Fluralaner solution preparation process and system

Abstract

This invention discloses a process and system for preparing fluorellana solution. The process is executed by a central control unit and monitored and fed back in real time by a process analysis module. The steps include: S1: Constructing an isoxazoline core in an addition cyclization reactor, dynamically adjusting the temperature to 70-85℃ based on near-infrared spectral data; S2: Transferring to a hydrolysis reactor, with coordinated control of the alkali addition rate and temperature, maintaining pH 12±0.5 and 95℃±1℃; S3: In an amidation reactor, adjusting the acyl chloride dropping rate and cooling intensity through exothermic curve fitting, controlling the temperature to 0℃±1℃; S4: Delivering to a purification system, crystallizing according to crystal characteristics using a nonlinear adaptive cooling curve. The central control unit is connected to a big data analysis and self-learning unit for post-batch optimization of the process parameter database, ensuring stable product quality. Simultaneously, the system dynamically adjusts the temperature setting by synchronously monitoring the spectral signals of by-products, reducing the generation of harmful by-products and improving raw material utilization and product purity.

Description

Technical Field

[0001] This invention relates to the synthesis technology of fluorenarian, specifically to the synthesis technology of fluorenarian droplets, and particularly to a process for preparing fluorenarian solution. Background Technology

[0002] Fluranar is a novel isoxazoline broad-spectrum insecticide that exerts its highly effective insecticidal action by irreversibly inhibiting the GABA-gated chloride ion channels of pests. It is widely used for the control of ectoparasites in pets and livestock. Its chemical structure comprises an isoxazoline core, a trifluoromethyl-substituted benzene ring, and an amide side chain. The synthetic route involves multiple intricate organic reactions, including key steps such as cyclization, alkaline hydrolysis, amidation, and crystallization purification.

[0003] Traditional fluranar synthesis processes rely heavily on batch reactions and manual experience control, resulting in unstable reaction conditions, high byproduct formation rates, and significant batch-to-batch quality fluctuations. For example, in the isoxazoline core construction stage, reaction temperature significantly affects product selectivity, but traditional temperature control methods are slow to respond and difficult to adapt to real-time changes in reaction kinetics. In the hydrolysis reaction, the amount of alkali added and the endpoint determination depend on offline sampling analysis, leading to insufficient or excessive hydrolysis. Amide formation is a strongly exothermic reaction, and improper control of the feeding rate and cooling can easily cause local overheating, resulting in product decomposition or safety accidents. Inaccurate control of crystal particle size distribution during crystallization affects the solubility and bioavailability of the final formulation. Summary of the Invention

[0004] In view of the above, the present invention provides a process and system for preparing a fluranar solution to solve the technical problems mentioned in the background.

[0005] The preparation process of fluorellaranoside solution includes the following steps:

[0006] The process is executed by a central control unit and monitored and fed back in real time using a process analysis module, including the following steps:

[0007] S1: In an addition cyclization reactor, methyl 2-methyl-4-cyanobenzoate is reacted with 1,3-dichloro-5-(1-trifluoromethylvinyl)benzene to construct an isoxazoline core. The central control unit, based on the near-infrared spectral data of the reaction solution collected in real time by the first reaction monitoring module, performs real-time comparison through the built-in process parameter database and instructs the first adaptive temperature control module to dynamically adjust the reaction temperature within the range of 70~85℃ to ensure that the reaction rate and by-product formation rate are within the preset optimal range.

[0008] S2: The product obtained in step S1 is transferred to the hydrolysis reaction tank through the first delivery valve and the first delivery pump for hydrolysis; the central control unit dynamically adjusts the alkali addition rate and hydrolysis temperature through the linkage control of the second adaptive temperature control module and the third metering pump based on the fusion feedback of the online pH sensor and the online conductivity meter in the second reaction monitoring module, so that the pH value is stabilized at 12±0.5 and the temperature is maintained at 95℃±1℃.

[0009] S3: The carboxylic acid intermediate obtained in step S2 is transferred to an amidation reaction vessel for amidation; the central control unit, based on the exothermic curve detected by the high-sensitivity reaction thermocouple in the third reaction monitoring module, and by fitting it with the standard exothermic curve in the process parameter database, proactively instructs the third adaptive temperature control module and the fifth metering pump to form a closed-loop control, and adjusts the acyl chloride drop acceleration rate and cooling intensity in real time to keep the reaction temperature precisely maintained at 0℃±1℃.

[0010] S4: The reaction mixture obtained in step S3 is sent to the purification system; the central control unit controls the program cooling module of the recrystallization device to crystallize according to the crystal particle size and morphology data from the in-situ particle characteristic monitoring module, and obtains Freranar crystals with the target particle size distribution.

[0011] S5: Dissolve the Frelanar crystals obtained in step S4 with an alcohol solvent at 25~40℃ according to the formula ratio to obtain a Frelanar solution;

[0012] The central control unit is also connected to a big data analysis and self-learning unit, which is used to optimize the process parameter database after each batch.

[0013] Furthermore, the first reaction monitoring module is configured to: acquire a spectrum every 30 seconds based on the built-in spectral acquisition and preprocessing unit; the central control unit uses partial least squares method to calculate the remaining concentration of key raw materials within 5 seconds; when three consecutive data points show that the main product generation rate is lower than 10% of the predicted value, a step-by-step temperature increase of 1~3℃ correction program is triggered, and the by-product spectral signal is monitored simultaneously to determine whether to maintain the new temperature.

[0014] Furthermore, the control logic of the second reaction monitoring module is as follows: the central control unit cross-verifies the progress of the hydrolysis reaction by integrating and analyzing the changing trends of pH value and conductivity; when the pH value is stable within the set range and the rate of change of conductivity approaches zero, it determines that the hydrolysis reaction is complete and automatically closes the alkali feed valve.

[0015] Furthermore, the execution logic of the third reaction monitoring module in the amidation reaction is as follows: the central control unit predicts the intensity of the reaction 100-300 milliseconds in advance through the real-time exothermic rate curve, and reduces the feeding rate or enhances cooling in advance. Its control algorithm adopts fuzzy PID control with online parameter self-tuning.

[0016] Furthermore, the in-situ particle characteristic monitoring module and the programmed cooling module form a closed-loop control; when the average particle size of the crystal reaches the lower limit of the target range, the cooling rate is automatically increased by 0.2℃ / min; when the average particle size of the crystal approaches the upper limit of the target range, the cooling rate is automatically reduced by 0.1℃ / min.

[0017] Furthermore, the process analysis module has a built-in process pre-configuration unit; before each batch of production starts, the process pre-configuration unit optimizes the initial process parameters by simulating the entire synthesis process based on the input raw material attribute code and initial process parameters, and preloads the optimized parameter set into the central control unit and the process parameter database.

[0018] Furthermore, the process analysis module has a built-in equipment status monitoring unit for real-time monitoring of the flow accuracy of all metering pumps and the torque of the agitator; based on the monitoring data, the central control unit automatically performs flow compensation when it predicts that the accuracy drift of any metering pump exceeds the set value, or issues a maintenance warning in advance when the torque of any agitator increases abnormally.

[0019] A fluorellarana solution preparation system is provided for the synthesis of fluorellarana solution prepared by the above-described process. The system includes a substitution cyclization unit, a hydrolysis unit, an amidation unit, and a purification unit connected sequentially along the material flow direction, as well as a control module for unified control of the entire system operation.

[0020] The substitution cyclization unit includes an addition cyclization reaction vessel, which is equipped with a first adaptive temperature control module, a first stirrer, a first metering pump for feeding methyl 2-methyl-4-cyanobenzoate, a second metering pump for feeding 1,3-dichloro-5-(1-trifluoromethylvinyl)benzene, and a first reaction monitoring module for real-time monitoring of the cyclization reaction process.

[0021] The hydrolysis unit includes a hydrolysis reaction tank, which is equipped with a second adaptive temperature control module, a second stirrer, a first delivery pump and a first delivery valve for receiving materials from the addition cyclization reaction tank, a third metering pump for adding alkali solution, and a second reaction monitoring module for monitoring the degree of hydrolysis. The second reaction monitoring module includes an online pH sensor and an online conductivity meter.

[0022] The amidation unit includes an amidation reaction vessel, which is equipped with a third adaptive temperature control module, a third stirrer, a second delivery pump and a second delivery valve for receiving materials from the hydrolysis reaction vessel, a fourth metering pump for adding thionyl chloride, a fifth metering pump for adding amine compounds, and a third reaction monitoring module for monitoring the exothermic reaction.

[0023] The purification and refining unit includes an extraction settling tank, a vacuum concentration device, and a recrystallization device connected in sequence; the recrystallization device is equipped with a programmed cooling module and an in-situ particle characteristic monitoring module for monitoring crystal growth; the extraction settling tank is connected to the outlet of the amidation reaction tank through a third delivery pump and a third delivery valve.

[0024] The control module is electrically connected to all metering pumps, delivery pumps, switching valves, adaptive temperature control modules, and reaction monitoring modules in the system. Based on the preset synthesis process program and real-time feedback monitoring data, it coordinates and controls the operation of each unit.

[0025] Furthermore, the first reaction monitoring module is a near-infrared spectral probe, and the spectral data it monitors is transmitted to the control module. The control module has a built-in chemometric algorithm for calculating the reactant concentration based on the spectral data, and dynamically adjusts the temperature setpoint of the first adaptive temperature control module according to the calculation results to achieve adaptive control of the cyclization reaction.

[0026] Furthermore, the monitoring signals from the online pH sensor and online conductivity meter of the second reaction monitoring module are transmitted to the control module. The control module controls the alkaline solution addition rate of the third metering pump in conjunction with the decreasing trend of pH value and the rate of change of conductivity, so as to achieve precise control of hydrolysis reaction and determination of endpoint.

[0027] Furthermore, the third reaction monitoring module monitors the reaction exothermic rate in real time, obtains a real-time exothermic curve, and transmits it to the control module. The control module compares the real-time exothermic curve with the built-in standard exothermic curve and adjusts the drip rate of the fifth metering pump and the cooling power of the third adaptive temperature control module in a forward-looking manner to form a preventive control over the strongly exothermic amidation reaction.

[0028] Furthermore, the in-situ particle characteristic monitoring module monitors the chord length distribution of the crystal in real time, and the control module dynamically adjusts the cooling rate of the programmed cooling module based on the real-time particle size data to achieve control over the particle size and distribution of the Freronar crystal.

[0029] Furthermore, the control module includes a central control unit, a process parameter database, and a big data analysis and self-learning unit; the process parameter database stores the expected trajectories of standard operating parameters and monitoring parameters for each synthesis step; the big data analysis and self-learning unit is used to optimize and update the process parameter database based on process data and product quality data after each batch of production is completed.

[0030] The beneficial effects of this invention are as follows:

[0031] This invention employs a first reaction monitoring module for real-time near-infrared spectral data acquisition during the cyclization reaction step. Combined with the process parameter database built into the central control unit and partial least squares method, it enables rapid calculation of key raw material concentrations and dynamic evaluation of the reaction rate. When the main product formation rate is detected to be lower than the predicted value, the system automatically triggers a stepped temperature increase correction program to ensure that the reaction rate and by-product formation rate remain within the optimal range. This real-time feedback and adaptive control mechanism solves the problems of incomplete reaction or excessive by-products caused by temperature fluctuations in traditional cyclization reactions. Specifically, near-infrared spectral data is acquired every 30 seconds, and the central control unit completes the analysis within 5 seconds, significantly shortening the response time and avoiding the lag of manual intervention. Simultaneously, the system dynamically adjusts the temperature setting by synchronously monitoring by-product spectral signals, reducing the formation of harmful by-products and improving raw material utilization and product purity.

[0032] In the hydrolysis reaction step, this invention integrates an online pH sensor and an online conductivity meter through a second reaction monitoring module, achieving cross-validation and precise control of the reaction process. The central control unit dynamically adjusts the alkali addition rate and hydrolysis temperature by analyzing the trends in pH and conductivity, stabilizing the pH at 12±0.5 and maintaining the temperature at 95℃±1℃. When the pH stabilizes and the rate of change in conductivity approaches zero, the system automatically determines that the hydrolysis reaction is complete and closes the alkali feed valve, avoiding over-hydrolysis or incomplete reaction. This multi-parameter fusion control strategy overcomes the limitations of single-sensor monitoring and improves the accuracy of endpoint determination.

[0033] In the amidation reaction step, this invention uses a third reaction monitoring module to monitor the exothermic curve in real time, and combines this with an online self-tuning fuzzy PID control algorithm in the central control unit to achieve proactive prediction of the reaction intensity. The system adjusts the acyl chloride droplet acceleration rate and cooling intensity 100-300 milliseconds in advance, precisely maintaining the reaction temperature at 0℃±1℃, forming a preventative control closed loop. This mechanism effectively avoids temperature runaway, product decomposition, or safety accidents caused by a strongly exothermic reaction.

[0034] In the purification step, this invention uses an in-situ particle characteristic monitoring module to monitor crystal particle size and morphology in real time, and controls the programmed cooling module to crystallize using a non-linear, adaptive cooling curve. When the average crystal particle size reaches the lower limit of the target range, the system automatically increases the cooling rate by 0.2℃ / min; when it approaches the upper limit, the cooling rate decreases by 0.1℃ / min. This closed-loop control ensures that the freranar crystals have the target particle size distribution, improving the product's solubility and bioavailability.

[0035] This invention achieves continuous optimization and adaptive adjustment of process parameters through big data analysis and self-learning units and process pre-configuration units. Before each batch of production, the process pre-configuration unit simulates the entire synthesis process and preloads an optimized parameter set based on raw material attribute codes and initial parameters. After each batch, the big data analysis and self-learning unit analyzes process data and product quality data, updates the process parameter database, and enables the system to continuously learn and improve. This self-learning mechanism solves the problem of production inconsistencies caused by batch differences in raw materials or environmental changes in traditional processes. Attached Figure Description

[0036] Figure 1 A flowchart of the method provided by the present invention;

[0037] Figure 2 This is a schematic diagram of the system framework principle provided by the present invention;

[0038] Figure 3 A schematic diagram of the structure of the substitution cyclization unit, hydrolysis unit and amidation unit provided by the present invention;

[0039] Figure 4 This is a schematic diagram of the purification unit provided by the present invention;

[0040] In the diagram: 100, Substitution Cyclone Unit; 200, Hydrolysis Unit; 300, Amide Unit; 400, Purification Unit; 101, Addition Cyclone Reaction Vessel; 102, Hydrolysis Reaction Vessel; 103, Amide Reaction Vessel; 107a, First Adaptive Temperature Control Module; 107b, Second Adaptive Temperature Control Module; 107c, Third Adaptive Temperature Control Module; 110a, First Stirrer; 110b, Second Stirrer; 110c, Third Stirrer; 201, Extraction Settling Tank; 301, Recrystallization Unit; 302, Reduced Pressure Concentration Unit; 305, Programmed Cooling Module; 401a, First Metering Pump; 401b, Second Metering Pump; 401c, Third Metering Pump; 401d, Fourth Metering Pump; 401e, Fifth Metering Pump; 402a, First Delivery Valve; 402b, Second Delivery Valve; 402 c. Third delivery valve; 403a. First delivery pump; 403b. Second delivery pump; 403c. Third delivery pump; 500. Control module; 501. Central control unit; 502. Process parameter database; 503. Big data analysis and self-learning unit; 504. Equipment status monitoring unit; 505. Process pre-configuration unit; 600. Process analysis module; 610. First reaction monitoring module; 610c. Spectrum acquisition and preprocessing unit; 610d. Ultraviolet-visible spectral probe; 611. Second reaction monitoring module; 611a. Online pH sensor; 611b. Online conductivity meter; 612. Third reaction monitoring module; 603. High-sensitivity reaction thermocouple; 612a. Reaction liquid transmission sound velocity sensor; 613. In-situ particle characteristic monitoring module; 613a. High-speed imaging probe. Detailed Implementation

[0041] Reference Figures 1 to 4 As shown, the present invention provides a process for preparing a fluorellaranoside solution, comprising the following steps:

[0042] The process is executed by the central control unit 501 and monitored and fed back in real time using the process analysis module 600, including the following steps:

[0043] S1: In addition cyclization reactor 101, methyl 2-methyl-4-cyanobenzoate is reacted with 1,3-dichloro-5-(1-trifluoromethylvinyl)benzene to construct an isoxazoline core; the central control unit 501 compares the near-infrared spectral data of the reaction liquid collected in real time by the first reaction monitoring module 610 with the built-in process parameter database 502 in real time, and instructs the first adaptive temperature control module 107a to dynamically adjust the reaction temperature within the range of 70~85℃ to ensure that the reaction rate and by-product formation rate are within the preset optimal range;

[0044] S2: The product obtained in step S1 is transferred to the hydrolysis reaction tank 102 for hydrolysis through the first delivery valve 402a and the first delivery pump 403a; the central control unit 501 dynamically adjusts the alkali addition rate and hydrolysis temperature through the linkage control of the second adaptive temperature control module 107b and the third metering pump 401c based on the fusion feedback of the online pH sensor 611a and the online conductivity meter 611b in the second reaction monitoring module 611, so that the pH value is stabilized at 12±0.5 and the temperature is maintained at 95℃±1℃.

[0045] S3: The carboxylic acid intermediate obtained in step S2 is transferred to the amidation reaction vessel 103 for amidation; the central control unit 501, based on the exothermic curve detected by the high-sensitivity reaction thermocouple 603 in the third reaction monitoring module 612, and by fitting it with the standard exothermic curve in the process parameter database 502, proactively instructs the third adaptive temperature control module 107c and the fifth metering pump 401e to form a closed-loop control, and adjusts the acyl chloride drop acceleration rate and cooling intensity in real time to keep the reaction temperature precisely maintained at 0℃±1℃;

[0046] S4: The reaction mixture obtained in step S3 is sent to the purification system; the central control unit 501 controls the program cooling module 305 of the recrystallization device 301 to crystallize according to the crystal particle size and morphology data from the in-situ particle characteristic monitoring module 613, and obtains Freranar crystals with the target particle size distribution.

[0047] S5: Dissolve the Frelanar crystals obtained in step S4 with an alcohol solvent at 25~40℃ according to the formula ratio to obtain a Frelanar solution;

[0048] The central control unit 501 is also connected to a big data analysis and self-learning unit 503, which is used to optimize the process parameter database 502 after each batch.

[0049] In step S1, the isoxazoline core construction reaction, methyl 2-methyl-4-cyanobenzoate and 1,3-dichloro-5-(1-trifluoromethylvinyl)benzene undergo a cyclization reaction in addition cyclization reactor 101 within the range of 70-85°C. The first reaction monitoring module 610 uses near-infrared spectroscopy to acquire spectral data of the reaction solution every 30 seconds. The process parameter database 502 built into the central control unit 501 stores the standard spectral characteristics and kinetic parameters of the reaction, and uses partial least squares regression to calculate the concentration of key raw materials and the formation rate of the main product in real time. If the formation rate of the main product is less than 10% of the predicted value for three consecutive monitoring cycles, the system automatically triggers a stepped temperature increase program, raising the temperature by 1-3°C each time, while simultaneously monitoring the intensity of the characteristic peaks of the by-products. If the by-product signal does not significantly increase, the new temperature is maintained until the reaction is complete; otherwise, the temperature is returned to the original temperature and the system parameters are adjusted. The first adaptive temperature control module 107a adopts a PID algorithm with feedforward compensation, which adjusts the flow rate of the jacket heat transfer oil or the electric heating power to achieve dynamic stability of the reaction temperature within the range of ±0.5℃.

[0050] In step S2, the product from S1 is transferred to the hydrolysis reactor 102 via the first delivery valve 402a and the first delivery pump 403a. The second reaction monitoring module 611 integrates an online pH sensor 611a and an online conductivity meter 611b, collecting data every 10 seconds. The central control unit 501 fuses the pH and conductivity signals using Kalman filtering to construct a dynamic trend of the hydrolysis reaction process. When the pH value is below 11.5, the pulse frequency of the third metering pump 401c is increased; when the pH is above 12.5, alkali addition is paused. Simultaneously, the conductivity change rate dC / dt is used to cross-validate the hydrolysis rate: if dC / dt remains below a threshold (e.g., 0.1 mS / cm·min) and the pH stabilizes at 12 ± 0.5, the reaction endpoint is determined, and the alkali feed valve is automatically closed. The second adaptive temperature control module 107b maintains the hydrolysis temperature at 95℃ ± 1℃ by adjusting the steam valve opening to avoid side reactions caused by high temperatures.

[0051] In step S3, the amidation reaction, the carboxylic acid intermediate obtained in S2 is transferred to the amidation reaction vessel 103 and amidation is carried out in the presence of acyl chloride. The third reaction monitoring module 612 uses a high-sensitivity reaction thermocouple 603 to monitor the reaction heat flow curve at a sampling frequency of 100Hz. The central control unit 501 performs sliding window correlation analysis between the real-time exothermic rate and the standard curve in the process parameter database 502. If it is predicted that the exothermic rate will exceed the safety threshold within the next 300 milliseconds, the stepper motor speed of the fifth metering pump 401e is immediately reduced, and the liquid nitrogen cooling flow rate of the third adaptive temperature control module 107c is simultaneously increased. The control algorithm adopts a fuzzy PID with online parameter self-tuning: based on the deviation of the exothermic rate and its rate of change, the proportional, integral, and derivative coefficients are dynamically adjusted to stabilize the reaction temperature at 0℃±1℃.

[0052] In step S4, purification and crystallization, the reaction mixture from S3 is fed to the purification system, where it is extracted, concentrated, and then enters the recrystallization unit 301. The in-situ particle characteristic monitoring module 613, based on the principle of laser diffraction, measures the chord length distribution and roundness of the crystals every 2 minutes. The central control unit 501 dynamically adjusts the cooling strategy of the programmed cooling module 305 based on real-time particle size data D10, D50, and D90: if D50 is below the target lower limit (e.g., 50 μm), the cooling rate is increased by 0.2 °C / min to promote nucleation; if D50 is close to the upper limit (e.g., 80 μm), the cooling rate is decreased by 0.1 °C / min to inhibit excessive growth. The cooling curve is fitted using a piecewise exponential function to ensure that the coefficient of variation (CV) of the crystal particle size distribution is less than 15%.

[0053] In step S5, the frelanar crystals obtained in step S4 are slowly added to an alcohol solvent (propylene glycol, polyethylene glycol 400, etc.) preheated to 25-40°C according to the formula ratio, and stirred at a constant speed until completely dissolved to obtain a frelanar solution.

[0054] After each batch, the big data analysis and self-learning unit 503 collects time-series data on temperature, pH, spectrum, particle size, etc., and correlates them with product quality indicators such as purity, yield, and particle size distribution. Through principal component analysis and random forest regression, it identifies the mapping relationship between critical process parameters (CPP) and critical quality attributes (CQA), updating the standard curve and control thresholds in the process parameter database 502. For example, if three consecutive batches of data show that the optimal temperature for the cyclization reaction should be adjusted to 78℃ (originally 80℃), the big data analysis and self-learning unit 503 will correct the temperature setpoint in the process parameter database 502 and preload the optimized parameters in the next batch.

[0055] In some embodiments, the first reaction monitoring module 610 is configured to: acquire a spectrum every 30 seconds based on the built-in spectral acquisition and preprocessing unit 610c; the central control unit 501 uses partial least squares method to calculate the remaining concentration of key raw materials within 5 seconds; when three consecutive data points show that the main product generation rate is lower than 10% of the predicted value, a step-by-step temperature increase of 1~3℃ correction procedure is triggered, and the by-product spectral signal is monitored simultaneously to determine whether to maintain the new temperature.

[0056] In the above-described reaction monitoring module 610, an ultraviolet-visible spectral probe 610d is further configured to work in conjunction with a near-infrared spectral probe to form a multispectral fusion analysis system. The near-infrared spectroscopy focuses on the overtone absorption of CH and NH bonds, used for quantitative analysis of the remaining concentration of the raw material methyl 2-methyl-4-cyanobenzoate; the ultraviolet-visible spectroscopy, with a wavelength range of 200-400 nm, specifically monitors the formation of byproduct dimers or oxidation products with conjugated structures. The central control unit 501 performs a fusion calculation on the dual-spectral data once per second using a built-in multivariate statistical analysis algorithm, generating a comprehensive index called the Reaction Health Index (RHI). The RHI is calculated by weighting three dimensions: the main product formation rate, the rate of decay of key raw material concentrations, and the absorbance of characteristic byproducts. When the RHI remains below the dynamic threshold for one minute, and this threshold is automatically set by the process parameter database 502 based on the initial raw material activity, the correction procedure is no longer limited to step-by-step heating. Instead, it initiates a multi-mode optimization sequence: First, the system fine-tunes the pulse frequency of the first metering pump 401a or the second metering pump 401b to compensate for potential local concentration inconsistencies or slight feeding errors in a particular raw material, with an adjustment range of ±5%. Second, in terms of temperature adjustment, a pulse heating strategy is adopted, i.e., rapidly increasing the temperature by 2°C within 2 minutes, maintaining it for 5 minutes, and then decreasing it by 1°C. This short-term high-temperature shock disrupts the reaction equilibrium, promoting the main reaction pathway. Simultaneously, ultraviolet spectroscopy is used to closely monitor byproduct signals. If the byproduct growth rate exceeds a critical value, the pulse heating is immediately terminated, and the system switches to a conservative isothermal control mode. This multi-spectral, multi-actuator linkage control achieves a leap from single temperature control to temperature-feed synergistic optimization, greatly improving the robustness and yield of the cyclization reaction.

[0057] In the above, the central control unit 501 performs time-series analysis on the spectral data collected by the first reaction monitoring module 610 to accurately predict the reaction endpoint and self-calibrate the prediction parameters. When the second derivative of the main product concentration changes from positive to negative and stabilizes close to zero, it indicates that the reaction is entering its final stage. The central control unit 501 will initiate the endpoint prediction program in advance: using the spectral data sequence of the past 15 minutes, it predicts the concentration trajectory for the next 5 minutes through a long short-term memory network algorithm. If the prediction shows that the main product concentration will reach a change rate of <0.1% / min, the system determines that the reaction is about to end and issues a transfer warning to the operator 2 minutes in advance. In addition, after each batch of reaction is completed, the system will send the complete spectral time-series data of this batch, as well as the product purity and yield data obtained from the final offline laboratory analysis, to the big data analysis and self-learning unit 503. The big data analysis and self-learning unit 503 calculates the prediction residual of the parameters for this batch by comparing the spectral prediction value with the actual laboratory value. If the residual continues to deviate, for example, if the mean residual of three consecutive batches exceeds twice the historical standard deviation, the parameter update process is automatically triggered.

[0058] The first reaction monitoring module 610 and the central control unit 501 also integrate a reaction safety early warning function. The system pre-stores dangerous spectral fingerprints representing abnormal reactions such as violent decomposition and catalyst deactivation in the process parameter database 502. During the reaction, the real-time acquired spectrum is quickly matched and calculated with these dangerous fingerprints. When any of the following situations occur: 1) the similarity between the real-time spectrum and any dangerous fingerprint exceeds 85%; 2) the calculated apparent activation energy increases abnormally by more than 20% within two consecutive monitoring cycles; the system will immediately trigger the highest level of safety warning. The central control unit 501 will instantly execute an emergency cooling and dilution procedure: the first adaptive temperature control module 107a switches to maximum cooling power, and at the same time starts an emergency solvent metering pump to quickly inject a predetermined amount of inert solvent into the addition cyclization reaction vessel 101 to dilute the reaction system, reduce the concentration of reactants, and rapidly remove heat. The entire process is completed within 30 seconds, and the raw material feed is simultaneously cut off, an audible and visual alarm is activated, and the operating interface is locked. This mechanism elevates process analysis technology from simple process optimization to proactive safety protection, effectively preventing serious safety accidents such as overheating and overflow that may be caused by exothermic reactions.

[0059] In some embodiments, the control logic of the second reaction monitoring module 611 is as follows: the central control unit 501 cross-verifies the progress of the hydrolysis reaction by integrating and analyzing the changing trends of pH value and conductivity; when the pH value is stable within the set range and the rate of change of conductivity approaches zero, it determines that the hydrolysis reaction is complete and automatically closes the alkali feed valve.

[0060] In the above process, during the initial stage of the reaction, due to the high substrate concentration and strong driving force, the system adopts a rapid-increase-decrease alkali addition strategy. This involves quickly raising the pH to 11.0 at a high flow rate, then gradually reducing the alkali addition rate to allow the pH to gradually approach and stabilize at 12.0. This avoids hydrolysis side reactions caused by localized overconcentration of the alkali solution in the initial stage. The central control unit 501 also analyzes the second derivative of the conductivity curve to predict and prevent scaling problems caused by the precipitation of inorganic salts, such as sodium chloride crystals generated in the reaction. If an inflection point is detected in the conductivity curve and the second derivative becomes significantly negative, indicating that the salt concentration may be approaching saturation, the system immediately fine-tunes the second adaptive temperature control module 107b, temporarily increasing the hydrolysis temperature by 1-2°C and increasing the rotation speed of the second stirrer 110b by 50 rpm for 10 minutes. This enhances mass transfer and increases solubility to dissolve any small crystal nuclei that may form, preventing their deposition on the vessel walls or sensor probe surfaces, ensuring the uniformity of the reaction and the long-term reliability of the monitoring data.

[0061] This application features a dual verification mechanism for the hydrolysis endpoint and an automated quality confirmation process before material transfer. The first endpoint determination is based on pH stability and a near-zero rate of change in conductivity. To prevent false positives, the system introduces a second verification: the central control unit 501 instructs an online micro-sampler installed on the hydrolysis reactor 102 to automatically aspirate 0.5 mL of the reaction solution after the system's initial endpoint determination. This sample is then transported through a heat-insulated capillary line to an integrated micro-refractive index meter to measure the refractive index. Because the solution composition changes significantly before and after hydrolysis, the refractive index exhibits a characteristic transition. The system compares the measured refractive index with the standard refractive index range under fully hydrolyzed conditions in the process parameter database 502. Only when the two match is the endpoint determination finally confirmed. After confirming complete hydrolysis, the system performs a precondition check before initiating material transfer. This includes confirming that the target amidation reactor 103 is idle and ready, confirming that the purging and cleaning of the delivery pipeline via the first delivery valve 402a and the first delivery pump 403a has been completed, and checking whether the temperature of the hydrolysis product is within the required delivery range, such as 90-100℃. Only when all conditions are met will the central control unit 501 issue a command to open the first delivery valve 402a and the first delivery pump 403a, achieving a safe, seamless, and fully automatic transfer from the hydrolysis unit 200 to the amidation unit 300.

[0062] The big data analysis and self-learning unit 503 is specifically designed to optimize the control parameters for alkali addition during the hydrolysis reaction. After each batch of hydrolysis reaction is completed, the big data analysis and self-learning unit 503 extracts the complete pH-time curve and conductivity-time curve for that batch. Through nonlinear fitting, it calculates the hydrolysis rate constant k and alkali consumption equivalent for that batch. These kinetic parameters are associated with the batch code of the raw material methyl 2-methyl-4-cyanobenzoate used in that batch and stored in a continuously expanding raw material database. When a new round of production begins, the process pre-configuration unit 505 queries the raw material database for matching or most similar historical data based on the input raw material code. If the estimated hydrolysis rate constant k of the new batch of raw material is found to deviate from the standard value by more than 10%, the system will automatically adjust the PID control parameters of the third metering pump 401c before the hydrolysis reaction starts: for raw materials with high reactivity, a more conservative integral term Ii is used to prevent overshoot; for raw materials with low reactivity, the proportional term Kp is appropriately increased to accelerate the response speed.

[0063] In some embodiments, the execution logic of the third reaction monitoring module 612 in the amidation reaction is as follows: the central control unit 501 predicts the intensity of the reaction 100-300 milliseconds in advance through the real-time exothermic rate curve, and reduces the feeding rate or enhances cooling in advance. Its control algorithm adopts fuzzy PID control with online parameter self-tuning.

[0064] The third reaction monitoring module 612 further integrates a reaction liquid transmission-type sound velocity sensor 612a, used to synchronously monitor the sound velocity change of the reaction system. This parameter is closely related to the density, viscosity, and composition of the reaction liquid and can serve as an auxiliary sensing signal for exothermic reactions. The central control unit 501 constructs a comprehensive reaction intensity index, which is a weighted fusion value of the exothermic rate and the sound velocity change rate. Based on this comprehensive reaction intensity index, a gradient feeding control strategy is adopted: the entire dropping process of acyl chloride is divided into three stages: initial, main reaction, and final stage. In the initial stage, the first 10% of the material uses a low fixed dropping rate to allow the system to be gently initiated; in the main reaction stage, the system switches to dynamic control based on fuzzy PID; in the final 15% of the material, when the comprehensive index shows that the reaction driving force is weakening, the system will actively increase the dropping rate by 10%-20% to shorten the total reaction time, but at the same time tighten the upper limit of temperature control to -0.5℃~0℃ to ensure safety in the later stages. This multimodal sensing-based, phased, gradient feeding strategy achieves the optimal balance between reaction efficiency and safety.

[0065] To address the potential for localized overheating during the highly exothermic reaction, the amidation reactor 103 is equipped with a composite stirring system featuring multiple impellers and baffles, as well as three temperature sensor points T1, T2, and T3 distributed at different heights along the reactor wall. The third reaction monitoring module 612 calculates the axial and radial temperature gradients within the amidation reactor 103 based on the temperature readings from these three points. The central control unit 501 combines this temperature field distribution information with the real-time exothermic curve. When a significant temperature difference (>2°C) is detected between the bottom temperature T1 and the top temperature T3, indicating temperature stratification, the system will automatically control the third stirrer 110c to perform variable-speed stirring: during peak exothermic periods, the stirring speed is automatically increased from the standard 200 rpm to 280 rpm to enhance heat transfer and mixing; when exothermic activity subsides, the speed returns to normal to conserve energy. Meanwhile, the acyl chloride feed pipe route has been changed from a single central dripping to a distributed coil dripping. The central control unit 501 can selectively open the dripping valves in different areas according to the temperature field, so that the reactants are more evenly distributed in the reactor, thus preventing the formation of local hot spots from both the equipment hardware and control module levels.

[0066] The control module 500 establishes a mechanistic model for the amidation unit 300, incorporating mass, energy conservation, and reaction kinetics. This model receives real-time initial conditions from the physical reactor, such as substrate concentration, volume, initial temperature, and the actual acyl chloride feeding rate. During each batch of reaction, the model runs synchronously at a speed faster than real-time, predicting the exothermic trend and temperature distribution of the reaction within the next 1-2 minutes. The central control unit 501 compares the measured data from the third reaction monitoring module 612 with the model's predicted data. If the two match well, the model prediction is trusted, and earlier, smoother, forward-looking control is implemented. If a significant deviation occurs, indicating that the actual reaction does not match the model's expectations, the system immediately switches to a more conservative control mode that relies more heavily on real-time sensors and records the deviation data for subsequent system correction.

[0067] In the above embodiment, the in-situ particle characteristic monitoring module 613 and the programmed cooling module 305 form a closed-loop control; when the average particle size of the crystal is detected to reach the lower limit of the target range, the cooling rate is automatically increased by 0.2℃ / min; when the average particle size of the crystal is detected to be close to the upper limit of the target range, the cooling rate is automatically reduced by 0.1℃ / min.

[0068] The in-situ particle characteristic monitoring module 613 is further equipped with a high-speed imaging probe 613a, capable of capturing dozens of real-time microscopic images of the crystals per second. The image recognition algorithm built into the control module 500, trained based on a convolutional neural network, can analyze the crystal morphology online, such as needle-like, plate-like, blocky, and aspect ratio. This deepens the control logic of the programmed cooling module 305: it is based not only on particle size but also on crystal morphology. For example, when the system detects that the proportion of needle-like crystals is too high and the average chord length is close to the lower limit, it will not simply increase the cooling rate, but will initiate a gentle growth program, reducing the cooling rate by 0.05℃ / min, and simultaneously fine-tune the stirring rate of the recrystallization device 301, such as reducing it by 20%, to reduce shear force and promote more uniform blocky crystal growth. Conversely, if crystal agglomeration is detected, appearing as irregular clumps on the image, the system will immediately increase the stirring rate and inject a small amount of surfactant solution, using an auxiliary anti-agglomeration metering pump to break up the agglomerates, ensuring the accuracy of particle size data and the flowability of the product.

[0069] The control module 500 no longer uses fixed cooling stages. Instead, it calculates and adjusts the supersaturation of the system in real time based on the particle size distribution and crystal number density fed back by the in-situ particle characteristic monitoring module 613. The system has a built-in supersaturation controller with a set value of an optimal range. When the real-time estimated supersaturation is lower than this range, the controller outputs a command to accelerate the cooling rate of the programmed cooling module 305 to increase the supersaturation and promote nucleation and growth. When the real-time supersaturation is higher than this range, the cooling is significantly slowed down or even stopped, allowing the system to consume the existing supersaturation, enabling crystal growth and preventing excessively wide particle size distribution caused by explosive nucleation.

[0070] The central control unit 501 sets a multi-attribute quality space for the final crystal product, including D50, span (D90-D10) / D50, and crystal roundness obtained through image analysis. The crystallization process can only be considered complete if all three indicators fall within the preset acceptable range. Near the end of crystallization, the control module 500 calculates the matching degree between the predicted final crystal product attributes, such as specific surface area and angle of repose, and the ideal raw material requirements for downstream formulation processes. If the matching degree is below 95%, the system will issue an early warning to the production planning department before the batch ends.

[0071] In some embodiments, the process analysis module 600 has a built-in process pre-configuration unit 505. Before each batch of production starts, the process pre-configuration unit 505 optimizes the initial process parameters by simulating the entire synthesis process based on the input raw material attribute code and initial process parameters, and preloads the optimized parameter set into the central control unit 501 and the process parameter database 502.

[0072] The process pre-configuration unit 505, built upon learning from historical production data, can instantly predict the final product quality and yield under any set of initial process parameters. During parameter pre-configuration, this unit not only provides a single optimal parameter set but also performs uncertainty quantification analysis to assess the process robustness when raw material properties fluctuate within tolerance limits. It outputs a master parameter set and several boundary parameter sets. At the start of production, the central control unit 501 prioritizes loading the master parameter set but simultaneously preloads the boundary parameter sets as backups. Once the first reaction monitoring module 610 detects a significant deviation between the actual reaction trajectory and the simulated prediction of the master parameter set in the early stages of the reaction, the system can quickly switch to a boundary parameter set that better matches the current situation.

[0073] In some embodiments, the process analysis module 600 has a built-in equipment status monitoring unit 504 for real-time monitoring of the flow accuracy of all metering pumps and the torque of the agitator; based on the monitoring data, the central control unit 501 automatically performs flow compensation when it predicts that the accuracy drift of any metering pump exceeds the set value, or issues a maintenance warning in advance when the torque of any agitator increases abnormally.

[0074] The central control unit 501 can adaptively control strategies based on the real-time health status provided by the equipment status monitoring unit 504 to protect equipment with slightly degraded performance and maintain production. For example, when slight wear is detected in the drive motor bearing of a metering pump, such as the third metering pump 401c, causing a slower response speed and slight overshoot, the central control unit 501 will not immediately perform large-scale flow compensation. Instead, it will proactively adjust the PID parameters in the pump's control loop: increasing the integral time to reduce overshoot and reducing the proportional gain to smooth the response. Simultaneously, the system may fine-tune the process parameters related to the pump, such as slightly widening the pH control range for the hydrolysis reaction from 12±0.5 to 12±0.7, to tolerate temporary slight declines in equipment performance. This ensures continuous operation of the production line while maintaining core product quality until planned maintenance is required.

[0075] The equipment status monitoring unit 504 extends the monitoring scope to the energy consumption of the entire system, and is equipped with sensors for total power, steam flow, and cooling water flow. The control module 500 calculates in real time the instantaneous energy consumption intensity of each production unit's cyclization, hydrolysis, amidation, and purification processes, such as power consumption per unit output. The central control unit 501 executes system-level energy efficiency optimization while ensuring process safety and product quality. For example, during the exothermic low point of the amidation reaction, the system automatically reduces the cooling power of the third adaptive temperature control module 107c to the minimum level required to maintain the temperature; during the recrystallization stage, if the ambient temperature allows, the programmed cooling module 305 prioritizes the use of free circulating cooling water rather than power-consuming refrigeration units. All these energy efficiency optimization commands are linked to the equipment health status; for example, a pump or agitator with a known potential malfunction will never be allowed to operate under overload in energy-saving mode. This forms an intelligent and green production execution system integrating equipment health, process control, and energy management.

[0076] The present invention also discloses a fluorellarana solution preparation system for the synthesis of fluorellarana solution preparation process. The system includes a substitution cyclization unit 100, a hydrolysis unit 200, an amidation unit 300 and a purification unit 400 connected in sequence along the material flow direction, as well as a control module 500 for unified control of the operation of the entire system.

[0077] The substitution cyclization unit 100 includes an addition cyclization reaction vessel 101, which is equipped with a first adaptive temperature control module 107a, a first stirrer 110a, a first metering pump 401a for feeding the raw material methyl 2-methyl-4-cyanobenzoate, a second metering pump 401b for feeding the raw material 1,3-dichloro-5-(1-trifluoromethylvinyl)benzene, and a first reaction monitoring module 610 for real-time monitoring of the cyclization reaction process.

[0078] The hydrolysis unit 200 includes a hydrolysis reaction tank 102, which is equipped with a second adaptive temperature control module 107b, a second stirrer 110b, a first delivery pump 403a and a first delivery valve 402a for receiving materials from the addition cyclization reaction tank 101, a third metering pump 401c for adding alkali solution, and a second reaction monitoring module 611 for monitoring the degree of hydrolysis. The second reaction monitoring module 611 includes an online pH sensor 611a and an online conductivity meter 611b.

[0079] The amidation unit 300 includes an amidation reaction vessel 103, which is equipped with a third adaptive temperature control module 107c, a third stirrer 110c, a second delivery pump 403b and a second delivery valve 402b for receiving materials from the hydrolysis reaction vessel 102, a fourth metering pump 401d for adding thionyl chloride, a fifth metering pump 401e for adding amine compounds, and a third reaction monitoring module 612 for monitoring the exothermic reaction.

[0080] The purification unit 400 includes an extraction settling tank 201, a vacuum concentration device 302, and a recrystallization device 301 connected in sequence; the recrystallization device 301 is equipped with a programmed cooling module 305 and an in-situ particle characteristic monitoring module 613 for monitoring crystal growth; the extraction settling tank 201 is connected to the outlet of the amidation reaction tank 103 through a third delivery pump 403c and a third delivery valve 402c.

[0081] The control module 500 is electrically connected to all metering pumps, delivery pumps, switching valves, adaptive temperature control modules, and reaction monitoring modules in the system. Based on the preset synthesis process program and real-time feedback monitoring data, it coordinates and controls the operation of each unit.

[0082] In the above, the cyclization unit 100, hydrolysis unit 200, amidation unit 300, and purification unit 400 all adopt a modular design. The core reaction tanks of each unit are connected to the piping system via quick-connect clamps and are equipped with independent local control units. These local control units communicate with the central control unit 501 of the control module 500 via the industrial Ethernet protocol, forming a distributed control network. Specifically, the local control unit of the addition cyclization reaction tank 101 is specifically responsible for processing the high-frequency spectral data of the first reaction monitoring module 610 and executing the fast PID control of the first adaptive temperature control module 107a; the local control unit of the hydrolysis reaction tank 102 is specifically responsible for processing the signals of the online pH sensor 611a and the online conductivity meter 611b and for controlling the third metering pump 401c in conjunction with them; the local control unit of the amidation reaction tank 103 is equipped with a high-speed processor, which is specifically used to run a fuzzy PID algorithm with online parameter self-tuning to achieve millisecond-level forward-looking control of the third adaptive temperature control module 107c and the fifth metering pump 401e.

[0083] Furthermore, the first reaction monitoring module 610 is a near-infrared spectral probe, and the spectral data it monitors is transmitted to the control module 500. The control module 500 calculates the reactant concentration based on the spectral data and dynamically adjusts the temperature setpoint of the first adaptive temperature control module 107a according to the calculation results, so as to realize adaptive control of the cyclization reaction.

[0084] The first reaction monitoring module 610 is not a single near-infrared spectral probe, but rather an array of three near-infrared spectral probes installed at different heights and radial positions within the addition cyclization reactor 101. These probes monitor the reaction liquid spectra at the top, middle, and bottom of the reactor, respectively. By comparing the spectral data from these three locations, the control module 500 can determine whether the reaction system is uniformly mixed and whether a concentration or temperature gradient exists. If the concentration difference of key raw materials between the bottom and top is detected to consistently exceed 5%, the control module 500 will instruct the first stirrer 110a to increase its rotation speed by 50 rpm and continue monitoring until the gradient disappears.

[0085] The first reaction monitoring module 610 further integrates an online near-infrared chemical imager mounted on the sight glass of the reaction vessel. This imager can acquire images of the chemical composition distribution on the reaction liquid surface at a rate of several frames per second. The image processing algorithm built into the control module 500 fuses spectral and spatial information to generate pseudo-color distribution maps of reactant and product concentrations on the liquid surface, thus visualizing the reaction process. By analyzing these chemical images, the system can calculate macroscopic kinetic parameters of the reaction in real time, such as the propulsion rate of the reaction front. When significant local inhomogeneities or stagnant zones are detected in the reaction, the system can not only adjust the stirring but also, in conjunction with the first adaptive temperature control module 107a, perform zoned temperature control of the reaction vessel jacket. For example, it can slightly increase the flow rate of the heat transfer medium in the low-temperature zone to promote the synchronization of the overall reaction, improving the control precision from the reactor level to the intra-reactor region level.

[0086] The addition cyclization reactor 101 is also equipped with a high-frequency acoustic sensor to collect sound signals such as stirring noise and bubbling during the reaction process. The control module 500 performs data fusion analysis of the acoustic signal spectrum and near-infrared spectral data. For example, specific acoustic spectral characteristics may be associated with the activation state of the catalyst or the release of trace gases. At the start of each batch of reaction, the system analyzes the acoustic fingerprint and the first batch of spectral data during the initial feeding and mixing stage, and quickly matches it with batch data of raw materials of different activity levels in the historical database, thereby making a preliminary assessment of the activity of the raw materials being fed in the early stages of the reaction, such as the first 5 minutes. If the assessment results show that the activity of the raw materials is significantly low, the control module 500 will immediately send a signal to the process pre-configuration unit 505 to invoke a more aggressive temperature control strategy pre-set for low-activity raw materials, such as increasing the initial temperature by 2°C, thus realizing a shift from passive response to active adaptation.

[0087] Furthermore, the monitoring signals from the online pH sensor 611a and the online conductivity meter 611b of the second reaction monitoring module 611 are transmitted to the control module 500. The control module 500 controls the alkaline solution addition rate of the third metering pump 401c in conjunction with the decreasing trend of pH value and the rate of change of conductivity, so as to achieve precise control of hydrolysis reaction and endpoint determination.

[0088] The second reaction monitoring module 611 employs a redundant design, comprising two independent online pH sensors 611a and online conductivity meters 611b, installed at different locations within the hydrolysis reactor 102. The control module 500 continuously compares the readings of the two sensors. When the deviation between the readings of the two online pH sensors 611a exceeds 0.1 units, or the deviation between the readings of the two online conductivity meters 611b exceeds 1%, the system determines that sensor drift or malfunction has occurred and automatically triggers a sensor reliability assessment program. This program temporarily freezes the control loop based on the suspected faulty sensor and prioritizes using data from the other sensor to maintain control. Simultaneously, it issues clear calibration and maintenance prompts to the operator through the control interface and records the fault event.

[0089] The system uses kernel partial least squares to calculate the input measurable variables and outputs the real-time concentration of the carboxylic acid intermediate, the core state variable of the hydrolysis reaction. The control module 500 cross-validates the measured concentration value with the endpoint determined by pH-conductivity fusion. If the measured concentration is close to zero but the pH-conductivity endpoint has not been triggered, or vice versa, the system performs in-depth diagnostics to determine whether the problem is system inaccuracy or sensor malfunction.

[0090] The control module of the hydrolysis unit 200 has the function of intelligent scheduling with upstream and downstream units. The control module 500 receives in real time the reaction completion forecast from the upstream substitution cyclization unit 100 and the readiness status of the downstream amidation unit 300. Based on the estimated remaining time of the hydrolysis reaction itself, the central control unit 501 performs global scheduling optimization. For example, if the downstream amidation unit 300 is not yet ready, the system can instruct the hydrolysis unit 200 to slightly lower the temperature at the end of the reaction, such as from 95°C to 90°C, to pause the reaction process and wait for the downstream; conversely, if the hydrolysis is about to be completed and the downstream is ready, the system can start the preheating program of the first delivery valve 402a and the first delivery pump 403a in advance.

[0091] Furthermore, the third reaction monitoring module 612 monitors the reaction exothermic rate in real time, obtains the real-time exothermic curve, and transmits it to the control module 500. The control module 500 compares the real-time exothermic curve with the built-in standard exothermic curve and adjusts the dripping acceleration rate of the fifth metering pump 401e and the cooling power of the third adaptive temperature control module 107c in a forward-looking manner to form a preventive control of the strongly exothermic amidation reaction.

[0092] In addition to the high-sensitivity reaction thermocouple 603, the third reaction monitoring module 612 also integrates an online sensor for the reaction liquid viscosity and a pressure sensor for the reaction vessel. The control module 500 integrates the exothermic rate, viscosity change, and pressure fluctuation from multiple dimensions to construct a more comprehensive reaction runaway risk index. The third adaptive temperature control module 107c is a graded cooling system, including three levels: main circulating water cooling, auxiliary chiller cooling, and emergency liquid nitrogen quenching. The central control unit 501 activates different levels of cooling according to the level of the reaction runaway risk index: at level one, only main circulating water cooling is used; at level two, the chiller auxiliary cooling is activated; at level three, while activating liquid nitrogen quenching, the power supply to the fifth metering pump 401e is simultaneously cut off.

[0093] The control module 500 analyzes the flow field distribution and acyl chloride concentration distribution within the reactor in real time based on the current agitator speed, material properties such as viscosity and density, and the location of the feeding point. The control module 500 uses the analysis results to optimize the feeding strategy. For example, when a flow dead zone exists below the feeding point, the system fine-tunes the drip rate of the fifth metering pump 401e or adjusts the speed of the third agitator 110c to improve local mixing and prevent side reactions caused by excessive local concentration of acyl chloride.

[0094] When managing the amidation reaction, the control module 500 incorporates an energy efficiency and safety balance factor. In the initial and final stages of the reaction, when the exothermic reaction is mild, the system prioritizes energy efficiency, primarily using the highly efficient main circulating water for cooling, and allowing slight temperature fluctuations within the control range of 0℃±1℃ to conserve cooling energy. During the peak exothermic phase, the system prioritizes safety, activating the auxiliary chiller unit regardless of cost, and tightening temperature control to 0℃±0.5℃ to ensure absolute safety.

[0095] Furthermore, the in-situ particle characteristic monitoring module 613 monitors the chord length distribution of the crystal in real time, and the control module 500 dynamically adjusts the cooling rate of the program cooling module 305 based on the real-time particle size data to achieve control over the particle size and distribution of the Freranar crystal.

[0096] The in-situ particle characteristic monitoring module 613 is a multifunctional particle system analyzer that integrates focused beam reflectance measurement, online imaging, and Raman spectroscopy probes. It not only monitors the chord length distribution of crystals in real time, but also analyzes the crystal's aspect ratio and roundness through imaging, and monitors the crystal form and solvate morphology through Raman spectroscopy. Therefore, the control module 500 can acquire real-time data on the three key quality attributes: particle size, crystal habit, and crystal form. The control objectives of the programmed cooling module 305 are also diversified, aiming to control particle size distribution while ensuring the acquisition of advantageous crystal habit and the correct stable crystal form, forming a multi-dimensional closed-loop control of crystal product quality.

[0097] The control module 500 employs a reinforcement learning algorithm to manage the crystallization process. This algorithm targets the multi-dimensional quality attributes of the final product, such as D50, span, and roundness, and uses the programmed cooling curve, stirring rate, and possible anti-agglomeration agent additions as its action space for online learning and optimization. By continuously trying small changes in control actions and observing their immediate impact on crystal quality attributes, the system gradually learns the optimal crystallization control strategy. This AI-based control method can automatically find complex cooling and operation curves that simultaneously satisfy multiple, sometimes conflicting, quality objectives, such as requiring both large granularity and narrow distribution, surpassing traditional control methods that rely on fixed parameters or human experience.

[0098] The control module 500 stores a dataset of the impact of different particle sizes and morphologies of fluoranarium crystals on key performance characteristics of the final drop product, such as dissolution rate, bioavailability, and suspension stability. During crystallization, the system predicts the performance of the current batch of crystalline product in downstream formulations based on real-time monitoring of crystal quality attributes. If the prediction shows that its performance precisely meets the requirements of a specific specification, such as high-concentration drops for large breed dogs, the system maintains the current strategy. If the prediction finds that it better meets the requirements of another specification, such as easily soluble drops for small breed dogs, the system can dynamically adjust the crystallization endpoint criteria, or even proactively change the control target, guiding the crystallization process to produce the product most suitable for that specification. This makes the purification unit 400 an intelligent unit capable of flexible production based on market demand, maximizing product value.

[0099] Furthermore, the control module 500 includes a central control unit 501, a process parameter database 502, and a big data analysis and self-learning unit 503; the process parameter database 502 stores the expected trajectory of standard operating parameters and monitoring parameters for each synthesis step; the big data analysis and self-learning unit 503 is used to optimize and update the process parameter database 502 based on process data and product quality data after each batch of production is completed.

[0100] The big data analysis and self-learning unit 503 possesses transfer learning capabilities. When the production line switches to produce other isoxazoline drugs with structures similar to fluorellana, the big data analysis and self-learning unit 503 can identify similarities between the old and new products in terms of reaction mechanisms and monitored spectral characteristics. Instead of learning the parameters of the new product from scratch, it uses pre-trained fluorellana process parameters, such as PLS parameters for cyclization reactions and fuzzy PID parameters for amidation, as a pre-training set. It then uses data from a small batch of new product batches for rapid fine-tuning and transfer, thereby significantly shortening the process modeling and optimization cycle for new products. This enables the transfer and reuse of production knowledge across different products, demonstrating the system's high level of intelligence and adaptability.

[0101] The control module 500 achieves a fully autonomous closed loop from production to management. After each batch of optimization, the big data analysis and self-learning unit 503 not only updates the process parameter database 502 but also automatically generates a structured batch production and optimization report. This report includes key process indicators for the current batch, comparative analysis with historical batches, the optimization actions performed and their effectiveness evaluation, and maintenance recommendations for equipment and system parameters. Furthermore, the system integrates with the enterprise resource planning system, automatically updating cost accounting based on optimized yield and energy consumption data; and sending pre-release signals to the quality management system based on predicted product quality. Ultimately, this system becomes an autonomous production organism capable of self-awareness, self-analysis, self-decision-making, self-optimization, and self-reporting, representing the highest form of intelligent manufacturing.

Claims

1. A process for preparing a fluranar solution, characterized in that, Includes the following steps: The process is executed by a central control unit (501) and monitored and fed back in real time using a process analysis module (600), including the following steps: S1: In the addition cyclization reaction vessel (101), methyl 2-methyl-4-cyanobenzoate and 1,3-dichloro-5-(1-trifluoromethylvinyl)benzene are subjected to an isoxazoline core construction reaction; the central control unit (501) compares the near-infrared spectral data of the reaction liquid collected in real time by the first reaction monitoring module (610) with the built-in process parameter database (502) in real time, and instructs the first adaptive temperature control module (107a) to dynamically adjust the reaction temperature within the range of 70~85℃ to ensure that the reaction rate and by-product formation rate are within the preset optimal range; S2: The product obtained in step S1 is transferred to the hydrolysis reaction tank (102) for hydrolysis through the first delivery valve (402a) and the first delivery pump (403a); the central control unit (501) dynamically adjusts the alkali addition rate and hydrolysis temperature according to the fusion feedback of the online pH sensor (611a) and online conductivity meter (611b) in the second reaction monitoring module (611), through the linkage control of the second adaptive temperature control module (107b) and the third metering pump (401c), so that the pH value is stabilized at 12±0.5 and the temperature is maintained at 95℃±1℃; S3: The carboxylic acid intermediate obtained in step S2 is transferred to the amidation reaction vessel (103) for amidation; the central control unit (501) according to the exothermic curve detected by the high-sensitivity reaction thermocouple (603) in the third reaction monitoring module (612) is fitted with the standard exothermic curve in the process parameter database (502), and proactively instructs the third adaptive temperature control module (107c) and the fifth metering pump (401e) to form a closed-loop control, and adjust the acyl chloride drop acceleration rate and cooling intensity in real time to keep the reaction temperature precisely at 0℃±1℃; S4: The reaction mixture obtained in step S3 is transported to the purification system; the central control unit (501) controls the program cooling module (305) of the recrystallization device (301) to crystallize with a non-linear, adaptive cooling curve based on the crystal particle size and morphology data from the in-situ particle characteristic monitoring module (613), and obtains Freranar crystals with the target particle size distribution. S5: Dissolve the Frelanar crystals obtained in step S4 with an alcohol solvent at 25~40℃ according to the formula ratio to obtain a Frelanar solution; The central control unit (501) is also connected to a big data analysis and self-learning unit (503) for optimizing the process parameter database (502) after each batch.

2. The process for preparing the fluorellaranoate solution according to claim 1, characterized in that: The first reaction monitoring module (610) is configured to: acquire a spectrum every 30 seconds based on the built-in spectral acquisition and preprocessing unit (610c); the central control unit (501) uses partial least squares method to calculate the remaining concentration of key raw materials within 5 seconds; when three consecutive data points show that the main product generation rate is lower than 10% of the predicted value, a step-by-step temperature increase of 1~3℃ correction program is triggered, and the by-product spectral signal is monitored simultaneously to determine whether to maintain the new temperature.

3. The process for preparing the fluorellaranoate solution according to claim 1, characterized in that: The control logic of the second reaction monitoring module (611) is as follows: the central control unit (501) cross-verifies the progress of the hydrolysis reaction by integrating and analyzing the changing trends of pH value and conductivity; when the pH value is stable within the set range and the rate of change of conductivity approaches zero, it determines that the hydrolysis reaction is complete and automatically closes the alkaline feed valve.

4. The process for preparing a fluorellaranoate solution according to claim 1, characterized in that: The execution logic of the third reaction monitoring module (612) in the amidation reaction is as follows: the central control unit (501) predicts the intensity of the reaction 100-300 milliseconds in advance through the real-time exothermic rate curve, and reduces the feeding rate or enhances cooling in advance. Its control algorithm adopts fuzzy PID control with online parameter self-tuning.

5. The process for preparing a fluorellaranoate solution according to claim 1, characterized in that: The in-situ particle characteristic monitoring module (613) and the programmed cooling module (305) form a closed-loop control; when the average particle size of the crystal reaches the lower limit of the target range, the cooling rate is automatically increased by 0.2℃ / minute; when the average particle size of the crystal approaches the upper limit of the target range, the cooling rate is automatically reduced by 0.1℃ / minute.

6. The process for preparing a fluorellaranoate solution according to claim 1, characterized in that: The process analysis module (600) has a built-in process pre-configuration unit (505). Before each batch of production starts, the process pre-configuration unit (505) optimizes the initial process parameters by simulating the entire synthesis process based on the input raw material attribute code and initial process parameters, and preloads the optimized parameter set into the central control unit (501) and the process parameter database (502).

7. The process for preparing a fluorellaranoside solution according to claim 1, characterized in that: The process analysis module (600) has a built-in equipment status monitoring unit (504) for real-time monitoring of the flow accuracy of all metering pumps and the torque of the agitator; the central control unit (501) automatically performs flow compensation when it predicts that the accuracy of any metering pump will drift beyond the set value, or issues a maintenance warning in advance when the torque of any agitator increases abnormally, based on the monitoring data.

8. A fluorellaranan solution preparation system for use in the synthesis of a fluorellaranan solution according to any one of claims 1-7, characterized in that, It includes a substitution cyclization unit (100), a hydrolysis unit (200), an amidation unit (300), and a purification unit (400) connected sequentially along the material flow direction, as well as a control module (500) that controls the operation of the entire system. The substitution cyclization unit (100) includes an addition cyclization reaction vessel (101), which is equipped with a first adaptive temperature control module (107a), a first stirrer (110a), a first metering pump (401a) for feeding methyl 2-methyl-4-cyanobenzoate, a second metering pump (401b) for feeding 1,3-dichloro-5-(1-trifluoromethylvinyl)benzene, and a first reaction monitoring module (610) for real-time monitoring of the cyclization reaction process. The hydrolysis unit (200) includes a hydrolysis reaction vessel (102), which is equipped with a second adaptive temperature control module (107b), a second stirrer (110b), a first delivery pump (403a) and a first delivery valve (402a) for receiving materials from the addition cyclization reaction vessel (101), a third metering pump (401c) for adding alkali solution, and a second reaction monitoring module (611) for monitoring the degree of hydrolysis, which includes an online pH sensor (611a) and an online conductivity meter (611b). The amidation unit (300) includes an amidation reaction vessel (103), which is equipped with a third adaptive temperature control module (107c), a third stirrer (110c), a second delivery pump (403b) and a second delivery valve (402b) for receiving materials from the hydrolysis reaction vessel (102), a fourth metering pump (401d) for adding thionyl chloride, a fifth metering pump (401e) for adding amine compounds, and a third reaction monitoring module (612) for monitoring the exothermic reaction. The purification and refining unit (400) includes an extraction settling tank (201), a vacuum concentration device (302), and a recrystallization device (301) connected in sequence; the recrystallization device (301) is equipped with a programmed cooling module (305) and an in-situ particle characteristic monitoring module (613) for monitoring crystal growth; the extraction settling tank (201) is connected to the outlet of the amidation reaction tank (103) through a third delivery pump (403c) and a third delivery valve (402c); The control module (500) is electrically connected to all metering pumps, delivery pumps, switching valves, adaptive temperature control modules and reaction monitoring modules in the system, and coordinates and controls the operation of each unit according to the preset synthesis process program and real-time feedback monitoring data.

9. The freranilar solution preparation system according to claim 8, characterized in that: The first reaction monitoring module (610) is a near-infrared spectral probe. The spectral data it monitors is transmitted to the control module (500). The control module (500) has a built-in chemometric algorithm for calculating the concentration of reactants based on the spectral data. Based on the calculation results, it dynamically adjusts the temperature setpoint of the first adaptive temperature control module (107a) to achieve adaptive control of the cyclization reaction.

10. The freranilar solution preparation system according to claim 8, characterized in that: The monitoring signals from the online pH sensor (611a) and online conductivity meter (611b) of the second reaction monitoring module (611) are transmitted to the control module (500). The control module (500) controls the alkaline addition rate of the third metering pump (401c) in conjunction with the decreasing trend of pH value and the rate of change of conductivity, so as to achieve precise control of hydrolysis reaction and endpoint determination.

Images