Feeding system and feeding method for intermediate synthesis

Abstract

The invention relates to the field of feeding, and discloses a feeding system and a feeding method for intermediate synthesis, and the feeding method comprises the following steps: based on monitoring and analyzing the distribution condition of a catalyst bed layer in an alkylation process, establishing an inactivation model in combination with pressure fluctuation to dynamically compensate the activity of a catalyst, and adjusting the dispersity and the feeding speed of the catalyst through the inactivation model; based on the end of alkylation, obtaining the oxidant concentration in the oxidation process, establishing a target function with the reaction time based on the oxidant change rate, calculating the reaction time deviation degree through the target function, and intelligently adjusting the oxidant supply frequency in combination with analysis; feeding parameters of a plurality of historical reaction periods are obtained, deep collaborative optimization is carried out on the reaction time and the feeding parameters by establishing a three-layer intelligent regulation and control model, and a time scale correlation inactivation model and a time scale coupling free radical prediction model are constructed to solve the optimal adjustment amount; and carrying out real-time self-adaptive regulation and control on feeding parameters, and integrating prediction maintenance to carry out feedback iteration.

Description

Technical Field

[0001] The present invention relates to the technical field of feeding, and particularly relates to a feeding system and a feeding method for intermediate synthesis. Background Art

[0002] Traditional feeding systems mostly rely on mechanical pumps and metering tanks. In traditional intermediate alkylation reactions, the monitoring of catalyst activity mainly relies on single-point thermocouple temperature measurement and timed sampling analysis; In the prior art, the differences in axial and radial temperature distributions cannot be captured, making it difficult to detect in a timely manner when the catalyst is locally overheated. The problem of chemical deactivation is often only noticed after the conversion rate of the reaction product decreases, showing a lag; there is a lack of a pressure-temperature multi-parameter coupling model, and the feeding speed is only passively reduced according to the temperature exceeding the threshold, without considering the mechanical deactivation (bed plugging) indicated by pressure fluctuations, resulting in the inability to actively adjust the vibration frequency or backflush when the catalyst dispersion deteriorates; relying on offline detection of the oxidant concentration after manual sampling, with a long sampling period, unable to capture the concentration change rate in real time, and a large deviation in the reaction time, resulting in excessive addition of the oxidant; the feeding frequency is fixed, without introducing a lag compensation algorithm, ignoring the delay of the oxidant from the feeding port to the reaction zone, which is likely to cause "peroxidation" side reactions in the later stage of the reaction; there is no time scale alignment mechanism for different reaction cycles, and the time deviation of the oxidant decaying to the lowest deactivation standard between batches is large, unable to predict the current deactivation trend through historical similar working conditions, and the stability of the catalyst life fluctuation range is poor; not associating the free radical kinetic equation with the feeding parameters, with a large deviation between the measured value and the theoretical value of the free radical concentration, and uneven consumption of the initiator resulting in an extended reaction induction period; the monitoring dimension is single, unable to achieve the fusion of multiple physical quantities, the model is static, lacking dynamic adaptability, the feeding control is extensive, lacking spatio-temporal scale coordination, and the data-driven ability is weak, unable to achieve full-cycle optimization; Therefore, there is a need to provide a feeding system and a feeding method for intermediate synthesis. Summary of the Invention

[0003] The purpose of the present invention is to provide a feeding system and a feeding method for intermediate synthesis. To solve the above-mentioned problems in the prior art, the present invention is achieved through the following technical solutions: In a first aspect, a feeding method for intermediate synthesis provided by an embodiment of the present invention specifically includes the following steps: Step 1: Based on monitoring and analyzing the distribution of the catalyst bed during the alkylation process, a deactivation model is established in combination with pressure fluctuations to dynamically compensate the catalyst activity, and the catalyst dispersion and feeding speed are adjusted through the deactivation model; Step 2: Based on the end of alkylation, obtain the oxidant concentration during the oxidation process, calculate the oxidant change rate to establish an objective function with the reaction time, calculate the reaction time deviation through the objective function, and intelligently adjust the oxidant feeding frequency in combination with the analysis; Step 3: Obtain the feeding parameters of multiple historical reaction cycles. By establishing a three-layer intelligent regulation model, deeply coordinate and optimize the reaction time and feeding parameters, construct a time-scale associated inactivation model and a time-scale coupled free radical prediction model to solve the optimal adjustment amount; Step 4: Based on the optimal adjustment amount, perform real-time adaptive regulation on the feeding parameters, and integrate predictive maintenance for feedback iteration.

[0004] Second, a feeding system for intermediate synthesis provided by an embodiment of the present invention specifically includes the following modules: Alkylation adjustment module: Based on monitoring and analyzing the distribution of the catalyst bed during the alkylation process, establish an inactivation model in combination with pressure fluctuations to dynamically compensate the catalyst activity, and adjust the catalyst dispersion and feeding speed through the inactivation model; Oxidation adjustment module: Based on the end of alkylation, obtain the oxidant concentration during the oxidation process, calculate the oxidant change rate to establish an objective function with the reaction time, calculate the reaction time deviation degree through the objective function, and intelligently adjust the oxidant feeding frequency in combination with the analysis; Adjustment and update module: Obtain the feeding parameters of multiple historical reaction cycles. By establishing a three-layer intelligent regulation model, deeply coordinate and optimize the reaction time and feeding parameters, construct a time-scale associated inactivation model and a time-scale coupled free radical prediction model to solve the optimal adjustment amount; Adaptive module: Based on the optimal adjustment amount, perform real-time adaptive regulation on the feeding parameters, and integrate predictive maintenance for feedback iteration.

[0005] Advantages of the present invention: Construct a three-dimensional temperature monitoring network for the catalyst bed by fusing infrared thermal imaging and thermocouples. Establish a multi-parameter inactivation model in combination with pressure fluctuations and the BP neural network to dynamically compensate the catalyst activity and adjust the feeding speed. At the same time, use ultraviolet spectroscopy and a dynamic objective function to realize real-time tracking of the oxidant concentration change rate. Through a three-layer intelligent regulation model: a time-scale associated inactivation model, a free radical prediction model, and a multi-objective optimization solver, deeply coordinate and optimize the feeding parameters of the reaction cycle, and finally form a closed loop of monitoring-modeling-regulation-iteration, realizing reducing the prediction error of catalyst activity, improving the utilization rate of the oxidant, and extending the catalyst life, providing an intelligent feeding solution with high precision and low energy consumption for intermediate synthesis. Description of the Drawings

[0006] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0007] Figure 1It is a flowchart of the steps of a feeding method for intermediate synthesis provided in Embodiment 1 of the present invention; Figure 2 It is a schematic structural diagram of a feeding system for intermediate synthesis provided in Embodiment 2 of the present invention. Detailed implementation manners

[0008] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0009] Embodiment 1

[0010] As Figure 1 shown, a feeding method for intermediate synthesis provided in an embodiment of the present invention specifically includes the following steps: Step 1: Based on monitoring and analyzing the distribution of the catalyst bed during the alkylation process, a deactivation model is established in combination with pressure fluctuations to dynamically compensate the catalyst activity, and the catalyst dispersion and feeding speed are adjusted through the deactivation model; In a specific embodiment, during the alkylation reaction for synthesizing the intermediate, the temperature distribution of the catalyst bed is monitored by an infrared thermal imager; Specifically, quartz glass windows are opened at equal intervals along the axial direction of the bed layer, and a nitrogen purge ring is installed on the inner wall to prevent the reactants from condensing and contaminating the windows; An inclined window is arranged at the head position, and the preset inclination angle of the inclined angle is 30°, to monitor the coking area at the top of the catalyst bed and the blockage condition of the distribution plate at the bottom; Monitoring is carried out by an infrared thermal imager that covers the normal working temperature of the catalyst and has a safety margin of more than 20%; The distance between the infrared thermal imager and the quartz glass window is kept at 2 m to ensure that the field of view covers the entire cross-section of the bed layer. The installation angle is calculated by the triangulation method, and the imaging edge coincides with the edge of the bed layer; Specifically, with the center of the cross-section of the bed layer as the reference, points A and B are marked on the edge of the window. The window is photographed by the infrared thermal imager, and the pitching angle and horizontal deflection angle of the device are adjusted so that points A and B in the imaging coincide with the actual edges of the bed layer. The pitching angle and horizontal deflection angle of the device are calculated using trigonometric functions to ensure that the optical axis of the infrared thermal imager is perpendicular to the center of the cross-section of the bed layer; For large reactors with a diameter > 1.5 m, 2 infrared thermal imagers are symmetrically installed, fixed with a rigid bracket, equipped with a three-dimensional adjustable pan-tilt head, for fine adjustment of the focal length and angle. The preset included angle is 90°, and the panoramic thermal image is stitched by software; Embed K-type thermocouples in the catalyst bed, with the depths preset to 20%, 50%, and 80% of the bed height. Synchronously collect the thermocouple data and thermal imaging data to establish a temperature-pixel gray-scale correction model; Adopt a BP neural network model. The input layer is the average pixel gray scale, the output layer is the temperature value, and the number of hidden layers and neurons is adjusted according to the data characteristics and the model training effect; Train the neural network with the training data set, use the gradient descent method to optimize the network weights, and reduce the prediction error of the model; Based on analyzing the relationship between temperature and pixel gray-scale value in the temperature-pixel gray-scale correction model of temperature measurement points at different depths, analyze whether there are changes in the parameters of the temperature-pixel gray-scale correction model caused by depth differences; If so, establish temperature-pixel gray-scale correction models for different depths; if not, establish a unified correction model; By obtaining the real-time temperature at different positions of the bed, obtain the temperature state of the catalyst during the reaction process, and analyze the influence of temperature on the catalyst activity; It should be noted that the change in temperature directly affects the activity and reaction rate of the catalyst. Different temperature regions result in different activity performances of the catalyst. Based on monitoring the temperature distribution and analyzing, establish an accurate catalyst activity model; Divide the thermal image into 3 axial regions, including the top preheating region, the middle reaction region, and the bottom cooling region; Set temperature thresholds for each axial region to identify abnormal reaction temperature characteristics; Exemplarily, the normal reaction temperature of the middle reaction region is set to [120°C, 130°C], the warning value temperature is preset to 135°C, and the interlock value temperature is preset to 140°C; It should be noted that the warning value temperature is used as an early warning signal for abnormal reaction temperature. When the temperature in the middle reaction region is monitored to reach 135°C, the system immediately issues an alarm. If the reaction temperature is close to the warning value, it indicates that the reaction system has a tendency to deviate from the normal operating state and there are potential risks. The potential risks include but are not limited to: changes in catalyst activity, fluctuations in reactant concentration, and poor heat dissipation; Within the normal operating temperature range of the catalyst, set catalytic temperature points, including the minimum operating temperature, the rated operating temperature, and the maximum operating temperature. The maximum operating temperature is the temperature after considering a 20% safety margin. Conduct operating condition tests at low load, rated load, and high load respectively under each catalytic temperature point to obtain temperature data and thermal imaging data under different operating conditions; It should be noted that the temperature data includes: the temperature value of each temperature measurement point, the acquisition time, and the corresponding operating conditions parameters. The operating conditions parameters include: the temperature set value, the load, and the reaction time; the thermal imaging data includes: the thermal imaging image file, the acquisition time, the corresponding operating conditions parameters, as well as the focal length, angle, and gain of the thermal imager; considering the characteristics of the catalyst at different usage stages, typical operating conditions are selected for data acquisition at each stage to reflect the impact of catalyst aging on the temperature distribution; Under stable operating conditions, the thermocouple data and thermal imaging data are collected synchronously, and the acquisition frequency is set to 1 time per minute to ensure the continuity and synchronization of the data; during the process of operating condition changes, such as heating up, cooling down, and load adjustment, the acquisition frequency is increased to 1 time per second to capture the dynamic process of temperature changes; Through the BP neural network, the input layer is the average pixel gray value, the output layer is the temperature value, and the number of hidden layers and neurons is adjusted according to the data characteristics and the training effect of the model; the neural network is trained using the training data set, and the gradient descent method is used to optimize the network weights; Specifically, the mean square error of calculating the average pixel gray value is used as the loss function, the initial learning rate is 0.001, and the learning rate is optimized through the Adam optimizer; Based on the established temperature-pixel gray correction model integrated into the monitoring system for real-time correction of thermal imaging data; through the bed layer thermal imaging image collected by the thermal imager, the system automatically extracts the gray values of each pixel point, calculates the corresponding temperature value according to the correction model, generates a temperature distribution cloud map, and displays the temperature of each area of the bed layer in real time; The reaction system pressure data is collected in real time through the pressure sensor, including the bed layer inlet / outlet pressure and the pressure difference, and is synchronously input into the system; The real-time pressure data is combined with the corrected temperature data as the input of the catalyst activity calculation model; In the initial model, the pressure fluctuation is used to compensate for the limitations of the temperature model; Exemplarily, a normal temperature in the middle reaction zone and a sudden increase in pressure indicate a local blockage; It should be noted that the foundation of multi-parameter monitoring is established, but the dynamic compensation part of the deactivation model has not been completed. The pressure data provided for subsequent writing can quantify the mechanical deactivation of the catalyst, while the temperature focuses on chemical deactivation, realizing high-precision three-dimensional temperature monitoring and data-driven modeling of the catalyst bed layer. The logical chain is: installing monitoring equipment, collecting multi-source data, constructing a correction model, analyzing and warning, and integrating pressure data; Using the integrated learning method, the input layer is: the corrected temperature data, the pressure fluctuation data, and the supplementary operating conditions parameters; the corrected temperature data includes: the regional average value and the gradient change; the pressure fluctuation data includes: the pressure difference and the fluctuation frequency; the supplementary operating conditions parameters include: the historical record of the feeding speed; The output layer is the catalyst activity coefficient, which is scaled from 0 to 1, with 1 indicating full activity; Proportionally adjust the catalyst dispersion and feed rate based on the difference between the catalyst activity coefficient and the standard activity coefficient; The model updates every 5 minutes by fine-tuning the weights with new data through an online learning mechanism to adapt to the dynamics of catalyst aging; Specifically, when the model detects that the activity coefficient drops to a preset activity threshold, dynamic compensation is triggered: If the temperature in the middle reaction zone > 135°C dominates deactivation, the feed rate is preferentially adjusted; If the pressure fluctuation differential pressure > 10% dominates deactivation, indicating blockage, the catalyst dispersion is preferentially adjusted and the bed vibrator is started to improve the distribution; Specifically, the catalyst dispersion is through an adjustable distribution plate built into the bed. When the activity coefficient drops and the temperature distribution is uneven, with the temperature difference in the top preheating zone > 15°C, the vibration frequency is increased to improve the catalyst uniformity; Combined with the pressure data, if the differential pressure suddenly increases, reverse purging is started to prevent blockage; The initial setting is corrected with real-time data through the reaction kinetics model: when the predicted decay rate of activity > 5% / min, the feed rate is gradually reduced by 0.5% each time, and vice versa; [[ID=?]] Step 2: Based on the end of alkylation, obtain the oxidant concentration during the oxidation process, calculate the oxidant change rate to establish an objective function with the reaction time, calculate the reaction time deviation degree through the objective function, and intelligently adjust the oxidant feed frequency in combination with the analysis; Install 3 groups of ultraviolet-visible spectroscopy sensors symmetrically at the top, middle, and bottom of the oxidation reactor, With an in-built flow-through cell sampling probe, set the sampling flow rate to control the sampling speed and avoid the influence of dead volume; Use oxidant solutions with standard concentrations, with concentration gradients of 0.1mol / L, 0.5mol / L, 1.0mol / L, and 2.0mol / L, for spectral calibration to establish a Lambert-Beer law calibration model: ; The monitored concentration C of the oxidant is calculated, is the absorbance, is the molar absorptivity, is the optical path length, is the offset correction amount; Apply a 5-point moving average filter to the original absorbance, and calculate the noise-reduced absorbance through the formula where, is the time index, is the original absorbance; ; Based on the obtained noise-reduced absorbance, construct an oxidant concentration change rate model through multi-level filtering and feature extraction; Specifically, based on 5-point moving average filtering, introduce wavelet transform for noise reduction to separate high-frequency noise and effective signals; For transient bubble interference, adopt a dynamic threshold rejection algorithm. If three consecutive data points deviate from the mean by ±3σ, they are marked as invalid and interpolated for completion, where σ is the standard deviation of deviation; The oxidant concentration change rate is obtained by taking the ratio of the change amount of the monitored oxidant concentration within a unit time to the change time; ; Fit a continuous curve through piecewise cubic spline interpolation to solve the derivative calculation error caused by discrete sampling; Based on the obtained oxidant concentration change rate; , through the formula: ; Construct a dynamic objective function; , where, is the theoretical concentration value generated by the power-law equation of the reaction kinetics model; Based on the obtained dynamic objective function, calculate the reaction time deviation degree through the formula ; ; Based on the reaction time deviation degree and real-time working conditions, design a multi-modal feeding frequency adjustment mechanism; Specifically, in the low deviation area: if < 0.1, maintain the current feeding frequency, pre-adjust the parameters through model predictive control to reduce subsequent fluctuations; In the medium deviation area: 0.1 ≤ < 0.3, start the fuzzy logic controller. The input parameters include the reaction time deviation degree, oxidant concentration gradient, and reactor temperature, and the output is the adjustment amount of the feeding frequency; In the high deviation area: ≥ 0.3, trigger the interlock mechanism, select the optimal strategy by combining the reinforcement learning decision tree, and urgently add oxidant or suspend feeding; Develop a dual-channel feeding system: the main channel feeds at the set feeding frequency f, and the auxiliary channel is dynamically fine-tuned according to the reaction time deviation degree; Introduce a lag compensation algorithm. Based on the residence time distribution of the oxidant in the reactor, calculate the feeding delay effect in advance to reduce the risk of overshoot; Specifically, use the two-parameter axial dispersion model to solve the RTD function, and take the time corresponding to the cumulative distribution F(t) = 90%; Dynamically fine-tune the real-time received target frequency according to the reaction time deviation degree, and calculate the dynamic time delay by multiplying the ratio of the current flow rate to the calibrated flow rate by the original delay; Retrieve the frequency command at the current time minus the dynamic delay time from the historical database, and output the compensated oxidant feeding frequency. Step 3: Obtain the feeding parameters of multiple historical reaction cycles. Through establishing a three-layer intelligent regulation model, deeply coordinate and optimize the reaction time and feeding parameters, and construct a time-scale associated inactivation model and a time-scale coupled free radical prediction model to solve the optimal adjustment amount. Specifically, obtain the feeding parameters of multiple historical reaction cycles. The feeding parameters include but are not limited to: catalyst dispersion, feeding speed, free radical concentration, initiator concentration, substrate concentration, and oxidant feeding frequency. Establish a three-layer intelligent regulation model: The first layer, construct a time-scale associated inactivation model: Use the dynamic time warping algorithm to align the time axes of different cycles, and divide them according to the reaction stage into: initiation stage, oxidant concentration from 0 to 80%; main reaction stage, concentration from 80% to 95%; decay stage, concentration from 95% to the end point. Obtain the ratio of the average time when the historical oxidant activity decays to 80% to the current time when the oxidant activity decays to 80% to get the time-scale scaling factor ; Through the attention mechanism, weight the historical similar cycle data and output the predicted inactivation curve: ; Among them, is the original inactivation rate, is the oxidant concentration change rate; The second layer, construct a time-scale coupled free radical prediction model: Based on the oxidation reaction chain mechanism, through the free radical kinetic equation: ; Obtain the free radical concentration compensation term , among which, is the initiation rate constant of the initiator, is the initiator concentration, is the termination rate constant of the free radical, is the free radical concentration, is the chain growth rate constant of the free radical, is the substrate concentration; Use the Transformer model to learn the mapping relationship between the historical free radical concentration and the feeding parameters, and output the compensation term; The third layer, construct a multi-objective optimization solver: Based on the multi-objective optimization solver, the formula is:

[0011] ; The constraint conditions are: predicting the free radical concentration , , , ; wherein, is the minimum reaction time, is the feeding frequency, is the vibration frequency of catalyst dispersion, is the economic weight factor, is the catalyst loss; The optimal adjustment amount is calculated by a multi-objective optimization solver; Step 4: Based on the optimal adjustment amount, perform real-time adaptive regulation on the feeding parameters, and integrate predictive maintenance for feedback iteration; Input compensation parameters: oxidant feeding frequency, vibration frequency of catalyst dispersion, and free radical concentration; Through the control equation: ; The adjusted compensation frequency is obtained, wherein, is the proportional gain, is the inactivation derivative compensation, is the actual value of free radical concentration, is the predicted value of free radical concentration, is the reference value of free radical concentration; Input the feeding parameters into the knowledge base, and write the operation log every cycle; generate a simulation scenario through digital twin in the knowledge base; recommend feeding parameter adjustment through the optimizer and update the control feeding parameters; Perform reinforcement learning based on catalyst deactivation acceleration, perform Bayesian parameter optimization based on insufficient oxidant utilization, perform fault tree analysis based on abnormal mechanical vibration and expand the rule base.

[0012] Example 2

[0013] As Figure 2 shown, a feeding system for intermediate synthesis provided by an embodiment of the present invention specifically includes the following modules: Alkylation adjustment module: Based on monitoring and analyzing the distribution of the catalyst bed during the alkylation process, establish an inactivation model to dynamically compensate the catalyst activity in combination with pressure fluctuations, and adjust the catalyst dispersion and feeding speed through the inactivation model; Oxidation adjustment module: Based on the end of alkylation, obtain the oxidant concentration during the oxidation process, calculate the oxidant change rate to establish an objective function with the reaction time, calculate the reaction time deviation degree through the objective function, and intelligently adjust the oxidant feeding frequency in combination with the analysis; Adjustment and update module: Obtain the feeding parameters of multiple historical reaction cycles, and through establishing a three-layer intelligent regulation model, deeply optimize the reaction time and feeding parameters in coordination, and construct a time-scale associated inactivation model and a time-scale coupled free radical prediction model to solve the optimal adjustment amount; Adaptive module: Based on the optimal adjustment amount, perform real-time adaptive regulation on the feeding parameters, and integrate predictive maintenance for feedback iteration.

[0014] The above has described an embodiment of the present invention in detail, but the content described is only a preferred embodiment of the present invention and cannot be considered as limiting the scope of implementation of the present invention; the above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the real situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation and historical experience and can be adjusted according to the actual situation; the above is only a preferred embodiment of the present invention and is not used to limit the present invention. All equivalent changes and improvements made according to the scope of the present invention application should still fall within the scope covered by the patent of the present invention.

Claims

1. A feeding method for intermediate synthesis, characterized in that, It includes the following steps: Based on monitoring and analyzing the distribution of the catalyst bed during the alkylation process, a deactivation model is established in combination with pressure fluctuations to dynamically compensate for the catalyst activity. The catalyst dispersion and feeding rate are adjusted through the deactivation model; Based on the end of alkylation, the oxidant concentration is obtained during the oxidation process, and the objective function of the oxidant change rate and reaction time is established. The reaction time deviation degree is calculated through the objective function, and the oxidant feeding frequency is intelligently adjusted in combination with the analysis; The feeding parameters of multiple historical reaction cycles are obtained. Through the establishment of a three-layer intelligent control model, the reaction time and feeding parameters are deeply coordinated and optimized, and a time-scale associated deactivation model and a time-scale coupled free radical prediction model are constructed to solve the optimal adjustment amount; Based on the optimal adjustment amount, the feeding parameters are adaptively regulated in real time, and predictive maintenance is integrated for feedback iteration.

2. The feeding method for intermediate synthesis according to claim 1, wherein, The method for compensating the catalyst activity is as follows: The model updates through an online learning mechanism and fine-tunes the weights every 5 minutes with new data to adapt to the dynamic aging of the catalyst; When the model detects that the activity coefficient drops to the preset activity threshold, dynamic compensation is triggered: If the temperature in the middle reaction zone > 135°C dominates deactivation, the feeding rate is preferentially adjusted; If the pressure fluctuation pressure difference > 10% dominates deactivation, indicating blockage, the catalyst dispersion is preferentially adjusted, and the bed vibrator is started to improve the distribution; The catalyst dispersion is through an adjustable distribution plate built into the bed. If the activity coefficient decreases and the temperature distribution is uneven, and the temperature difference in the top preheating zone > 15°C, the vibration frequency is increased; Combined with the pressure data, if the pressure difference suddenly increases, reverse purging is started; The initial setting is corrected by real-time data through the reaction kinetics model: when the predicted decay rate of activity > 5% / min, the feeding rate is gradually reduced by 0.5% each time, and vice versa.

3. A feeding method for intermediate synthesis according to claim 1, characterized in that, The method for adjusting the catalyst dispersion and feeding rate is as follows: The pressure data of the reaction system are collected in real time through a pressure sensor, including the inlet / outlet pressure and pressure difference of the bed layer, and are synchronously input into the system; The real-time pressure data are combined with the corrected temperature data and used as the input of the catalyst activity calculation model; An integrated learning method is adopted, and the input layer is: corrected temperature data, pressure fluctuation data, and supplementary operating condition parameters; The output layer is the catalyst activity coefficient, and the catalyst activity coefficient is scaled from 0 to 1, where 1 represents full activity; Based on the difference between the catalyst activity coefficient and the standard activity coefficient, the catalyst dispersion and feeding rate are adjusted proportionally.

4. A feeding method for the synthesis of an intermediate according to claim 1, characterized in that, The method for obtaining the objective function is as follows: Based on the 5-point moving average filter, wavelet transform is introduced for noise reduction to separate high-frequency noise and effective signals; For transient bubble interference, a dynamic threshold rejection algorithm is adopted. If 3 consecutive data points deviate from the mean by ±3σ, they are marked as invalid and interpolated and complemented, where σ is the deviation standard deviation; The oxidant concentration change rate is obtained by processing the ratio of the change amount of the monitored oxidant concentration within a unit time to the change time ; The continuous curve is fitted by piecewise cubic spline interpolation to solve the derivative calculation error caused by discrete sampling; Based on the obtained oxidation agent concentration change rate , through the formula: ; Construct a dynamic objective function , where is the theoretical concentration value, which is generated by the power-law equation of the reaction kinetics model.

5. A feeding method for intermediate synthesis according to claim 1, characterized in that, The method for calculating the reaction time deviation degree is as follows: Based on the obtained dynamic objective function, through the formula calculate the reaction time deviation .

6. A feeding method for intermediate synthesis according to claim 1, characterized in that, The method for adjusting the oxidant feeding frequency is as follows: Based on the reaction time deviation degree and real-time operating conditions, a multi-modal feeding frequency adjustment mechanism is designed; Low deviation area: If < 0.1, maintain the current feeding frequency and pre-adjust the parameters through model predictive control; Middle deviation range: 0.1 ≤ <0.3, start the fuzzy logic controller, and the input parameters include the reaction time deviation degree, the oxidant concentration gradient, and the reactor temperature, and the output is the adjustment amount of the feeding frequency; High deviation area: ≥0.3, trigger the interlock mechanism, select the optimal strategy by combining the reinforcement learning decision tree, and urgently supplement the oxidant or suspend the feeding; A dual-channel feeding system is developed: the main channel feeds at the set feeding frequency f, and the auxiliary channel is dynamically fine-tuned according to the reaction time deviation degree; Introduce a lag compensation algorithm, and based on the residence time distribution of the oxidant in the reactor, calculate the feed delay effect in advance; Use a two-parameter axial dispersion model to solve the RTD function, and take the time corresponding to the cumulative distribution F(t)=90%; Dynamically fine-tune the real-time received target frequency according to the reaction time deviation, and calculate the dynamic time delay by multiplying the ratio of the current flow rate to the calibrated flow rate by the original delay; Retrieve the frequency command at the time obtained by subtracting the dynamic time delay from the current time from the historical database, and output the compensated oxidant feed frequency.

7. A feeding method for the synthesis of an intermediate according to claim 1, characterized in that, The method for establishing a three-layer intelligent control model is as follows: Establish a three-layer intelligent control model: The first layer is to construct a time-scale associated deactivation model; Use the dynamic time warping algorithm to align the time axes of different cycles, and divide them according to the reaction stage: the oxidant concentration in the initiation stage is 0→80%; the concentration in the main reaction stage is 80%→95%; the concentration in the attenuation stage is 95%→end point; Obtain the ratio of the average time for the historical oxidant activity to decay to 80% to the time for the current oxidant activity to decay to 80% to obtain the time scale scaling factor ; Output the predicted deactivation curve by weighting the historical similar cycle data through the attention mechanism; The second layer is to construct a time-scale coupled free radical prediction model; Based on the chain mechanism of the oxidation reaction, obtain the free radical concentration compensation term through the free radical kinetic equation; Use the Transformer model to learn the mapping relationship between the historical free radical concentration and the feed parameters, and output the compensation term The third layer is to construct a multi-objective optimization solver.

8. A feeding method for synthesizing an intermediate according to claim 1, characterized in that, The method for solving the optimal adjustment amount is as follows: Based on the multi-objective optimization solver, the formula is: ; The constraint is: predicting the free radical concentration , , , ; Among them, is the minimum reaction time, is the feeding frequency, is the vibration frequency of catalyst dispersion, is the economic weight factor, is the catalyst loss; Calculate the optimal adjustment amount through the multi-objective optimization solver.

9. A feeding method for intermediate synthesis according to claim 1, characterized in that The method for the integrated predictive maintenance to perform feedback iteration is as follows: Input compensation parameters: oxidant feed frequency, catalyst dispersion vibration frequency, and free radical concentration; Through the control equation: ; Obtain the adjusted compensation frequency , where is the proportional gain, is the deactivation derivative compensation, is the actual value of the free radical concentration, is the predicted value of the free radical concentration, is the reference value of the free radical concentration; Input the feed parameters into the knowledge base, write the operation log every cycle; perform digital twin through the knowledge base to generate a simulation scenario; recommend feed parameter adjustments through the optimizer and update the control feed parameters; Perform reinforcement learning based on the acceleration of catalyst deactivation, perform Bayesian parameter optimization based on insufficient utilization of the oxidant, and perform fault tree analysis and rule base expansion based on abnormal mechanical vibration.

10. A feeding system for intermediate synthesis, which is used to execute the feeding method described in any one of the above claims 1-9, characterized in that, Include: Alkylation adjustment module: Based on monitoring and analyzing the distribution of the catalyst bed during the alkylation process, establish a deactivation model in combination with pressure fluctuations to dynamically compensate the catalyst activity, and adjust the catalyst dispersion and feed rate through the deactivation model; Oxidation adjustment module: Based on the end of alkylation, obtain the oxidant concentration during the oxidation process, calculate the oxidant change rate to establish an objective function with the reaction time, calculate the reaction time deviation through the objective function, and intelligently adjust the oxidant feed frequency in combination with the analysis; Adjustment and update module: Obtain the feed parameters of multiple historical reaction cycles, deeply co-optimize the reaction time and feed parameters through the establishment of a three-layer intelligent control model, construct a time-scale associated deactivation model and a time-scale coupled free radical prediction model to solve the optimal adjustment amount; Adaptive module: Based on the optimal adjustment amount, perform real-time adaptive control on the feed parameters, and perform feedback iteration through integrated predictive maintenance.

Images