determined. Using the characteristic spectral signal as the feedback benchmark, and under the conditions of meeting the hydrolysis and chelation rate matching constraints and precipitation formation constraints, the deviation between the reaction process index and the target process is calculated using a multivariate collaborative solution to generate a set of operational variables; among them, when abnormal scattering characteristics appear in the characteristic spectral signal, the system switches to the protection linkage solution mode to generate protection linkage operation instructions. The feed rate, reaction temperature, and system pH of the microchannel reaction unit are synchronously controlled according to the set of operating variables or protective linkage operation commands.

[0005] Specifically, the step of acquiring the characteristic spectral signal collected in real time by each spectral detection unit includes: performing differential processing on the original spectral signal collected by each spectral detection unit and the corresponding dark spectral reference signal, and performing baseline correction on the differentially processed signal to obtain the characteristic spectral signal.

[0006] Specifically, determining the reaction progress index of the current reaction system based on the comparison results between the characteristic spectral signal and the preset multi-stage reference spectral features includes: The absorption intensity ratio of at least two characteristic bands is extracted from the characteristic spectral signal, wherein the first characteristic band is sensitive to the bonding state of titanium-oxygen oligomers and the second characteristic band is sensitive to the charge transfer transition state from ligand to metal. The absorption intensity ratio is matched with the reference ratio range of the corresponding band in the preset multi-stage reference spectrum characteristics to determine the reaction process index of the current reaction system.

[0007] Specifically, determining the reaction progress index of the current reaction system also includes: Obtain the rate of change of scattering intensity of the characteristic spectral signal at the third characteristic band, which reflects the precipitation scattering effect; When the rate of change of scattering intensity is greater than the preset critical slope threshold, the corresponding absorption intensity ratio change trajectory is retrieved, and the reaction process index is compensated and corrected based on the change trajectory and the rate of change of scattering intensity.

[0008] Specifically, the multivariate collaborative solution of the deviation between the reaction process index and the target process to generate a set of operational variables includes: Based on the deviation between the reaction progress index and the target progress, under the conditions that the hydrolysis rate and chelation rate of each microchannel reaction unit are limited to a preset dynamic ratio range by the hydrolysis and chelation rate matching constraint, and the product of the concentration of titanium oxide oligomer and the concentration of hydroxyl group in the system is limited to below the precipitation dissolution volume based on the precipitation formation constraint, the cost function of feed rate, reaction temperature and system pH is optimized in the rolling time domain to generate a sequence of operating variables that makes the deviation converge in the prediction time domain, and the set of operating variables for the current control cycle is extracted from the sequence of operating variables.

[0009] Specifically, when abnormal scattering characteristics appear in the characteristic spectral signal, the system switches to the protection linkage calculation mode and generates protection linkage operation instructions, including: The response level of the protection linkage is determined based on the preset range of the amplitude of the abnormal scattering characteristics. When the response level indicates a warning state, the first protective linkage operation command generated includes simultaneously reducing the temperature setpoint of each microchannel reaction unit to the value corresponding to the preset safety temperature curve, and simultaneously increasing the feed rate of the chelating agent. When the response level indicates a critical state, the generated second protective linkage operation command, based on the first protective linkage operation command, triggers the opening of the emergency dilution valve located at the interstage connection of each microchannel reaction unit to inject anhydrous inert solvent into the main channel.

[0010] Specifically, the multivariate collaborative solution includes: in each control cycle, updating each constraint boundary value in real time according to the set of operating variables of the previous control cycle; when any variable among the feed rate, reaction temperature and system pH is limited to its corresponding constraint boundary and the deviation between the reaction progress index and the target progress does not converge, the constraint relaxation logic is triggered, and the constraint boundary with the lowest priority is extended to the second boundary value according to the preset constraint priority before resolving.

[0011] Specifically, the synchronous control of the feed rate, reaction temperature, and system pH of the microchannel reaction unit according to the set of operating variables or protective linkage operation commands includes: The temperature setpoint in the group of operating variables is used as a feedforward signal to control the jacket medium temperature of the corresponding microchannel reaction unit to form a preset axial temperature gradient along the material flow direction. The feed rate in the set of operational variables is used as a feedforward signal to control the stroke frequency of the water injection feed pump and the rotation speed of the chelating agent feed pump. The system pH value in the manipulated variable group is used as a feedback signal to control the pulse width modulation duty cycle of the acid-base regulator injection pump.

[0012] Specifically, after controlling the pulse width modulation duty cycle of the acid-base regulator injection pump, the method further includes: The pH microelectrode detection signal is acquired in real time at the inlet of the corresponding microchannel reaction unit, which is located downstream of the regulator injection point. The injection deviation is obtained by comparing the pH microelectrode detection signal with the system pH value in the manipulated variable group. The pulse width modulation duty cycle is corrected based on the injection deviation, so that the actual pH value at the injection point reaches the system pH value set by the operating variable group within the preset relaxation time.

[0013] A catalyst preparation process control system, used to implement the aforementioned catalyst preparation process control method, includes: The data acquisition module acquires the characteristic spectral signals collected in real time by each spectral detection unit, wherein the characteristic spectral signals are used to reflect the chemical state of titanium species; The reaction analysis module determines the reaction progress index of the current reaction system based on the comparison results between the characteristic spectral signal and the preset multi-stage reference spectral features. The operation analysis module uses the characteristic spectral signal as the feedback benchmark. Under the conditions of meeting the hydrolysis and chelation rate matching constraints and precipitation formation constraints, it performs multi-variable collaborative calculation on the deviation between the reaction process index and the target process to generate a set of operation variables. When abnormal scattering characteristics appear in the characteristic spectral signal, it switches to the protection linkage calculation mode and generates protection linkage operation instructions. The preparation process control module synchronously regulates the feed rate, reaction temperature, and system pH of the microchannel reaction unit based on the set of operating variables or protective linkage operation commands.

[0014] The beneficial effects of this application are as follows: By integrating microchannel continuous flow reaction hardware, online spectroscopy, and multivariate collaborative control, and using characteristic spectral signals as a unified feedback benchmark, the feed rate, reaction temperature, and system pH are optimized in the rolling time domain under the constraints of hydrolysis-chelation rate matching and precipitation formation. This can eliminate uncontrollable hydrolysis and precipitation caused by local concentration gradients in batch reactors, and achieve dynamic matching of hydrolysis rate, chelation efficiency, and maturation stability. The protective linkage mode responds in stages according to scattering characteristics, and timely suppresses precipitation risk through stepwise interventions such as cooling, increasing chelating agent, and emergency dilution. This reduces batch-to-batch differences in catalyst activity, improves product yield, and enhances the stability and robustness of the catalyst preparation process. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating a catalyst preparation process control method as described in this application. Figure 2 This is a schematic diagram of a catalyst preparation process control system in an embodiment of this application. Detailed Implementation

[0016] The present application will be further described in detail below with reference to the accompanying drawings and embodiments.

[0017] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0018] Hereinafter, the terms "first," "second," and other generic terms are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0019] refer to Figure 1 The illustration shows a specific embodiment of a catalyst preparation process control method according to this application. It is applied to a control system comprising at least three cascaded microchannel reaction units and multiple spectral detection units. The multiple spectral detection units are respectively located between stages and at the final stage outlet of each microchannel reaction unit, including: S101. Acquire the characteristic spectral signal collected in real time by each spectral detection unit, wherein the characteristic spectral signal is used to reflect the chemical state of the titanium species; S102. Based on the comparison results between the characteristic spectral signal and the preset multi-stage reference spectral features, determine the reaction process index of the current reaction system. S103. Using the characteristic spectral signal as the feedback benchmark, under the conditions of satisfying the hydrolysis and chelation rate matching constraints and precipitation formation constraints, the deviation between the reaction process index and the target process is calculated using a multivariate collaborative solution to generate a set of operational variables; among them, when abnormal scattering characteristics appear in the characteristic spectral signal, the system switches to the protection linkage solution mode to generate a protection linkage operation command. S104. Based on the set of operating variables or protective linkage operation instructions, the feed rate, reaction temperature and system pH of the microchannel reaction unit are synchronously controlled.

[0020] This embodiment provides a method for controlling a catalyst preparation process. The method operates within a control system comprising three cascaded microchannel reaction units and multiple spectral detection units. The three microchannel reaction units are arranged sequentially along the material flow direction, each undertaking continuous reactions in three stages: hydrolysis, chelation, and ripening. An interstage connecting pipe section with an optical detection window is provided between each adjacent two-stage microchannel reaction units, and the outlet pipe section of the final-stage microchannel reaction unit also has a detection window. Ultraviolet-visible fiber optic spectral probes are integrated into these detection windows, forming a distributed chain of spectral detection units along the material flow direction. Each spectral detection unit transmits the collected spectral signal to a spectrometer via optical fiber. The spectrometer converts the optical signal into a digital signal and sends it to the controller.

[0021] At the beginning of each control cycle, the controller sends an acquisition command to each spectral detection unit to obtain the raw spectral signals acquired in real time by each spectral detection unit. In order to suppress the influence of light source fluctuations and dark current drift of the detection unit itself on the useful signal, the controller performs two-step preprocessing of differential and baseline correction on each raw spectral signal.

[0022] Specifically, differential processing is achieved using the dark spectral reference signals pre-stored by each spectral detection unit. The dark spectral reference signal is the average value of a segment of spectral signal acquired by the corresponding spectrometer under the condition that the light source is turned off and no sample flows through the detection window. When the controller performs differential processing, it subtracts the value of the dark spectral reference signal at the corresponding wavelength point from the value of each wavelength point of the current original spectral signal to obtain the differential spectrum.

[0023] Specifically, baseline correction targets the gradually varying trend term in the differential spectrum caused by the spectral profile of the light source itself. The controller selects two wavelength intervals in the differential spectrum that are outside the characteristic absorption band of titanium species and where no obvious absorption peaks were detected as baseline reference intervals. Linear fitting is performed on the spectral data within these two intervals to obtain a baseline. The value of each wavelength point in the entire differential spectrum is subtracted from the value of the corresponding wavelength on the baseline to obtain the characteristic spectral signal. The characteristic spectral signal has been stripped of the influence of the light source profile and dark current, and its band absorption and scattering characteristics directly reflect the chemical state of titanium species in the liquid flowing through the detection window.

[0024] The processed characteristic spectral signal contains at least three information-carrying bands: a first characteristic band located between approximately 350 nm and 450 nm, where the absorption intensity mainly originates from the electronic transitions of oxygen bridge bonds in titanium-oxygen oligomers, and its intensity variation reflects the number of titanium-oxygen bonds formed by the hydrolysis-condensation reaction; a second characteristic band located between approximately 500 nm and 600 nm, where the absorption intensity corresponds to the charge transfer transitions of chelated ligands to the titanium central ion, and its intensity increases with the degree of chelation; and a third characteristic band located above approximately 700 nm, where the intrinsic absorption of titanium species is very weak, and the light attenuation is mainly caused by Rayleigh scattering of titanium dioxide nanoparticles, and its scattering intensity serves as an indicator of the risk of precipitation formation.

[0025] After the controller obtains the characteristic spectral signals of each spectral detection unit, it extracts the absorption intensity of the first characteristic band and the second characteristic band. For each characteristic band, the absorption intensity is the integral value of the absorbance at each wavelength point within the band. In order to eliminate common-mode interference such as flow rate fluctuations and slow changes in the transmittance of the detection window, the ratio of the absorption intensity of the first characteristic band to the absorption intensity of the second characteristic band is calculated to obtain the absorption intensity ratio.

[0026] Specifically, the multi-stage reference spectral characteristics are obtained in advance through calibration experiments and stored in the controller. The calibration experiments use the same microchannel reactor hardware configuration and raw materials as normal production. Throughout the entire process from initial feeding to complete termination of the reaction, characteristic spectral signals from each stage of the spectral detection unit are collected at preset time intervals, while simultaneously sampling and analyzing the actual progress state of the reaction system at that moment. For the characteristic spectral signal at each calibration moment, its absorption intensity ratio is calculated and correlated with the corresponding actual progress state. The absorption intensity ratio of the entire process is divided into multiple stages according to the sequence of the process, with each stage corresponding to a ratio interval and a stage number. For example, stage 1 is the low-temperature pre-hydrolysis induction period, with a corresponding absorption intensity ratio ranging from 0.15 to 0.25; stage 2 is the first selective chelation period, with a corresponding ratio ranging from 0.40 to 0.55; stage 3 is the temperature-controlled main hydrolysis period, with a corresponding ratio ranging from 0.70 to 0.85; stage 4 is the second deep chelation stabilization period, with a corresponding ratio ranging from 0.55 to 0.65; and stage 5 is the isothermal maturation period, with a corresponding ratio ranging from 0.30 to 0.40. These values ​​are only for illustrating the interval division method; in actual applications, they should be determined through calibration based on the specific catalyst formulation and process conditions.

[0027] The controller matches the currently calculated absorption intensity ratio with the reference ratio range for each stage. If the current ratio falls within the reference ratio range of a certain stage and is not in the overlapping area of ​​adjacent stages, the stage number of that stage is directly assigned to the reaction progress index. If the current ratio falls in the overlapping area of ​​two adjacent stages, the controller uses linear interpolation to calculate a continuous value between the two stage numbers as the reaction progress index. For example, when there is an overlapping area between the reference ratio ranges of stages 2 and 3, if the absorption intensity ratio is within this overlapping area, the reaction progress index slides proportionally between 2.0 and 3.0, thus transforming the process representation from a discrete stage to a continuous quantity.

[0028] When the initial formation of titanium dioxide precipitate occurs in the catalytic reaction system due to local anomalies, scattering will cause an increase in the intensity of the third characteristic band of the characteristic spectral signal. The appearance of precipitation indicates that some titanium precursors have not participated in the reaction along the expected path, and the apparent progress judged solely by the absorption intensity ratio may overestimate the actual effective reaction depth. To address this situation, the controller also introduces a compensation correction mechanism from the third characteristic band when determining the reaction progress index.

[0029] The controller calculates the rate of change of scattering intensity in the third characteristic band at the current moment, specifically by taking the difference between the scattering intensity at the current sampling moment and the scattering intensity at the previous sampling moment, and dividing it by the sampling interval. A critical slope threshold is pre-set within the controller. This threshold is determined based on statistical results of the scattering intensity fluctuation range under normal reaction conditions and can be taken as six times the standard deviation of the normal fluctuation. When the rate of change of scattering intensity exceeds the critical slope threshold, it indicates that the scattering intensity has experienced an abnormally rapid increase, and small precipitate particles have been formed. At this time, the controller retrieves previously stored data on the change trajectory of the absorption intensity ratio, which records the evolution trend of the absorption intensity ratio over the past dozen or so control cycles.

[0030] The controller generates a negative progress index compensation based on the extent to which the rate of change of the current scattering intensity exceeds a threshold, combined with the reaction propagation speed reflected by the trajectory of the absorption intensity ratio change. The negative sign of the compensation indicates that the actual effective reaction progress is lower than the apparent absorption progress, and its absolute value is proportional to the extent of the exceedance. The compensation is superimposed on the initial progress index obtained through absorption intensity ratio matching to obtain the corrected reaction progress index. This compensation correction enables the controller to obtain a reaction state assessment result closer to the actual situation when precipitation is still in the early reversible stage, providing accurate input for subsequent control decisions.

[0031] The controller uses the deviation between the corrected reaction process index and the preset target process as the driving force to perform multivariate collaborative calculation. The calculation adopts a rolling time-domain optimization method, which calculates the adjustment sequence of feed rate, reaction temperature and system pH value in the future prediction time domain in each control cycle, so as to minimize the weighted sum of squares of the deviation in the prediction time domain. The weighting weight gradually decreases from near to far along the prediction time domain to reflect the priority correction of recent deviations.

[0032] Specifically, the calculation process is carried out under the joint constraints of two conditions. The hydrolysis-chelation rate matching constraint stipulates that within the same microchannel reaction unit, the ratio of the hydrolysis rate to the chelation rate must be within a preset dynamic ratio range. The ratio range changes exponentially with the reaction progress; in the early stage of the process, the hydrolysis rate is allowed to be slightly faster than the chelation rate, with an upper limit of approximately 2.0 to 2.5, to ensure the generation of sufficient titanium oxide oligomers for subsequent chelation; in the middle stage of the process, the two are required to be close to equilibrium, and the ratio range narrows to 0.8 to 1.2; in the late stage of the process, the chelation rate can moderately exceed the hydrolysis rate, and the lower limit of the ratio range is widened to 0.4 to 0.5. Precipitation formation constraints stipulate that the product of the concentration of titanium oxide oligomer and the concentration of hydroxyl groups in the reaction system must not exceed the precipitation-dissolution volume. The precipitation-dissolution volume is a function of temperature; the higher the temperature, the greater the solubility of titanium oxide oligomer, and the higher the allowable upper limit of the precipitation-dissolution volume. The controller obtains the corresponding upper limit of the precipitation-dissolution volume based on the current temperature setpoint of each microchannel reaction unit, and updates the boundary of the precipitation formation constraint in real time based on the current concentration of titanium oxide oligomer and hydroxyl groups calculated from the online spectral signal.

[0033] The rolling time-domain optimization solver searches for the sequence of operational variables that minimizes the cost function within the feasible solution space defined by the two constraints using a sequential quadratic programming method. The search starts at the set of operational variables actually executed in the previous control cycle and the extrapolated prediction. Once the optimal sequence is obtained, the operational variables corresponding to the upcoming first control cycle are extracted from the sequence to form a set of operational variables, which serves as the output command under normal control mode.

[0034] During the optimization search, if any variable among the feed rate, reaction temperature, or system pH value has reached its constraint boundary, and the reaction progress exponential deviation has not yet converged to the allowable range, the controller initiates a constraint relaxation mechanism. The constraint priorities are set as follows: the priority of precipitation formation constraints is higher than the upper limit constraint of the ratio in the hydrolysis-chelation rate matching constraints, and the upper limit constraint of the ratio has a higher priority than the lower limit constraint of the ratio. When relaxation is needed, the controller expands the boundary value of the lowest priority constraint outwards, for example, expanding the lower limit from 0.4 to 0.3, and re-executes the solution within the new feasible region after relaxation. If convergence still fails after relaxation, the next lowest priority constraint is relaxed in order of priority; this ensures that the controller prioritizes preventing precipitation under extreme operating conditions.

[0035] In addition to the normal collaborative calculation mode, the controller also has a built-in protective linkage calculation mode to deal with abnormal scattering characteristics in the characteristic spectral signal. The controller continuously monitors the increase in scattering intensity of the third characteristic band relative to the normal baseline; the normal baseline is the moving average of the scattering intensity over the previous ten minutes; when the increase enters the range of 30% to 80%, the controller determines it to be in a warning state, switches to the protective linkage calculation mode, and generates the first protective linkage operation command. This command includes two synchronous actions: rapidly reducing the temperature setpoint of each microchannel reaction unit to 60% to 70% of the normal operating temperature curve, and simultaneously increasing the chelating agent feed rate to 1.5 to 2 times the current normal feed rate. The cooling action is achieved by adjusting the opening of the jacket heat transfer medium flow valve, and the chelating agent increment is achieved by increasing the speed of the corresponding metering pump.

[0036] When the scattering intensity increases by more than 80%, the controller determines it to be in a critical state. In addition to executing the first protective linkage operation command mentioned above, it issues a second protective linkage operation command: triggering the opening of the emergency dilution valves located on the inter-stage connecting pipes of each microchannel reaction unit, injecting anhydrous isopropanol or anhydrous butanol, dried with molecular sieves, into the main flow channel within the connecting pipe. The injection flow rate is controlled between 0.5 and 1 times the current flow rate of the main flow channel, and the injection pulse lasts for three to five seconds. The injection of anhydrous inert solvent instantly reduces the local concentration of titanium species within the pipe section, causing the product of the titanium oxide oligomer concentration and the hydroxyl concentration to rapidly fall below the precipitation dissolution volume, kinetically blocking the continued growth of already formed crystal nuclei.

[0037] The controller sends the set of operating variables or protective linkage commands to each actuator to achieve synchronous control of the feed rate, reaction temperature, and system pH. For the reaction temperature, the controller uses the temperature setpoint from the set of operating variables as a feedforward control signal and outputs it to the temperature control loop of the jacket of each stage of the microchannel reaction unit. The jacket of each stage of the microchannel reaction unit is divided into several independent temperature control sections, each set with a different temperature value sequentially along the material flow direction, thereby establishing an axial temperature gradient within the microchannel reaction unit. For example, the first stage of the microchannel reaction unit maintains a lower temperature at the front to meet the requirements of low-temperature pre-hydrolysis, while the rear stage gradually increases the temperature to connect to the next stage of the reaction unit; the second stage corresponds to the medium temperature required for main hydrolysis; and the third stage maintains a constant temperature to meet the ripening requirements.

[0038] For the feed rate, the controller uses the corresponding feed rate value from the controlled variable group as a feedforward control signal, outputting it to the drive controllers of the water injection pump and the chelating agent pump respectively. The water injection pump is a plunger metering pump, whose mass flow rate is linearly related to the stroke frequency; the chelating agent pump is a gear metering pump, whose mass flow rate is linearly related to the pump speed. The controller directly calculates the required stroke frequency and speed values ​​and writes them into the corresponding drive controllers.

[0039] For pH control of the system, the controller uses the system pH setpoint in the operation variable group as the reference input for the feedback loop; the acid-base regulator is a diluted inorganic acid or organic base solution, which is added by an injection pump controlled by pulse width modulation; the pH control loop of the controller targets the setpoint and outputs a pulse width modulation duty cycle signal to the injection pump driver; the larger the duty cycle, the more regulator is injected per unit time.

[0040] A pH microelectrode is installed in the pipe section downstream of the regulator injection point and before the inlet of the microchannel reaction unit. This microelectrode monitors the actual pH value of the liquid after the injection point in real time and feeds the detection signal back to the controller. The controller compares the actual pH value fed back by the microelectrode with the system pH setpoint in the manipulated variable group to obtain the injection deviation. Based on the injection deviation, the controller performs a secondary closed-loop correction on the pulse width modulation duty cycle. The correction algorithm adjusts the duty cycle in a proportional-integral manner so that the actual pH value at the injection point approaches and stabilizes at the setpoint within three to five control cycles. The closed-loop correction compensates for the dead zone delay in the regulator delivery pipeline and the dynamic deviation introduced by the uneven mixing of the regulator and the main fluid, ensuring that the liquid entering the microchannel reaction unit has reached the set pH condition.

[0041] This application integrates microchannel continuous flow reaction hardware, online spectroscopy, and multivariate collaborative control. Using characteristic spectral signals as a unified feedback benchmark, it performs rolling time-domain optimization of feed rate, reaction temperature, and system pH while satisfying constraints on hydrolysis-chelation rate matching and precipitation formation. This eliminates uncontrollable hydrolysis and precipitation caused by local concentration gradients in batch reactors, achieving dynamic matching of hydrolysis rate, chelation efficiency, and maturation stability. The protective linkage mode responds in stages based on scattering characteristics, promptly suppressing precipitation risks through stepwise interventions such as cooling, increasing chelating agents, and emergency dilution. This reduces batch-to-batch differences in catalyst activity, improves product yield, and enhances the stability and robustness of the catalyst preparation process.

[0042] Furthermore, the characteristic spectral signal acquired in real time by each spectral detection unit is obtained, including: performing differential processing on the original spectral signal acquired by each spectral detection unit and the corresponding dark spectral reference signal, and performing baseline correction on the differentially processed signal to obtain the characteristic spectral signal.

[0043] In this embodiment, after receiving the light signal collected by the probe, the ultraviolet-visible fiber optic spectrometer in the spectral detection unit converts the light intensity distribution within the wavelength range into a digitized raw spectral signal according to a fixed integration time. In addition to containing the absorption and scattering information of light of a specific wavelength by the liquid flowing through it, the signal also includes the gradual change trend of the spectral profile of the light source itself, the background shift caused by the dark current of the spectrometer, and the drift introduced by the slow change of the transmittance of the detection window with the use time.

[0044] The controller stores a set of dark spectral reference signals for each spectral detection unit. These signals are obtained by arithmetically averaging one hundred continuously acquired spectra with the same integration time as normal acquisition, under conditions where the light source of the detection unit is off and no sample flows through the detection window. This reflects the baseline of the dark current response generated by the detection circuit itself when there is no light signal input. During differential processing, the controller subtracts the intensity value corresponding to each wavelength point in the original spectral signal acquired in the current control cycle from the intensity value of that wavelength point in the dark spectral reference signal. The result of the subtraction operation constitutes the differential spectrum. Since the dark spectral reference signal and the original spectral signal come from the same spectrometer and have the same integration time, the dark current noise components they contain are highly consistent, and this component is effectively canceled out after subtraction. Simultaneously, slowly varying factors such as changes in the transmittance of the detection window, which have an approximately proportional scaling effect on the overall light intensity level, are recorded during the acquisition of the dark spectral reference signal, and the differential processing also removes their contribution from the signal.

[0045] Specifically, the differential spectrum retains the influence of the light source's own spectral profile. The radiant intensity of the ultraviolet-visible light source varies at different wavelengths; the radiant intensity is typically low in the longer wavelength regions beyond the characteristic absorption band of titanium species, and higher in the mid-wavelength range. This non-flat spectral profile introduces a slowly varying trend term with wavelength into the differential spectrum. To eliminate this slowly varying trend term, the controller performs baseline correction on the differential spectrum. During baseline correction, the controller selects two reference wavelength intervals in the differential spectrum: the first interval is located approximately 50 nanometers before the start wavelength of the characteristic absorption band of titanium species, and the second interval is located approximately 50 nanometers after the end wavelength of the characteristic absorption band. Within these intervals, the absorption of light by the flowing liquid is extremely weak, and the recorded intensity signals are primarily derived from the light source's own radiation. The controller extracts the differential spectral intensity values ​​corresponding to each wavelength point within these two reference wavelength intervals and performs linear regression fitting to obtain a baseline that varies linearly along the wavelength dimension. The linear regression uses wavelength as the independent variable and the differential spectral intensity value as the dependent variable, determining the slope and intercept of the baseline based on the criterion of minimizing the sum of squared residuals. After obtaining the baseline, the controller subtracts the intensity value of the corresponding wavelength point on the baseline from the intensity value of each wavelength point in the entire differential spectrum, and the difference obtained is the characteristic spectral signal.

[0046] After differential and baseline correction processing, the intensity variations of the characteristic spectral signals obtained are decoupled from non-reactive factors such as fluctuations in light source radiation intensity, spectrometer dark current drift, and window transmittance decay. They change only with the chemical state of titanium species in the flowing liquid. The increase or decrease in the absorption peak intensity in the 350-450 nm range directly corresponds to the increase or decrease in the number of oxygen bridges in the titanium-oxygen oligomer; the intensity variation of the absorption band in the 500-600 nm range tracks the strength of the charge transfer transition from the chelated ligand to the titanium central ion; and the baseline rise above 700 nm reflects the change in transparency due to the scattering effect of titanium dioxide nanoparticles. This preprocessing provides a high-fidelity, interference-free feedback signal source for the subsequent determination of the reaction progress index and multivariate collaborative calculations. This ensures that even detector response drift caused by light source aging or changes in ambient temperature will not be incorrectly interpreted as a change in the chemical state of titanium species within the reaction system, guaranteeing the reliability and consistency of the control system during long-term continuous operation.

[0047] Furthermore, based on the comparison results between the characteristic spectral signal and the preset multi-stage reference spectral characteristics, the reaction progress index of the current reaction system is determined, including: S301. Extract the absorption intensity ratio of at least two characteristic bands from the characteristic spectral signal, wherein the first characteristic band is sensitive to the bonding state of titanium-oxygen oligomers and the second characteristic band is sensitive to the charge transfer transition state from ligand to metal. S302. Match the absorption intensity ratio with the reference ratio range of the corresponding band in the preset multi-stage reference spectrum characteristics to determine the reaction process index of the current reaction system.

[0048] In this embodiment, the controller extracts the absorption intensities of two characteristic bands from the preprocessed characteristic spectral signal and calculates their ratio to determine the reaction progress. The characteristic spectral signal presents as a continuous curve in the wavelength dimension. The first characteristic band is located in the 350 nm to 450 nm range. In this band, the electronic transitions of oxygen bridge bonds in the titanium-oxygen oligomer produce characteristic absorption, and the absorption intensity increases monotonically with the increase of the number of titanium-oxygen bonds formed by the hydrolysis-condensation reaction. The second characteristic band is located in the 500 nm to 600 nm range. In this band, the charge transfer transitions of the chelated ligand to the titanium central ion produce characteristic absorption, and the absorption intensity increases monotonically with the increase of the coordination degree between the chelated ligand and the titanium central ion.

[0049] The controller performs a trapezoidal integration on the absorbance values ​​at each wavelength point within the first characteristic band. The starting and ending wavelengths of the integration are taken as the absorbance values ​​corresponding to the wavelength points at the left and right boundaries of the band, respectively. The absorbance values ​​at intermediate wavelength points are weighted and accumulated according to wavelength intervals to obtain the absorption intensity of the first characteristic band. The same integration operation is performed on the second characteristic band to obtain the absorption intensity of the second characteristic band. The controller divides the absorption intensity of the first characteristic band by the absorption intensity of the second characteristic band to obtain the absorption intensity ratio. The reason for choosing a ratio instead of a single band intensity is that common-mode interference factors such as overall fluctuations in light source intensity, slow decay of transmittance of the detection window, and optical path fine-tuning caused by changes in flow rate will have a nearly proportional impact on the two bands. By using the ratio calculation, these interference factors can be canceled out as common factors in the numerator and denominator, resulting in a robust spectral characteristic quantity that is insensitive to interference.

[0050] Preferably, the absorption intensity ratio integrates the independent chemical information carried by the two bands into a single characteristic quantity, which simultaneously encodes information on both the hydrolysis-condensation depth and the chelation progress. When the hydrolysis reaction leads and the chelation reaction lags behind, the absorption intensity of the first characteristic band is higher while the absorption intensity of the second characteristic band is still lower, resulting in a higher ratio. As the chelation reaction gradually catches up, the absorption intensity of the second characteristic band increases, and the ratio falls back. When the two tend to be in equilibrium, the ratio stabilizes in a certain intermediate range. The numerical change trajectory provides a continuously trackable quantitative basis for subsequent stage matching, enabling the controller to perceive the relative advancement relationship of the two reaction streams of hydrolysis and chelation in the reaction system in real time based solely on the online spectral signal without relying on indirect means such as reaction time timing or cumulative feed measurement.

[0051] The controller matches the calculated absorption intensity ratio with a pre-stored multi-stage reference ratio range, outputting the reaction progress index of the current reaction system. The multi-stage reference ratio range originates from a dataset established by calibration experiments. These calibration experiments are conducted using the same microchannel reactor hardware configuration, employing the same batch of titanate precursor, water for hydrolysis, and chelating ligands as those used in formal production. In the calibration experiments, characteristic spectral signals from each spectral detection unit are sequentially acquired at fixed time intervals throughout the entire process from feed to maturity, and the absorption intensity ratio is calculated. Simultaneously, samples are taken from the corresponding detection locations, and the conversion and chelation rates of the titanate are determined by chemical titration. The combined index of conversion and chelation rates is used as the true process state label at that moment. All calibration data are arranged in the order of evolution from the initial state to the final state according to the true process state label. Inflection points where the slope of the absorption intensity ratio changes significantly are identified on the curve of the process evolution, and these inflection points serve as the boundaries between adjacent stages. The fluctuation range of the absorption intensity ratio in each stage is statistically determined. The mean of the ratios of all calibrated sample points within that stage, plus or minus two standard deviations, is taken as the lower and upper limits of the reference ratio for that stage. A multi-stage reference spectral characteristic table is generated in the controller, consisting of three fields: stage number, lower limit of the reference ratio, and upper limit of the reference ratio. For example, the reference ratio range for stage number 1 is 0.15 to 0.25, for stage number 2 it is 0.40 to 0.55, for stage number 3 it is 0.70 to 0.85, for stage number 4 it is 0.55 to 0.65, and for stage number 5 it is 0.30 to 0.40.

[0052] Specifically, the matching process is executed once by the controller in each control cycle. The controller reads the currently calculated absorption intensity ratio and iterates through the reference ratio ranges of each stage in the multi-stage reference spectral characteristic table. If the current ratio falls between the lower and upper limits of the reference ratio for a certain stage and there is no overlap between that stage and adjacent stages, the controller directly assigns the stage number of that stage to the reaction progress index. If the current ratio falls within the overlap of two adjacent stages, the controller uses a linear interpolation method to calculate the continuous progress index value. Specifically, the lower limit of the overlap corresponds to the stage number of the previous stage, and the upper limit of the overlap corresponds to the stage number of the next stage. The floating-point value between the two stage numbers is determined proportionally to the relative position of the current ratio within the overlap. This continuous processing transforms the process representation from a discrete step transition to a smooth slide, more closely reflecting the physical nature of the continuous evolution of the reaction process.

[0053] Preferably, multidimensional spectral information is compressed into a single-dimensional process index to achieve a quantitative expression of the overall progress of the reaction system. The calibration experiment uses chemical titration to give the spectral ratio a precise meaning of the process state, establishing a mapping bridge between the spectral signal and the chemical reality. This allows the controller to infer the current chemical progress depth of the reaction based solely on the online spectral signal, avoiding the control command jumps that may occur at the stage boundaries when judging discrete stages. This provides a smoothly changing deviation input signal for the downstream multivariate collaborative solver, which is conducive to the solver generating a sequence of operational variables with smaller fluctuations, reducing the frequent actions of the actuator, and extending the service life of the equipment.

[0054] Furthermore, determining the reaction progress index of the current reaction system also includes: S401. Obtain the rate of change of scattering intensity of the characteristic spectral signal at the third characteristic band that reflects the precipitation scattering effect; S402. When the rate of change of scattering intensity is greater than the preset critical slope threshold, retrieve the corresponding absorption intensity ratio change trajectory, and compensate and correct the reaction process index based on the change trajectory and the rate of change of scattering intensity.

[0055] In this embodiment, during the process of determining the reaction progress index, in addition to tracking the normal progress of hydrolysis and chelation using the absorption intensity ratio, the controller also continuously monitors the changes in scattering intensity in the third characteristic band to capture the initial precipitation signal. The third characteristic band is located in the wavelength region above 700 nanometers. In this wavelength range, the intrinsic electronic transition absorption of titanium species is extremely weak, and the light attenuation is mainly caused by Rayleigh scattering of incident light by titanium dioxide nanoparticles suspended in the flowing liquid. The scattering intensity is positively correlated with the total volume concentration of the particles.

[0056] The controller extracts the scattering intensity of the third characteristic band from the characteristic spectral signal by integrating the absorbance values ​​at each wavelength point within the band. The integration interval is from the starting wavelength of the band to the upper limit of the spectrometer's detection range. After obtaining a scattering intensity value in each control cycle, the controller stores it in a rolling buffer with a length equal to the last ten control cycles. The rate of change of scattering intensity is calculated by dividing the difference between the scattering intensity value of the current control cycle and the scattering intensity value of the previous control cycle by the duration of a single control cycle. The result is the rate of change per second, expressed in absorbance units. When no precipitate has formed in the reaction system, the scattering intensity fluctuates only within a small random range, with the rate of change oscillating slightly around zero. Once titanium dioxide crystal nuclei begin to form and gradually grow, the scattering intensity immediately shows a unidirectional increase, and the rate of change enters a sustained positive value region from the near-zero region.

[0057] Preferably, the rate of change in scattering intensity provides the controller with a sensitive indicator for early warning of the precipitation process. The rate of change responds more quickly to initial weak scattering enhancement and is not falsely triggered by slow drift in the scattering intensity baseline caused by gradual contamination of the detection window or long-term aging of the light source. The amplitude of the slow drift during adjacent control cycles is much lower than the amplitude during precipitation formation. Using the rate of change in scattering intensity as the trigger condition for process index correction allows the controller to intervene before the precipitated particles cause irreversible catalyst deactivation, thus securing a critical reaction time window for protective linkage operations.

[0058] When the rate of change of scattering intensity exceeds the critical slope threshold, the controller initiates a compensation and correction program for the reaction progress index. The critical slope threshold is automatically set by the controller based on the statistical characteristics of the rate of change of scattering intensity under normal reaction conditions. Specifically, in the initial stage of each batch of production, before any precipitated products are present in the reaction system, the controller continuously collects samples of the rate of change of scattering intensity for the first thirty control cycles, calculates the arithmetic mean and standard deviation of these samples, and takes the arithmetic mean plus six times the standard deviation as the critical slope threshold. The basis for this six-times factor is that, under the assumption of random fluctuations within a normal distribution, the probability of the rate of change exceeding the mean plus or minus six times the standard deviation is extremely low.

[0059] After triggering the compensation correction procedure, the controller retrieves previously stored data on the absorption intensity ratio variation trajectory from the rolling buffer. This trajectory is a time-series sequence of the absorption intensity ratio over the last fifteen control cycles, reflecting the recent evolution trend of the hydrolysis and chelation progression relationship within the reaction system before the scattering alarm occurs. Based on the recent trend of the absorption intensity ratio in the trajectory, the controller determines the approximate progression stage of the current reaction: if the ratio shows a continuous upward trend, it indicates that the system is in the hydrolysis acceleration phase; if the ratio remains flat, it indicates that the system is in the equilibrium progression phase; if the ratio continues to decrease, it indicates that the system is in the chelation catch-up phase.

[0060] Specifically, the compensation correction logic is based on the physical understanding of the consumption of titanium precursors by precipitation side reactions. The formation of titanium dioxide precipitate consumes some of the titanate precursor. The consumed precursor neither participates in hydrolysis to form titanium oxide oligomers nor enters the subsequent chelation reaction channel, resulting in lower absorption intensities in the first and second characteristic bands compared to the case without precipitation. The apparent progress index calculated from the ratio of these two bands fails to reflect the loss of titanium precursors due to precipitation, and the apparent value overestimates the actual effective reaction depth. The compensation correction amount is negative to lower the progress index, and its absolute value is determined by two factors: the first factor is the ratio of the magnitude of the scattering intensity change rate exceeding the critical slope threshold to the threshold. The larger the magnitude of the exceedance, the faster the particle formation rate, and the compensation amount increases linearly accordingly. The second factor is a stage sensitivity coefficient selected based on the stage determination result of the absorption intensity ratio change trajectory. A higher sensitivity coefficient is selected during the hydrolysis acceleration period, the impact of titanium precursor loss on the process is more significant during the rapid product accumulation stage, a medium sensitivity coefficient is selected during the equilibrium advancement period, and a lower sensitivity coefficient is selected during the chelation catch-up period. The absolute value of the compensation is limited to no more than half the span of the current process index corresponding to the current stage, to prevent control oscillations caused by excessively large single corrections. The controller adds the calculated compensation to the initial process index obtained by matching the absorption intensity ratio, and outputs the corrected reaction process index.

[0061] Preferably, the consumption of titanium precursor by precipitation side reactions is incorporated into the reaction process evaluation model, so that the process index delivered by the controller to the downstream solver under abnormal operating conditions is closer to the actual effective reaction depth. Through compensation correction, the process index is reasonably reduced, and the downstream collaborative solver obtains the corresponding set of operating variables. The stage determination of the change trajectory provides the compensation logic with information on the current kinetic stage of the reaction, so that the same intensity of scattering alarm obtains a stronger correction force in the reaction acceleration period than in the equilibrium advancement period, realizing an adaptive correction strategy that matches the dynamic characteristics of the reaction system.

[0062] Furthermore, a multivariate collaborative solution is performed on the deviation between the reaction progress index and the target progress to generate a set of operational variables. This includes: based on the deviation between the reaction progress index and the target progress, under the conditions that the hydrolysis rate and chelation rate of each microchannel reaction unit are limited to a preset dynamic ratio range by the hydrolysis and chelation rate matching constraint, and the product of the concentration of titanium oxide oligomers and the concentration of hydroxyl groups in the system is limited to below the precipitation dissolution volume by the precipitation formation constraint, the cost function of the feed rate, reaction temperature and system pH is optimized in the rolling time domain to generate a sequence of operational variables that makes the deviation converge in the prediction time domain, and the set of operational variables for the current control cycle is extracted from the sequence of operational variables.

[0063] In this embodiment, the multivariate collaborative solution process receives the deviation between the reaction progress index and the target progress as the driving input. Under constraints, it generates a set of operational variables to be executed in the current control cycle through rolling time-domain optimization. The deviation is obtained by subtracting the preset target progress from the corrected reaction progress index. The target progress is the expected process trajectory unfolded along the time axis. A lower value is taken in the initial stage of the batch to allow for reaction start-up margin, a linear ramp-up is taken in the middle of the batch to match capacity demand, and a constant value close to the final value is taken at the end of the batch to ensure complete reaction.

[0064] The controller inputs the deviation into the rolling time-domain optimization solver. The prediction time domain is the length of the next eight control cycles, and the control time domain is the length of the next three control cycles. Within the prediction interval beyond the control time domain, the operated variables maintain their values ​​from the end of the control time domain. The cost function is defined as the square of the difference between the predicted reaction progress index and the target progress value for each control cycle within the prediction time domain, multiplied by the weight corresponding to that cycle, and then summed. The weights decrease from near to far along the prediction time domain direction, with a weight of 1.0 for the first cycle. Each subsequent cycle's weight is multiplied by a decay factor of 0.85, ensuring the solver prioritizes rapid correction of recent deviations.

[0065] Specifically, the multivariate collaborative solution seeks the sequence of operational variables that minimizes the cost function under two constraints. The first constraint is a hydrolysis-chelation rate matching constraint, which requires that within the same microchannel reaction unit, the ratio of the hydrolysis rate to the chelation rate must not exceed the dynamic ratio range corresponding to the current reaction progress index. The dynamic ratio range, which varies with the progress index, is determined by statistical analysis of calibration experimental data. When the progress index is between 0 and 0.2, the hydrolysis rate is allowed to moderately lead the chelation rate, with the ratio range set at 1.0 to 2.5 to ensure sufficient initial generation of titanium-oxygen oligomers for subsequent chelation. When the progress index is between 0.2 and 0.6, the two are required to tend to balance, and the ratio range narrows to 0.8 to 1.2. After the progress index exceeds 0.6, the chelation rate is allowed to moderately exceed the hydrolysis rate, and the ratio range is adjusted to 0.4 to 1.0. The reaction rate is determined by the temperature setpoint and feed rate in the current set of operating variables. The controller calculates the rate constant from the temperature based on the Arrhenius relation, and calculates the reactant concentration from the feed rate combined with the microchannel residence time distribution, thus estimating the current values ​​of the hydrolysis rate and chelation rate.

[0066] Specifically, the second constraint is the precipitation formation constraint, which requires that the product of the titanium oxide oligomer concentration and the hydroxyl concentration in the system must not exceed the precipitation-dissolution volume at any given time. The functional relationship between precipitation-dissolution volume and temperature is pre-established through isothermal solubility testing. In this test, hydroxyl-containing components are added dropwise to a titanium oxide oligomer solution of known concentration under isothermal conditions. The critical point of precipitation is monitored using online scattering spectroscopy, and the critical concentration product at different temperatures is recorded, forming a temperature-precipitation-dissolution volume comparison table stored in the controller. The controller retrieves the corresponding upper limit of precipitation-dissolution volume based on the current temperature setpoint of each microchannel reaction unit, and calculates the current titanium oxide oligomer concentration by combining the absorption intensity of the first characteristic band and the scattering intensity of the third characteristic band in the characteristic spectral signal. The current hydroxyl concentration is calculated from the feed rate and reaction conversion rate, and the precipitation formation constraint is checked in real-time to ensure it is not exceeded.

[0067] Furthermore, the rolling time-domain optimization solver searches for the sequence of operands that minimizes the cost function within the feasible region. The feasible region is jointly defined by the physical constraints of the operands themselves, the matching constraints of the hydrolysis-chelation rate, and the precipitation formation constraints. The search employs a sequential quadratic programming method. The starting point is the optimal operand sequence obtained from the previous control cycle, shifted forward by one cycle, and supplemented with the unchanged predicted values. The sequential quadratic programming method approximates the nonlinear constrained optimization problem as a constrained quadratic programming subproblem at the current iteration point. By solving this subproblem, the search direction is obtained, and a one-dimensional search is performed along this direction to determine the step size and update the iteration point. This search is repeated until the decrease in the cost function value between two adjacent iterations is less than 0.1% or the preset maximum number of iterations is reached. The optimal operand sequence obtained at the end of the search includes the feed rate setpoint, the temperature setpoint of each stage of the microchannel reaction unit, and the system pH setpoint for each control cycle in the prediction time domain. The controller extracts the operand set of the first control cycle from this sequence as the instruction output to the execution layer for the current control cycle.

[0068] If, during the solution process, any variable among the feed rate, reaction temperature, or system pH value is found to touch its operating boundary or constraint boundary, and the exponential deviation of the reaction process still has not converged to the preset allowable range, the controller triggers constraint relaxation logic. The constraint priority is set as follows: precipitation formation constraint cannot be relaxed; the lower limit constraint in the hydrolysis-chelation rate matching constraint has a lower priority; and the upper limit constraint has a medium priority. During relaxation, the boundary value of the lower limit constraint is first expanded outward by 20%, and the solution is restarted. If convergence is still not achieved, the boundary value of the upper limit constraint is expanded outward by 10%, and the solution is restarted.

[0069] Preferably, by using the exponential deviation of the reaction process as the driving force, and under the explicit protection of two process constraints, the optimal coordinated adjustment of three sets of operating variables—feed rate, reaction temperature, and system pH—is solved simultaneously, eliminating coupling conflicts that arise when each parameter is adjusted separately. Rolling time-domain optimization plans the operation trajectory for multiple future steps in each control cycle, giving the adjustment of operating variables feedforward predictive capabilities and avoiding over-adjustment and oscillations caused by the large inertial lag of the reaction process. Precipitation formation constraints directly incorporate the product of titanium oxide oligomer concentration and hydroxyl concentration into the optimization constraints. Constraint relaxation logic releases secondary constraints in order of priority under extreme operating conditions, ensuring that the system prioritizes preventing precipitation when the degree of freedom of adjustment is insufficient.

[0070] Furthermore, when abnormal scattering characteristics appear in the characteristic spectral signal, the system switches to the protection linkage calculation mode to generate protection linkage operation commands, including: S601. Determine the response level of the protection linkage based on the preset range of the amplitude of the abnormal scattering characteristics. S602. When the response level indicates a warning state, the generated first protective linkage operation command includes synchronously reducing the temperature setpoint of each microchannel reaction unit to the value corresponding to the preset safety temperature curve, and synchronously increasing the feed rate of the chelating agent. S603. When the response level indicates a critical state, the generated second protective linkage operation command, based on the first protective linkage operation command, triggers the opening of the emergency dilution valve located at the interstage connection of each microchannel reaction unit to inject anhydrous inert solvent into the main channel.

[0071] In this embodiment, the controller continuously tracks the scattering intensity of the third characteristic band in the characteristic spectral signal. When an abnormal scattering feature is detected, the controller determines the response level of the protection linkage based on the preset range in which its amplitude falls. The amplitude of the abnormal scattering feature is taken as the percentage increase of the current scattering intensity relative to the normal baseline. The normal baseline is the moving average of the scattering intensity over the most recent sixty control cycles. The calculation window for this average value is sixty control cycles to cover the natural fluctuations in scattering intensity caused by minor fluctuations in feed and gradual changes in ambient temperature under normal operating conditions. The percentage increase is obtained by subtracting the normal baseline from the current scattering intensity and then dividing by the normal baseline. When this percentage exceeds the first preset boundary of 30%, the controller determines that the response level of the protection linkage enters the warning state; when this percentage further exceeds the second preset boundary of 80%, the response level is upgraded to a critical state. The 30% and 80% boundary values ​​are derived from observations of precipitation kinetics experiments. The experiments show that when the scattering intensity increase is below 30%, the titanium dioxide particles appearing in the system are mainly primary crystal nuclei with a diameter of less than ten nanometers. At this point, a gentle adjustment of process parameters is sufficient to redissolve them. When the increase exceeds 80%, the crystal nuclei have grown into particles larger than tens of nanometers, and the Brownian motion deagglomeration is insufficient to dissolve them; a more forceful dilution intervention is necessary. The interval between these two boundary values ​​constitutes a warning state, providing a window of opportunity for process parameter intervention.

[0072] Preferably, the protection linkage solution mode achieves a step-by-step response that matches the severity of precipitation evolution, avoiding the radical measures of drastically reducing or stopping production when precipitation is still in a reversible stage, and also preventing the delay in treatment due to relying only on mild adjustment after precipitation has entered an irreversible growth stage; the method of calculating the growth percentage based on the normal baseline automatically adapts to the rise in scattering intensity baseline caused by slow contamination during long-term use of the detection window, and will not generate false alarms due to natural baseline drift.

[0073] When the response level indicates a warning state, the controller generates a first protective linkage operation command, which simultaneously executes two actions: cooling and chelating agent enhancement. The cooling action synchronously adjusts the temperature setpoints of each microchannel reaction unit to the values ​​corresponding to the preset safe temperature curve. The safe temperature curve is set to 60% to 70% of the current normal operating temperature curve. The specific percentage is determined by linear interpolation based on the relative position of the abnormal scattering characteristic amplitude within the warning range. The closer the amplitude is to the critical boundary of 80%, the closer the cooling ratio is to the lower limit of 60%. The cooling rate is determined by the maximum dynamic response capability of the heat transfer medium flow regulating valve in the jacket of each microchannel reaction unit. The controller decomposes the step change in the temperature setpoint into multiple control cycle ramp commands according to the upper limit of the cooling rate to avoid thermal stress impact on the microchannel wall caused by a sudden drop in jacket temperature. The chelating agent enhancement action increases the speed setpoint of the chelating agent feed pump to 1.5 to 2 times the current normal value. The specific multiple is also determined by linear interpolation based on the position of the abnormal scattering characteristic amplitude within the warning range. The cooling and chelating agent enhancement actions are implemented simultaneously within this control cycle, without any sequential order, to ensure the timeliness of the intervention measures.

[0074] The synergistic intervention of cooling and chelating agents works based on the following mechanisms: lowering the temperature reduces the hydrolysis reaction rate constant through the Arrhenius relation, slowing down the formation rate of titanium oxide oligomers from the source; increasing the supply of chelating agents enhances the collision coordination probability between chelating ligands and titanium central ions, causing the already formed titanium oxide oligomers to be rapidly chelated and captured, thus escaping the polycondensation reaction chain; these two pathways simultaneously compress the net accumulation rate of titanium oxide oligomers from the formation and consumption ends, respectively, causing their concentration to fall below the critical concentration corresponding to the precipitation dissolution volume. The formed microcrystals, due to the reduced concentration of surrounding titanium oxide oligomers, lose their material supply for continued growth and gradually depolymerize and dissipate under Brownian motion. While maintaining continuous production line operation, the rapid and coordinated adjustment of process parameters curbs the worsening trend of precipitation, providing a window for the reaction system to self-repair and avoiding material losses and production interruptions caused by shutdown and slag discharge.

[0075] When the response level indicates a critical state, the controller, while maintaining all the contents of the first protective linkage operation command, generates a second protective linkage operation command. This command triggers the opening of the emergency dilution valves located on the inter-stage connecting pipe sections of each microchannel reaction unit, injecting anhydrous inert solvent into the main flow channel within the connecting pipe section. The emergency dilution valve is a normally closed pneumatic diaphragm valve, connected upstream to an anhydrous isopropanol storage tank that has undergone molecular sieve drying, with a nitrogen back pressure maintained at 0.3 MPa to 0.5 MPa. After the controller issues the opening command, the solenoid reversing valve drives the diaphragm valve to fully open within 0.2 seconds. Anhydrous isopropanol, driven by back pressure, is injected into the inter-stage connecting pipe section through a check valve. The injection flow rate is determined by the orifice diameter of the downstream throttling orifice plate and the back pressure, designed to be 0.5 to 1 times the current flow rate of the main flow channel. The injection pulse duration is 3 to 5 seconds, determined through residence time distribution experiments. This duration is sufficient to ensure that the injected inert solvent slug completely covers the entire reaction liquid volume of the interstage connecting pipe section and the inlet section of the next-stage microchannel reaction unit. After the anhydrous inert solvent is injected, the concentration of titanium oxide oligomers in the pipe section is instantaneously diluted. The product of its concentration and the hydroxyl concentration is forcibly reduced to below the precipitation product. The formed titanium dioxide crystal nuclei lose the support of the supersaturation driving force and stop growing. Some smaller crystal nuclei redissolve because their solubility increases with the decrease in concentration.

[0076] Preferably, the scattering alarm of the third characteristic band provides a buffer zone between normal control and emergency dilution within the warning range. The first protective linkage operation command operates at full capacity during this period, and most minor precipitation disturbances can be eliminated within this range. The dilution valve is only activated as a last line of defense under extreme conditions, reducing solvent consumption and interference with the residence time distribution of the main channel. The combination of cooling and chelating agents complements the inert solvent dilution in terms of mechanism of action. By changing the reaction kinetic parameters, it limits precipitation formation at the source, and by changing the local concentration field, it inhibits the growth of existing precipitation.

[0077] Furthermore, the multivariate collaborative solution updates the constraint boundary values ​​in real time according to the set of operating variables from the previous control cycle within each control cycle. When any variable among the feed rate, reaction temperature, and system pH is limited to its corresponding constraint boundary and the deviation between the reaction progress index and the target progress does not converge, the constraint relaxation logic is triggered. The constraint boundary with the lowest priority is extended to the second boundary value according to the preset constraint priority, and then the solution is recalculated.

[0078] In this embodiment, at the start of each control cycle, the multivariate collaborative solution performs a complete refresh calculation of the boundary values ​​of the two constraints based on the actual set of operational variables executed in the previous control cycle and the feedback of the current characteristic spectral signal. The boundary value update of the hydrolysis-chelation rate matching constraint uses the current reaction process index as an index to retrieve the corresponding upper and lower limits of the ratio from a pre-stored dynamic ratio interval table. This dynamic ratio interval table is generated statistically from calibration experimental data, dividing the entire process range into twenty process index intervals. For the calibration sample points within each interval, the percentile distribution of the ratio of hydrolysis rate to chelation rate is calculated, and the fifth percentile is taken as the lower limit of the ratio for that interval, and the ninety-fifth percentile as the upper limit of the ratio. The boundary values ​​of adjacent intervals are smoothed by moving average to eliminate segment jumps.

[0079] Specifically, the boundary value update for precipitation formation constraints uses the highest value among the current temperature settings of each microchannel reaction unit as the search key, and retrieves the corresponding upper limit of the solubility volume from the temperature-precipitation solubility volume lookup table. The temperature-precipitation solubility volume lookup table is pre-established through isothermal solubility measurement experiments. These experiments involve titrating a known concentration of titanium oxide oligomer solution with hydroxyl groups under isothermal conditions. The critical point of precipitation is detected by online scattering spectroscopy, and the critical concentration product corresponding to different temperature points is recorded. The temperature sampling interval is 5 degrees Celsius, and the solubility volume value at intermediate unmeasured temperature points is obtained by logarithmic linear interpolation between two adjacent measured points. The hydroxyl concentration is calculated jointly from the current water injection rate and reaction conversion rate, while the titanium oxide oligomer concentration is calculated jointly from the absorption intensity of the first characteristic band and the scattering intensity of the third characteristic band. The product of these two values ​​is compared with the upper limit of the solubility volume obtained from the lookup table, and the current boundary value of the precipitation formation constraint is set as the maximum allowable concentration product corresponding to this upper limit of the solubility volume.

[0080] Preferably, the constraint boundary value is always synchronized with the actual thermodynamic and kinetic state of the reaction system, and will not result in the constraint being too tight, limiting the degree of freedom of adjustment, or too loose, losing its protective effect due to the progress of the reaction or temperature changes. The dynamic ratio interval is established using the percentile statistical method, using the actual dispersion range of experimental data rather than empirical fixed values ​​as the boundary, so that the constraint matches the actual process dispersion and avoids a large number of false constraint alarms generated by the fixed boundary when the raw material batch fluctuates.

[0081] During the iterative search process of the rolling time-domain optimization solver, the controller checks after each iteration whether each operational variable has reached its respective constraint boundary, while simultaneously monitoring whether the cost function value is within the preset convergence tolerance range. The constraint boundaries of the operational variables fall into three categories: The first category is physical limit boundaries, determined by the stroke range of the actuator. For example, a feed pump has a minimum start / stop flow rate and a maximum rated flow rate; the jacket medium temperature has heating and cooling limits; and the pH adjuster injection pump has a minimum pulse width and a maximum full-on duty cycle. The second category is operational variable boundaries derived from hydrolysis-chelation rate matching constraints. During iteration, the solver transforms rate ratio constraints into feasible domain restrictions on the feed rate and temperature setpoints. The third category is operational variable boundaries derived from precipitation formation constraints. The solver transforms concentration product constraints into feasible domain restrictions on the water injection feed rate and the system pH value. When one or more operational variables are on their constraint boundaries for two consecutive iterations and the relative decrease in cost function value compared to the previous iteration is less than 0.5%, the controller determines that the operational variable has been confined to its corresponding constraint boundary and the deviation has not converged, thus triggering the constraint relaxation logic.

[0082] Specifically, the constraint priorities are preset to three levels in the controller. The precipitation formation constraint is at the highest protection level and cannot be relaxed; its boundary value remains unchanged regardless of whether the optimization converges. The upper limit constraint of the ratio between the hydrolysis and chelation rates is at the second priority, and the lower limit constraint is at the lowest priority, the third priority. The priority setting is based on the level of process safety: precipitation formation directly leads to irreversible deactivation of the catalyst's active components, which is a bottom-line risk that must be eliminated; excessively high hydrolysis rate leading chelation rate will lead to excessive accumulation of titanium oxide oligomers, increasing the exposure to precipitation risk, which is a factor that needs to be controlled but is not immediately disastrous; a chelation rate that is moderately lower than the hydrolysis rate only affects the sufficiency of chelation and does not immediately lead to precipitation, which is a factor that can be tolerated for short-term deviations.

[0083] When constraint relaxation is triggered, the controller expands the constraint boundaries in ascending order of priority. First, the lower bound constraint of the third priority is expanded downwards by 20% of the current boundary value; for example, if the current lower bound is 0.4, it expands to 0.32, forming the second boundary value and replacing the original boundary value in the solver's constraint set. The solver starts from the current iteration point and re-executes the complete sequential quadratic programming iterative search under the new constraints. If the deviation converges to within the acceptable range after re-solving, the current set of operand variables is output, and the solution for this control cycle ends. If the deviation still does not converge after re-solving, the controller expands the upper bound constraint of the second priority by 10% upwards; for example, if the current upper bound is 1.2, it expands to 1.32, and re-solves. If convergence still fails after two layers of relaxation, the controller maintains the relaxed constraint boundaries, outputs the currently found optimal feasible solution as the set of operand variables, and continues to use the relaxed boundaries as initial constraints in the solution of the next control cycle.

[0084] Preferably, the feasible region is autonomously expanded under extreme conditions through multivariate collaborative solution, avoiding deadlock caused by the simultaneous tightening of multiple constraints, which leads to the feasible region degenerating into an empty set and thus being unable to output any effective control commands. The priority-based relaxation design ensures that the system follows an order from light to heavy and from the edge to the core when sacrificing process optimality, prioritizing the relaxation of the lower limit requirement of chelation rate, followed by the relaxation of the upper limit restriction of hydrolysis rate, to meet the precipitation formation constraint. Before each control cycle recalculation, the constraint boundary is refreshed before triggering relaxation, so that relaxation occurs only when necessary. Once the operating conditions return to normal in the next cycle, the constraint boundary automatically returns to the original set value, without generating permanent constraint relaxation offset.

[0085] Furthermore, based on the set of operating variables or protective linkage operation commands, the feed rate, reaction temperature, and system pH of the microchannel reaction unit are synchronously controlled, including: S801. Use the temperature setpoint in the operation variable group as a feedforward signal to control the jacket medium temperature of the corresponding microchannel reaction unit to form a preset axial temperature gradient along the material flow direction. S802. Use the feed rate in the operation variable group as a feedforward signal to control the stroke frequency of the water injection feed pump and the rotation speed of the chelating agent feed pump. S803. Use the system pH value in the operation variable group as a feedback signal to control the pulse width modulation duty cycle of the acid-base regulator injection pump.

[0086] In this embodiment, the controller uses the temperature setpoints of each stage of the microchannel reaction unit in the operational variable group as feedforward signals, and establishes a preset axial temperature gradient along the material flow direction by adjusting the flow rate of the heat transfer medium in the jacket. The jacket of each stage of the microchannel reaction unit is divided into several independent temperature control sections along the material flow direction. Each temperature control section is equipped with a heat transfer medium inlet valve, an outlet valve, and a temperature transmitter. After receiving the temperature setpoints corresponding to each temperature control section of the microchannel reaction unit in the operational variable group, the controller compares them with the measured heat transfer medium return temperature at the outlet of the corresponding temperature control section. The deviation is then calculated using proportional-integral operations to generate the opening command of the flow control valve.

[0087] Specifically, the valve opening determines the flow rate of the heat transfer medium entering the jacket cavity of the temperature control section. As the heat transfer medium flows through the jacket cavity, it exchanges heat with the reaction liquid through the microchannel wall, ultimately causing the temperature of the reaction liquid within that microchannel section to approach the set value. The set values ​​of each temperature control section change sequentially along the material flow direction, forming a continuous temperature slope inside the microchannel. The specific shape of the temperature slope is determined by the requirements of the reaction process: in the first-stage microchannel reaction unit dominated by hydrolysis, a lower temperature is set at the inlet section to match the requirement of low-temperature pre-hydrolysis to suppress the hydrolysis rate, and the temperature gradually increases along the flow direction to connect to the subsequent chelation section; in the second stage dominated by chelation, the temperature is maintained in the medium temperature range to balance the chelation rate and precipitation suppression; in the maturation section, the temperature is kept constant at a moderate maturation temperature to promote the structural rearrangement and stabilization of titanium oxide oligomers. When the controller issues a temperature setpoint change command, it uses the maximum allowable heating and cooling rate of the jacketed heat transfer medium system as a constraint to break down the step setpoint change into ramp commands within multiple control cycles, thus avoiding thermal shock exceeding the allowable stress caused by sudden temperature changes of the heat transfer medium within the microchannel metal wall.

[0088] Preferably, the temperature setpoint is directly taken from the result of multivariate collaborative calculation, avoiding the problem that the temperature in each segment cannot change in a coordinated manner along the flow direction due to the lag of local feedback correction in independent temperature loops; the establishment of the axial temperature gradient transforms the heating curve that originally needed to be achieved through time programming in a single batch reactor into a steady-state temperature distribution unfolded along the spatial coordinates of the microchannel, so that the reaction liquid experiences the designed temperature sequence when flowing through different axial positions, eliminating the influence of temperature differences of different fluid particles within a batch on the uniformity of catalyst activity; decomposing the temperature setpoint change into ramp commands avoids thermal fatigue damage to the equipment and extends the service life of the jacket and microchannel structural components.

[0089] The controller uses the setpoints for the water injection rate and the chelating agent feed rate in the operational variable group as feedforward signals. These are converted into stroke frequency commands for the water injection pump and speed commands for the chelating agent feed pump, respectively, and then sent to the corresponding pump drivers. The water injection pump is a plunger metering pump, whose reciprocating motion is driven by a variable frequency motor via a reduction gear. The volume of liquid discharged per stroke is determined by the plunger diameter and stroke length, and is recorded in the controller during factory calibration. The controller divides the water injection rate setpoint by the volume discharged per stroke, and then by the number of pump heads connected in parallel, to obtain the required number of strokes per unit time, i.e., the stroke frequency, which is directly written into the frequency command register of the variable frequency motor. The chelating agent feed pump is a gear metering pump, whose rotational motion is directly driven by a servo motor. The volume of liquid discharged per revolution is determined by the gear module and tooth width, and is recorded during factory calibration. The controller divides the chelating agent feed rate setpoint by the volume discharged per revolution to obtain the required number of revolutions per unit time, i.e., the speed, which is written into the speed command register of the servo motor. Commands to both pumps are issued synchronously within the current control cycle. The pump driver begins to respond in the next servo cycle after receiving the command, aligning the changes in the two feed flow rates in time.

[0090] Preferably, the feed rate is adjusted in real time based on spectral feedback and multivariate calculation results, ensuring that the supply rate of titanium source and water always matches the actual consumption rate of the reaction system, thus avoiding reactant accumulation or shortage when the reaction kinetics fluctuate due to a fixed feed curve. The stroke frequency control of the plunger metering pump and the speed control of the gear metering pump do not rely on the closed-loop feedback of the flow meter, eliminating the constraint of flow meter measurement lag on the dynamic response speed. The synchronous issuance of the two feed commands ensures that the water for hydrolysis and the chelating ligand enter the reactor at the same time with a new ratio, maintaining the instantaneous matching relationship between the hydrolysis rate and the chelation rate at the feed inlet.

[0091] The controller uses the system pH setpoint from the controlled variable group as the target value to form a pH feedback control loop. It outputs a pulse-width modulation (PWM) duty cycle command to control the acid-base regulator injection pump. The pH feedback control loop uses the setpoint as a reference input and the measured signal from the pH microelectrode located downstream of the regulator injection point and before the inlet of the microchannel reaction unit as the feedback quantity. In each control cycle, the controller reads the output voltage of the pH microelectrode, converts it to the actual pH value after temperature compensation, and subtracts it from the setpoint to obtain the pH deviation. This deviation is fed into the proportional-integral (PI) controller. The proportional gain and integral time constant are tuned based on the liquid residence time in the pipe section between the regulator injection point and the pH microelectrode installation position, and the kinetic time constant of the acid-base neutralization reaction. The output of the PI controller is a pulse-width modulation duty cycle, ranging from 5% to 95%. When the duty cycle is 5%, the injection pump is activated for only a very short time in each modulation cycle, resulting in the minimum injection rate per unit time. When the duty cycle is 95%, it is activated almost continuously, with the injection rate per unit time approaching the pump's rated flow rate. The carrier frequency of the pulse width modulation is selected near the upper limit of the allowable switching frequency of the injection pump solenoid valve to minimize the injection volume of a single pulse while ensuring reliable valve core operation, thereby improving the resolution of pH adjustment. The controller writes the duty cycle instruction into the pulse width register of the injection pump driver, which then controls the on / off timing of the solenoid valve to achieve fine adjustment of the amount of regulator injected per unit time.

[0092] Preferably, the output of the regulator from the injection pump is controlled by pulse width modulation (PWM), overcoming the problem of extremely small flow rates of acid-base regulators and the difficulty in accurately controlling them with ordinary flow control valves. Linear regulation of minute flow rates is achieved by utilizing the time ratio of rapid opening and closing of the solenoid valve. The feedback loop uses a pH microelectrode adjacent to the inlet of the microchannel reaction unit as the detection point. The measured pH value already includes the actual acidity or alkalinity after the regulator is mixed with the main fluid, eliminating the interference of regulator delivery pipeline delays and uneven mixing on pH control accuracy. In parallel with feedforward control of the feed rate and temperature, pH feedback control completes the control method combination of the manipulated variable execution layer, enabling the manipulated variable to have a dynamic response speed matching the calculation cycle.

[0093] Furthermore, after controlling the pulse width modulation duty cycle of the acid-base regulator injection pump, the following is also included: S901. Real-time acquisition of pH microelectrode detection signal located downstream of the regulator injection point and at the inlet of the corresponding microchannel reaction unit; S902. Compare the pH microelectrode detection signal with the system pH value in the manipulated variable group to obtain the injection deviation; S903. Based on the injection deviation, the pulse width modulation duty cycle is corrected so that the actual pH value at the injection point reaches the system pH value set by the operating variable group within the preset relaxation time.

[0094] In this embodiment, after the controller writes a pulse width modulation duty cycle command to the acid-base regulator injection pump driver, it synchronously reads the pH microelectrode detection signal located downstream of the regulator injection point and at the inlet of the microchannel reaction unit. The installation position of the pH microelectrode is such that the length of the flow channel along the pipe section axis from the regulator injection point is ten to fifteen times the inner diameter of the pipe section. This distance has been confirmed by computational fluid dynamics simulation to be sufficient to allow the injected regulator droplets to complete turbulent mixing with the main fluid before reaching the surface of the electrode sensitive membrane. The pH microelectrode adopts an integrated encapsulation structure of antimony-antimony oxide sensitive membrane and silver-silver chloride reference electrode, and its response time constant does not exceed 0.3 seconds at 25 degrees Celsius, meeting the timing requirements of the control cycle.

[0095] The potential signal output by the microelectrode is transmitted to the analog input module of the controller via a coaxial shielded cable. Within the module, it is first differentially amplified by an instrumentation amplifier with a 10¹² ohm input impedance to suppress common-mode interference. Then, it is filtered by a fourth-order Bessel low-pass filter to remove high-frequency noise introduced by the solenoid valve switching. The cutoff frequency is set to five times the reciprocal of the control cycle to retain dynamic pH change information. The filtered signal is quantized into a digital value by a 16-bit analog-to-digital converter. The controller reads this digital value and converts it into the actual pH value based on the electrode calibration data stored in non-volatile memory. The electrode calibration data is updated periodically through an online calibration mechanism. Every two hundred control cycles or at the start of each production batch, the controller automatically triggers a three-point calibration program, rinsing the electrode surface with standard buffer solutions at pH values ​​of 4.00, 6.86, and 9.18 and recording the steady-state potential. A linear regression is used to fit the slope and intercept of the potential-pH conversion, refreshing the calibration parameters.

[0096] Preferably, the pH detection point is set at the inlet of the microchannel reaction unit, so that the feedback signal reflects the true pH value of the liquid that has been fully mixed and is about to enter the reaction zone; the combination of high input impedance differential amplification and Bessel filter effectively suppresses electromagnetic interference and valve switching noise common in industrial environments while maintaining fast signal response capability, and avoids frequent fluctuations in duty cycle commands due to signal glitches; periodic automatic three-point calibration eliminates long-term drift caused by changes in the oxide layer state on the antimony electrode surface, so that the absolute accuracy of pH detection is maintained within the specified tolerance range throughout the entire production batch.

[0097] The controller subtracts the actual pH value obtained from the pH microelectrode detection signal (after calibration) from the system pH setpoint in the manipulated variable group, digit by digit, to obtain the injection deviation. The system pH setpoint in the manipulated variable group is generated by the multivariate collaborative solver in the current control cycle or a previous protection linkage solver mode, representing the target acidity / alkalinity of the inlet liquid of the microchannel reaction unit required to meet the process constraints of the entire process section. The sign of the deviation directly indicates the adjustment direction: a positive deviation indicates that the actual pH is higher than the setpoint, requiring an increase in the amount of acidic regulator injected or a decrease in the amount of alkaline regulator injected; a negative deviation indicates that the actual pH is lower than the setpoint, requiring a decrease in the amount of acidic regulator injected or an increase in the amount of alkaline regulator injected. The controller sends the deviation to the secondary closed-loop correction controller, which forms the inner loop of a cascade structure in the original pH feedback control loop. The outer loop is the main pH feedback control loop, and the inner loop is specifically designed to compensate for the mixing and transport dynamics within the pipe section between the injection point and the microelectrode detection point.

[0098] Specifically, the secondary closed-loop correction controller employs a proportional-integral (PI) algorithm. The proportional gain and integral time constant are individually tuned based on the characteristics of the lag phase, from the opening of the regulator injection valve to the detection of a change by the pH microelectrode. The lag time is measured during system debugging. The controller sends a 100-millisecond test pulse to the injection pump, and the time difference from the pulse start to the point where a discernible change in the pH microelectrode signal is detected is recorded. This process is repeated five times, and the average is taken as the lag time. The integral time constant is set to three times the lag time to prevent excessive accumulation of integral action within the lag interval, which could lead to overshoot. The proportional gain is gradually increased from a conservative value. After each increase, a small step change in the setpoint is applied, and the overshoot of the response curve is observed. The increase is stopped when the overshoot reaches 10%, and 80% of this gain is taken as the final proportional gain.

[0099] Within each control cycle, the proportional term is incremented by multiplying the current injection deviation by the proportional gain, while the integral term is incremented by multiplying the current injection deviation by the control cycle duration and dividing by the integral time constant, then accumulated in the integral register. The sum of these two terms serves as the correction increment for the pulse width modulation duty cycle. The corrected duty cycle value is limited to 5% to 95% of the allowable range of the pulse width modulation driver to prevent the solenoid valve from overheating due to lack of intermittent cooling when fully closed or continuously fully open.

[0100] Preferably, the actual mixed pH at the injection point outlet, rather than the regulator flow rate at the injection pump outlet, is used as feedback to adjust the duty cycle in a closed loop, compensating for the effects of regulator concentration fluctuations, changes in the main fluid pH buffer capacity, and the nonlinearity of the injection valve flow characteristics. The proportional-integral algorithm in the lag system differentiates the action periods of the proportional and integral terms to process deviations in frequency bands. The proportional term responds quickly to the current deviation to provide timely correction, while the integral term continues to act after the lag time to eliminate steady-state residuals, ensuring that the actual pH value approaches and stabilizes at the set value within a preset relaxation time. The cascade structure decouples the dynamic response speed and accuracy of pH control to two loops: the outer loop senses changes in pH demand across the entire process and outputs the set value, while the inner loop quickly overcomes local disturbances at the injection end. This ensures that the pH condition at the inlet of the microchannel reaction unit remains consistent with the calculated expected value despite the varying update frequency of the operating variable group, guaranteeing that the pH model upon which the multi-variable collaborative solution relies holds true in actual operation.

[0101] like Figure 2 As shown, a catalyst preparation process control system is used to implement a catalyst preparation process control method, comprising: The data acquisition module acquires the characteristic spectral signals collected in real time by each spectral detection unit, wherein the characteristic spectral signals are used to reflect the chemical state of titanium species; The reaction analysis module determines the reaction progress index of the current reaction system based on the comparison results between the characteristic spectral signal and the preset multi-stage reference spectral features. The operation analysis module uses the characteristic spectral signal as the feedback benchmark. Under the conditions of meeting the hydrolysis and chelation rate matching constraints and precipitation formation constraints, it performs multi-variable collaborative calculation on the deviation between the reaction process index and the target process to generate a set of operation variables. When abnormal scattering characteristics appear in the characteristic spectral signal, it switches to the protection linkage calculation mode and generates protection linkage operation instructions. The preparation process control module synchronously regulates the feed rate, reaction temperature, and system pH of the microchannel reaction unit based on the set of operating variables or protective linkage operation commands.

[0102] The above description is merely a preferred embodiment of this application. The scope of protection of this application is not limited to the above embodiments. All technical solutions falling within the scope of this application's concept are within the scope of protection of this application. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of this application should also be considered within the scope of protection of this application.

Claims

1. A method for controlling a catalyst preparation process, applied in a control system comprising at least three tandem microchannel reaction units and multiple spectral detection units, wherein the multiple spectral detection units are respectively disposed between stages and at the final stage outlet of each microchannel reaction unit, characterized in that, include: The characteristic spectral signals acquired in real time by each spectral detection unit are obtained, whereby the characteristic spectral signals are used to reflect the chemical state of the titanium species; Based on the comparison results between the characteristic spectral signal and the preset multi-stage reference spectral features, the reaction process index of the current reaction system is determined. Using the characteristic spectral signal as the feedback benchmark, and under the conditions of meeting the hydrolysis and chelation rate matching constraints and precipitation formation constraints, the deviation between the reaction process index and the target process is calculated using a multivariate collaborative solution to generate a set of operational variables; among them, when abnormal scattering characteristics appear in the characteristic spectral signal, the system switches to the protection linkage solution mode to generate protection linkage operation instructions. The feed rate, reaction temperature, and system pH of the microchannel reaction unit are synchronously controlled according to the set of operating variables or protective linkage operation commands.

2. The method for controlling the catalyst preparation process according to claim 1, characterized in that, The process of acquiring the characteristic spectral signal collected in real time by each spectral detection unit includes: performing differential processing on the original spectral signal collected by each spectral detection unit and the corresponding dark spectral reference signal, and performing baseline correction on the differentially processed signal to obtain the characteristic spectral signal.

3. The method for controlling the catalyst preparation process according to claim 1, characterized in that, The reaction progress index of the current reaction system is determined based on the comparison results between the characteristic spectral signal and the preset multi-stage reference spectral features, including: The absorption intensity ratio of at least two characteristic bands is extracted from the characteristic spectral signal, wherein the first characteristic band is sensitive to the bonding state of titanium-oxygen oligomers and the second characteristic band is sensitive to the charge transfer transition state from ligand to metal. The absorption intensity ratio is matched with the reference ratio range of the corresponding band in the preset multi-stage reference spectrum characteristics to determine the reaction process index of the current reaction system.

4. The method for controlling the catalyst preparation process according to claim 3, characterized in that, The determination of the reaction process index of the current reaction system also includes: Obtain the rate of change of scattering intensity of the characteristic spectral signal at the third characteristic band, which reflects the precipitation scattering effect; When the rate of change of scattering intensity is greater than the preset critical slope threshold, the corresponding absorption intensity ratio change trajectory is retrieved, and the reaction process index is compensated and corrected based on the change trajectory and the rate of change of scattering intensity.

5. The method for controlling the catalyst preparation process according to claim 1, characterized in that, The multivariate collaborative solution of the deviation between the reaction process index and the target process, generating a set of operational variables, includes: Based on the deviation between the reaction progress index and the target progress, under the conditions that the hydrolysis rate and chelation rate of each microchannel reaction unit are limited to a preset dynamic ratio range by the hydrolysis and chelation rate matching constraint, and the product of the concentration of titanium oxide oligomer and the concentration of hydroxyl group in the system is limited to below the precipitation dissolution volume based on the precipitation formation constraint, the cost function of feed rate, reaction temperature and system pH is optimized in the rolling time domain to generate a sequence of operating variables that makes the deviation converge in the prediction time domain, and the set of operating variables for the current control cycle is extracted from the sequence of operating variables.

6. The method for controlling the catalyst preparation process according to claim 5, characterized in that, When abnormal scattering features appear in the characteristic spectral signal, the system switches to the protection linkage calculation mode and generates protection linkage operation instructions, including: The response level of the protection linkage is determined based on the preset range of the amplitude of the abnormal scattering characteristics. When the response level indicates a warning state, the first protective linkage operation command generated includes simultaneously reducing the temperature setpoint of each microchannel reaction unit to the value corresponding to the preset safety temperature curve, and simultaneously increasing the feed rate of the chelating agent. When the response level indicates a critical state, the generated second protective linkage operation command, based on the first protective linkage operation command, triggers the opening of the emergency dilution valve located at the interstage connection of each microchannel reaction unit to inject anhydrous inert solvent into the main channel.

7. The method for controlling the catalyst preparation process according to claim 6, characterized in that, The multivariate collaborative solution includes: in each control cycle, updating each constraint boundary value in real time according to the set of operating variables of the previous control cycle; when any variable among the feed rate, reaction temperature and system pH is limited to its corresponding constraint boundary and the deviation between the reaction process index and the target process does not converge, the constraint relaxation logic is triggered, and the constraint boundary with the lowest priority is extended to the second boundary value according to the preset constraint priority before resolving.

8. The method for controlling the catalyst preparation process according to claim 1, characterized in that, The synchronous control of the feed rate, reaction temperature, and system pH of the microchannel reaction unit based on the set of operating variables or protective linkage operation commands includes: The temperature setpoint in the group of operating variables is used as a feedforward signal to control the jacket medium temperature of the corresponding microchannel reaction unit to form a preset axial temperature gradient along the material flow direction. The feed rate in the set of operational variables is used as a feedforward signal to control the stroke frequency of the water injection feed pump and the rotation speed of the chelating agent feed pump. The system pH value in the manipulated variable group is used as a feedback signal to control the pulse width modulation duty cycle of the acid-base regulator injection pump.

9. The method for controlling the catalyst preparation process according to claim 8, characterized in that, After controlling the pulse width modulation duty cycle of the acid-base regulator injection pump, the method further includes: The pH microelectrode detection signal is acquired in real time at the inlet of the corresponding microchannel reaction unit, which is located downstream of the regulator injection point. The injection deviation is obtained by comparing the pH microelectrode detection signal with the system pH value in the manipulated variable group. The pulse width modulation duty cycle is corrected based on the injection deviation, so that the actual pH value at the injection point reaches the system pH value set by the operating variable group within the preset relaxation time.

10. A catalyst preparation process control system, characterized in that, A method for controlling a catalyst preparation process as described in any one of claims 1 to 9, comprising: The data acquisition module acquires the characteristic spectral signals collected in real time by each spectral detection unit, wherein the characteristic spectral signals are used to reflect the chemical state of titanium species; The reaction analysis module determines the reaction progress index of the current reaction system based on the comparison results between the characteristic spectral signal and the preset multi-stage reference spectral features. The operation analysis module uses the characteristic spectral signal as the feedback benchmark. Under the conditions of meeting the hydrolysis and chelation rate matching constraints and precipitation formation constraints, it performs multi-variable collaborative calculation on the deviation between the reaction process index and the target process to generate a set of operation variables. When abnormal scattering characteristics appear in the characteristic spectral signal, it switches to the protection linkage calculation mode and generates protection linkage operation instructions. The preparation process control module synchronously regulates the feed rate, reaction temperature, and system pH of the microchannel reaction unit based on the set of operating variables or protective linkage operation commands.

Images