Intelligent controlled-release nano-drug carrier preparation device and method

Abstract

The application belongs to the technical field of drug-loaded materials, and discloses a smart controlled-release nano drug carrier preparation device and method, which comprises the following steps: initialization and process parameter setting are performed by a smart control center; raw materials are pretreated by a raw material pretreatment unit; smart carrier directional assembly is realized by a multi-modal dynamic regulation and control synthesis system; online characterization and quality control are performed by an online characterization and quality control unit; post-treatment is performed by a post-treatment integrated unit; images are collected by an image collection module during the preparation process of the nano drug carrier; image analysis and identification are performed by an image analysis module; data analysis is performed by a data analysis module; comprehensive analysis is performed by a comprehensive analysis module; and potential risks are predicted. The raw material pretreatment unit provides high-quality raw materials for subsequent synthesis through separate channels; the multi-modal dynamic regulation and control synthesis system meets the directional synthesis requirements of different types of carriers through modularization and controllable design, and improves the synthesis stability; online quality control is real-time and comprehensive, and ensures that the core indicators of the carrier are qualified.

Description

Technical Field

[0001] This application relates to the field of drug delivery materials technology, and more specifically, to a device and method for preparing intelligent controlled-release nanomedicine carriers. Background Technology

[0002] With the rapid development of interdisciplinary research in materials chemistry and biochemistry, biocompatible amphiphilic polymer nanoparticles containing small-molecule anticancer drugs or frequently used to package anticancer drugs in their hydrophobic inner core or inner layer for the preparation of highly efficient drug delivery systems. Studies have shown that nanoscale carriers have strong adsorption and penetration capabilities to tissue mucosa. Polymer-based micelles, vesicles, and other nanostructures can "passively target" to tumor sites through enhanced permeation and retention (EPR) effects within tumor tissue, achieving precise drug release.

[0003] Existing technology publication CN114948900A discloses a method for preparing a thermosensitive, dynamically bond-regulated, in-situ anchored drug-intelligent release nanocarrier. The method includes the following steps: preparation of furan-modified polyethylene glycol hydrophilic chains; preparation of maleimide hydrophobic prodrug molecules; preparation of amphiphilic polymer chains; and preparation of nanoparticles. The nanocarrier obtained by this method is thermosensitive, transforming from a large size to a small size in response to thermal stimulation, thus enhancing its penetration into tumor tissue. The maleimide groups on the surface of the small-sized nanocarrier ensure prolonged drug retention on the tumor cell surface, overcoming the deficiency of insufficient drug concentration near the tumor. Thermal and glutathione stimulation responses enable intelligent and controllable drug release, while chemotherapy-photothermal synergistic therapy and integrated diagnosis and treatment through fluorescence and photoacoustic imaging.

[0004] While the existing technical solutions described above can achieve the relevant beneficial effects through their structure, they still have the following drawbacks: 1. Insufficient targeted pretreatment of raw materials, failing to guarantee the stability of raw material quality; 2. Weak controllability of the synthesis process, unable to achieve directional assembly of the carrier; 3. Lack of online quality control, allowing substandard intermediates to easily flow into subsequent stages; 4. Incomplete monitoring coverage and lack of intelligent early warning capabilities. They also lack risk prediction functions, only passively alarming after an anomaly occurs, unable to predict potential risks based on historical data, easily leading to production interruptions or material scrapping.

[0005] In view of this, we propose an intelligent controlled-release nanomedicine carrier preparation device and method. Summary of the Invention

[0006] 1. The technical problems to be solved.

[0007] The purpose of this application is to provide an intelligent controlled-release nanomedicine carrier preparation device and method, which solves the technical problems mentioned in the background art above. It realizes that the raw material pretreatment unit provides high-quality raw materials for subsequent synthesis through a specialized channel design; the multimodal dynamic control synthesis system meets the directional synthesis needs of different types of carriers through modular and controllable design, and improves the synthesis stability; and the online quality control is real-time and comprehensive, ensuring that the core indicators of the carrier are qualified.

[0008] 2. Technical solution.

[0009] This application provides a smart controlled-release nanomedicine carrier preparation device, including...

[0010] Raw material pretreatment unit: pretreatment of raw materials, including lipid raw material pretreatment channel, polymer raw material pretreatment channel and inorganic nano-raw material pretreatment channel.

[0011] Multimodal dynamic control synthesis system: realizes the directional assembly of intelligent carriers, including a precision fluid delivery module, a switchable multimodal hybrid chip assembly, a dynamic pressure feedback regulating valve, and an in-situ modification reaction chamber.

[0012] Online characterization and quality control unit: includes dynamic light scattering (DLS) particle size detection module, zeta potential detection module, and UV-Vis spectrophotometric drug loading rate detection module.

[0013] The integrated post-processing unit includes a staged ultrafiltration module, an intermediate storage tank, a vacuum freeze-drying module, and a finished product collection module. These modules are connected via quick-release piping.

[0014] Image acquisition module: Includes a high-definition camera and LED lights to acquire high-definition images of the nanomedicine carrier preparation process.

[0015] Image analysis module: Analyzes and identifies the acquired images to promptly detect anomalies.

[0016] Data analysis module: Analyzes the data collected by various sensors to promptly detect anomalies.

[0017] Comprehensive Analysis Module: This module integrates the monitoring results from the image analysis module and the data analysis module, combines them with historical data, and performs comprehensive analysis to accurately identify anomalies and predict potential risks.

[0018] Alarm module: Includes an alarm that promptly issues an alert when abnormal situations or potential risks are detected.

[0019] Intelligent control center: It is network connected to the raw material pretreatment unit, multimodal dynamic regulation and synthesis system, online characterization and quality control unit, post-processing integrated unit, image acquisition module, image analysis module, data analysis module, comprehensive analysis module and alarm module.

[0020] This invention provides a method for preparing intelligent controlled-release nanomedicine carriers, comprising the following steps.

[0021] S1. The intelligent control center performs initialization and sets process parameters.

[0022] S2, The raw material pretreatment unit pretreatments the raw materials.

[0023] S21, Lipid Pretreatment Channel: Lipid pretreatment is performed using temperature-controlled stirring and ultrasonic dispersion to ensure stable lipid activity and concentration.

[0024] S22, Polymer raw material pretreatment channel for polymer pretreatment. Wet grinding and two-stage filtration ensure uniform polymer particle size and absence of impurities.

[0025] S23, Inorganic Nanomaterial Pretreatment Channel: Inorganic nanomaterial pretreatment is performed. High-pressure homogenization and degassing are used to achieve uniform dispersion of inorganic nanoparticles without bubbles.

[0026] S3. The multimodal dynamic control synthesis system enables the directional assembly of intelligent carriers and performs multimodal synthesis.

[0027] S31, the precision fluid delivery module accurately delivers materials; it simultaneously delivers pre-treated raw materials and modifying reagents.

[0028] S32. Switchable multimodal mixing chip assembly for multimodal mixing. Select impingement jet / vortex / laminar mixing mode according to carrier type.

[0029] S33, the dynamic pressure feedback regulating valve regulates the pressure in real time to avoid particle size fluctuations.

[0030] S34. In-situ modification is performed in the reaction chamber. This completes the modification of the support and prevents its sedimentation.

[0031] S4. The online characterization and quality control unit performs online characterization and quality control. The dynamic light scattering (DLS) particle size detection module detects particle size to ensure the carrier particle size meets standards. The zeta potential detection module monitors the zeta potential of the carrier to ensure stability. The UV-Vis spectrophotometric drug loading rate detection module detects the drug loading rate and calculates the carrier encapsulation efficiency and free drug concentration.

[0032] S5, the post-processing integrated unit performs post-processing.

[0033] S51. The nanomedicine carrier that has passed online characterization and quality control is introduced into the fractional ultrafiltration module through a quick-release tubing; the fractional ultrafiltration module performs fractional ultrafiltration purification to remove impurities and concentrate the carrier.

[0034] S52. The ultrafiltration carrier liquid flows into the intermediate storage tank through a quick-release pipeline for intermediate storage.

[0035] S53. The carrier liquid in the intermediate storage tank is transported to the freeze-drying chamber of the vacuum freeze-drying module through a quick-release pipeline for vacuum freeze-drying and shaping to remove moisture and extend storage life.

[0036] S54. The freeze-dried solid nanomedicine carrier is introduced into the finished product collection module through a quick-release tubing (with a sterile quick connector) for aseptic dispensing and quantitative collection of the finished product.

[0037] S6. The image acquisition module acquires high-definition images of the nanomedicine carrier preparation process.

[0038] S7, the image analysis module analyzes and identifies the acquired images, promptly detecting any abnormalities.

[0039] S8, the data analysis module analyzes the data collected by each sensor and promptly detects anomalies.

[0040] S9. The comprehensive analysis module integrates the monitoring results of the image analysis module and the data analysis module, combines them with historical data, performs comprehensive analysis, accurately identifies abnormal situations, and predicts potential risks.

[0041] S10. When an abnormal situation or potential risk is detected, the alarm module will issue an alarm in a timely manner.

[0042] 3. Beneficial effects.

[0043] One or more technical solutions provided in this application have at least the following technical effects or advantages.

[0044] 1. The raw material pretreatment unit of this invention solves the pretreatment pain points of different types of raw materials through a specialized channel design, providing high-quality raw materials for subsequent synthesis.

[0045] 2. The synthesis process is highly controllable, enabling targeted assembly of the vector. The multimodal dynamic control synthesis system, through modular and controllable design, meets the targeted synthesis requirements of different types of vectors and improves synthesis stability.

[0046] 3. Real-time and comprehensive online quality control ensures that the core indicators of the carrier are qualified; the online characterization quality control unit realizes lossless, real-time, and closed-loop adjustment and control to prevent unqualified intermediates from flowing into subsequent stages.

[0047] 4. The integrated post-processing unit ensures the quality and storage stability of the finished product through efficient purification, precise freeze-drying, and aseptic packaging.

[0048] 5. Full-process monitoring and intelligent early warning reduce production risks; image analysis and data analysis modules identify anomalies in real time, and the comprehensive analysis module combines historical data to predict potential risks.

[0049] 6. The intelligent control center achieves production automation, traceability, and compliance through model prediction, visual monitoring, and data management. Attached Figure Description

[0050] Figure 1 This is a schematic diagram of the double-layer spiral stirring paddle mechanism of the intelligent controlled-release nanomedicine carrier preparation device disclosed in a preferred embodiment of this application.

[0051] Figure 2 This is a schematic diagram of the lower spiral stirrer, upper spiral stirrer, and transmission mechanism of the intelligent controlled-release nanomedicine carrier preparation device disclosed in a preferred embodiment of this application. Figure 3 This is a schematic diagram of a temperature-controlled stirring and dissolving tank of a smart controlled-release nanomedicine carrier preparation device disclosed in a preferred embodiment of this application.

[0052] Reference numerals in the attached diagram: 1. Temperature-controlled stirring and dissolving tank; 2. Sealing cover; 3. Drive motor; 4. Electric push rod; 5. Feed pump; 6. Feed pipe; 7. Lower spiral stirring paddle; 8. Upper spiral stirring paddle; 9. Transmission mechanism; 91. Driving bevel gear; 92. Driven bevel gear; 93. Transmission bevel gear; 11. Tank wall guide plate; 12. Shaft; 13. Ultrasonic probe. Detailed Implementation

[0053] The present application will be further described in detail below with reference to the accompanying drawings.

[0054] This application provides an intelligent controlled-release nanomedicine carrier preparation device, including...

[0055] Raw material pretreatment unit: This unit pretreats the raw materials, including channels for lipid raw material pretreatment, polymer raw material pretreatment, and inorganic nanomaterial pretreatment. These channels are all connected to the multimodal dynamic control synthesis system via pipelines.

[0056] A multimodal dynamic control synthesis system enables the directional assembly of intelligent carriers. It comprises a precision fluid delivery module, a switchable multimodal hybrid chip assembly, a dynamic pressure feedback regulating valve, and an in-situ modification reaction chamber. These components are sequentially connected via piping. Flow rate, pressure, and temperature sensors are installed on the piping.

[0057] Online characterization and quality control unit: includes dynamic light scattering (DLS) particle size detection module, zeta potential detection module, and UV-Vis spectrophotometric drug loading rate detection module.

[0058] The integrated post-processing unit includes a staged ultrafiltration module, an intermediate storage tank, a vacuum freeze-drying module, and a finished product collection module. These modules are connected via quick-release piping. It integrates weighing, level, and aseptic sensors, with data fed into an intelligent control center.

[0059] Image acquisition module: Includes a high-definition camera and LED lights to acquire high-definition images of the nanomedicine carrier preparation process. High-definition cameras are installed above the hybrid chip, in-situ modification reaction chamber, ultrafiltration membrane assembly, freeze-drying chamber, and pipeline connector area to achieve full coverage acquisition of high-definition images of key processes.

[0060] Image analysis module: Analyzes and identifies the acquired images to promptly detect anomalies.

[0061] Data analysis module: Analyzes the data collected by various sensors to promptly detect anomalies.

[0062] Comprehensive Analysis Module: This module integrates the monitoring results from the image analysis module and the data analysis module, combines them with historical data, and performs comprehensive analysis to accurately identify anomalies and predict potential risks.

[0063] Alarm module: Includes an alarm that promptly issues an alert when abnormal situations or potential risks are detected.

[0064] Intelligent Control Center: This center is networked with the raw material pretreatment unit, multimodal dynamic control synthesis system, online characterization and quality control unit, integrated post-processing unit, image acquisition module, image analysis module, data analysis module, comprehensive analysis module, and alarm module. It includes an industrial-grade PLC, touchscreen, and data storage module. The industrial-grade PLC uses a Siemens S7-1200, the touchscreen is 15 inches, and the data storage module is a 1.0TB hard drive supporting USB export. The industrial-grade PLC has a built-in LSTM neural network model that predicts carrier particle size and encapsulation efficiency based on historical synthesis data. After inputting the carrier type and targeting requirements, it automatically generates initial process parameters. The touchscreen supports visual process monitoring, displaying real-time pressure, temperature, and flow rate curves for each module, and has access control and audit tracking functions.

[0065] Furthermore, the lipid raw material pretreatment channel includes a temperature-controlled stirring dissolving tank and an ultrasonic disperser; the ultrasonic disperser is equipped with an ultrasonic probe, which is a titanium alloy probe with adjustable power from 0-300W. The ultrasonic disperser is installed on the temperature-controlled stirring dissolving tank and is connected to the multimodal dynamic control synthesis system through pipelines.

[0066] The temperature-controlled stirring dissolving tank is equipped with a double-layer spiral stirring paddle mechanism, which, together with the tank wall guide plates (four pieces, tilted at 15°), forms an upward and downward convection, avoids lipid aggregation, and improves the dissolution uniformity by 40%.

[0067] The ultrasonic probe uses a lifting mounting bracket (driven by an electric push rod), which can automatically adjust the insertion depth according to the liquid level to avoid cavitation damage to lipid activity. Temperature control accuracy is ±0.5℃ (jacket with circulating water bath, temperature control range 5-60℃).

[0068] The temperature-controlled stirring and dissolving tank integrates a temperature sensor. When the temperature exceeds the threshold due to ultrasonic heat generation, it automatically reduces the ultrasonic power by 10%-20% to prevent lipid oxidation and degradation, thus increasing the activity retention rate to over 95%. The outlet of the temperature-controlled stirring and dissolving tank is equipped with a near-infrared spectroscopy probe (wavelength 900-1700nm) to monitor lipid concentration in real time. When the concentration falls below the set value, the PLC automatically starts the feed pump to replenish lipid powder, ensuring stable raw material concentration for subsequent synthesis.

[0069] Furthermore, the polymer raw material pretreatment channel includes a wet grinding chamber and a high-pressure delivery pump, with the high-pressure delivery pump being a plunger-type high-pressure delivery pump. The high-pressure delivery pump is fixedly installed on the wet grinding chamber, and its outlet is connected to the multimodal dynamic control synthesis system via a pipeline.

[0070] The wet grinding chamber is equipped with a removable grinding bead filter with a pore size of 0.2mm and made of zirconium oxide. The filter is fixed by a buckle, making it easy to replace grinding beads of different sizes to meet the grinding needs of different polymers such as PLGA and polyacrylic acid.

[0071] An online impurity removal filter with a 0.22μm PTFE membrane and a quick-release filter housing is connected in series at the outlet of the wet grinding chamber to remove incompletely ground polymer particles and prevent subsequent clogging of the synthesis channel.

[0072] The wet grinding chamber supports quick replacement of grinding beads with different particle sizes from 0.1 to 0.5 mm: 100-200 μm beads are suitable for PLGA (low glass transition temperature, requires gentle grinding), and 300-500 μm beads are suitable for polyacrylic acid (high hardness, requires strong shearing). The particle size uniformity CV value is ≤10%.

[0073] The wet grinding chamber outlet features a two-stage filtration system: a first-stage 0.22μm PTFE quick-release filter housing to remove unground particles, and a second-stage 0.1μm ceramic membrane filter to trap micro-abrasive particles from the grinding beads, achieving a filtration efficiency of ≥99.9% and completely solving the problem of clogging in the synthesis channel.

[0074] Furthermore, the inorganic nanomaterial pretreatment channel includes a high-pressure homogenizer and a degassing tank, with the vacuum degree of the degassing tank being 0-0.09 MPa.

[0075] The homogenization chamber of the high-pressure homogenizer uses a counter-impact diamond valve seat, and the valve seat spacing is adjusted by a servo motor, which can dynamically adjust the shear strength according to the type of raw material.

[0076] The degassing tank is equipped with an umbrella-shaped distributor (0.5mm aperture) and multi-layer wire mesh packing made of 316L stainless steel with a mesh size of 200. After the raw material is dispersed into a thin film by the distributor, it is further broken into microdroplets by the wire mesh packing. With the help of a 7.5kW vacuum pump, the residual amount of air bubbles is <0.1mL / L, which avoids the formation of hollow carriers during synthesis.

[0077] The high-pressure homogenizer outlet is equipped with a dynamic light scattering (DLS) probe to monitor particle size distribution in real time. If PDI > 0.2 (target ≤ 0.15), the PLC automatically reduces the valve seat spacing by 0.005-0.01mm, forming a homogenization-detection-adjustment closed loop to ensure the uniformity of inorganic nanoparticle dispersion.

[0078] Furthermore, the precision fluid delivery module includes four servo-driven pumps, a non-contact flow meter, and anti-bubble tubing. Three of the servo-driven pumps are connected to the outlets of the lipid raw material pretreatment channel, the polymer raw material pretreatment channel, and the inorganic nano-raw material pretreatment channel, respectively. The fourth pump is connected to the modifier storage tank and is used to deliver modification reagents such as targeted antibodies and responsive molecules, achieving synchronous and precise delivery of basic raw materials and modifiers.

[0079] The pump body of the servo-driven pump adopts a linear guide rail double guide rod guiding mechanism, which, together with the servo motor, improves the flow rate stability compared to the single guide rod design.

[0080] The non-contact flow meter uploads flow rate data to the PLC in real time via RS485 communication. When the flow rate deviation of a single channel exceeds 1%, the PLC immediately adjusts the motor speed of the corresponding pump. At the same time, it integrates a pipeline pressure sensor with a range of 0-5MPa. If the pressure suddenly rises to >0.5MPa, the pump group will automatically stop and trigger an early warning to avoid overload damage to the pump body.

[0081] The anti-bubble pipeline uses a spiral venting PTFE pipe, with a miniature venting valve at the highest point of the pipeline. The automatic venting pressure is 0.1MPa. Combined with the bubble sensor at the inlet of the servo-driven pump, the bubble detection rate is ≥99%, preventing bubbles from entering the mixing chip and causing uneven mixing.

[0082] Supports five-pump upgrades. The module has a reserved installation position for one pump, and a fifth pump can be added according to process requirements to deliver additional components such as buffer solutions and solubilizers, or to achieve independent and precise delivery of two modifiers, such as delivering targeted antibodies and fluorescent labels simultaneously, thereby improving process flexibility.

[0083] Furthermore, the switchable multimodal hybrid chip assembly includes a hybrid chip and a chip mount.

[0084] The hybrid chip is a pluggable hybrid chip, including multiple impingement jet type, vortex type, and laminar flow type hybrid chips, all made of PMMA material, with channel sizes ranging from 100-500μm. The chip holder is made of aluminum alloy and features a locating pin and O-ring seal. The hybrid chips use quick-release snap-fit ​​connections, with a press-type locking connection preferred. Different chips are compatible with different carriers.

[0085] Impact jet chip: dual channels with an included angle of 60°, inlet flow rate of 5-50 mL / min, suitable for highly homogeneous carriers such as LNP, with particle size deviation <5%.

[0086] Eddy current chip: It has three sets of spiral grooves with a depth of 50μm and a pitch of 100μm to generate eddy current mixing and is suitable for stepwise assembly of polymer micelles.

[0087] The chip holder has a built-in temperature control module with a ceramic heating element, and the temperature control range is 10-80℃. The reaction temperature is regulated by heating the chip channel to meet the synthesis requirements of temperature-responsive carriers, such as temperature-sensitive PLGA nanoparticles.

[0088] The mounting base has a built-in miniature pressure sensor and a platinum resistance temperature sensor to directly collect the pressure difference between the chip inlet and outlet and the channel temperature. If the pressure difference is >2.0MPa (chip blockage) or the temperature deviates from the target value by >2℃, the PLC will be immediately fed back to switch to a backup chip, thus shortening downtime for maintenance.

[0089] Furthermore, the switchable multimodal hybrid chip assembly also includes a switching block and an electric actuator.

[0090] The switching block is made of 6061 aluminum alloy with an anodized surface. It serves as the chip carrier and movement platform, featuring three chip slots on its top, spaced to match the chip mounting interface. This allows for the simultaneous mounting of hybrid chips of impact jet, vortex, and laminar flow types. O-rings are embedded in the slots for a sealed fit with the bottom of the hybrid chip. Guide rails on both sides of the switching block mate with the guide rails of the chip mounting base, ensuring no misalignment during movement. The electric actuator is a ball screw type.

[0091] In this technical solution, the electric actuator is driven by the pulse signal output by the PLC and is linked with the miniature pressure sensor and temperature sensor of the chip holder. Before switching, the PLC first controls the pipeline to depressurize. After the switching block moves to the target chip position, the sensor detects the fit and sealing between the chip and the holder. After confirming that there is no leakage, the normal working pressure is restored to avoid material leakage or chip damage during the switching process.

[0092] Furthermore, the dynamic pressure feedback regulating valve includes a piezoelectric ceramic driven valve and a pressure sensor; the valve core material of the piezoelectric ceramic driven valve is zirconium oxide.

[0093] The piezoelectric ceramic driven valve employs a proportional control design. The piezoelectric ceramic controls the valve core displacement via a 0-10.0V voltage, thereby adjusting the mixing chamber pressure in real time. Using PID and feedforward compensation algorithms, the valve opening is adjusted in advance when a change in flow rate is detected, keeping pressure fluctuations within ±0.05MPa. A pressure sensor is installed at the mixing chip outlet. When a sudden pressure increase is detected (e.g., channel blockage), the valve automatically opens wider to release pressure; when a sudden pressure drop is detected (e.g., insufficient raw materials), the valve closes slightly to maintain pressure, ensuring stable mixing intensity and preventing particle size fluctuations.

[0094] Furthermore, the in-situ modified reaction chamber includes a reaction chamber, a multi-channel reagent injector, and an ultrasonic aid.

[0095] The reaction chamber is made of 316L stainless steel. The chamber is equipped with three spiral baffles, ensuring that the raw material makes full contact with the injected modifying reagent as it flows along the baffles, resulting in a modification efficiency >90%.

[0096] The number of ultrasonic auxiliary devices is four. The probes of the ultrasonic auxiliary devices are installed around the outer wall of the reaction chamber to generate low-frequency ultrasound (20-50kHz) to promote the modification reaction, avoid excessive local concentration, and prevent carrier sedimentation.

[0097] The reaction chamber outlet is equipped with a UV-Vis fiber optic probe to monitor the residual concentration of modifying reagents in real time (e.g., the characteristic absorption of Her2 antibody at 280 nm). If the residual concentration is >5 μg / mL, it indicates incomplete modification. The PLC automatically extends the sonication time or increases the reagent injection volume to ensure complete modification. The inner wall of the reaction chamber is coated with a 5 μm thick PTFE layer deposited by plasma deposition to reduce the adsorption rate between the carrier and the chamber wall, thereby improving the product recovery rate.

[0098] Furthermore, the dynamic light scattering (DLS) particle size detection module includes a semiconductor laser, a photodetector, and a miniature sample cell. The miniature sample cell is made of quartz material, has a volume of 100 μL, and is equipped with a temperature control sleeve.

[0099] The sample cell adopts a flow-through design, with raw materials flowing directly into the sample cell from the synthesis channel and then flowing back to the post-processing unit after detection, resulting in no sample loss.

[0100] The temperature control system uses circulating water to maintain a temperature of 25±0.1℃, preventing temperature changes from affecting particle size detection results.

[0101] Furthermore, the eta potential detection module includes an electrode cell, a high-voltage power supply, and a current sensor. Employing an alternating electric field design to avoid electrode polarization, it can monitor the charge transitions of pH-responsive carriers in real time, such as the potential changing from -20.0 mV to +15.0 mV at pH 5.0.

[0102] The electrode cell is connected in series with the sample cell for particle size detection, sharing the same fluid channel, eliminating the need for additional sampling. The detection sequence is particle size first, then potential, with a detection cycle of less than ten seconds per detection, meeting real-time monitoring requirements. When a deviation of the zeta potential from the target value is detected, the PLC automatically controls the buffer pH adjustment pump in the raw material pretreatment unit to achieve closed-loop potential control, with a potential adjustment accuracy of ±1mV.

[0103] Furthermore, the UV-Vis spectrophotometric drug loading detection module includes an UV light source, a fiber optic probe, and a spectrometer. The fiber optic probe is directly inserted into the synthesis channel, 50 mm from the outlet of the modification reaction chamber, and uses transmission detection without the need for sample splitting. The system has a built-in drug characteristic spectral database and automatically calculates the free drug concentration and encapsulation efficiency. When the encapsulation efficiency is <80%, a raw material ratio adjustment command is triggered.

[0104] Furthermore, the staged ultrafiltration module includes a tangential flow ultrafiltration membrane assembly and a peristaltic pump.

[0105] The tangential flow ultrafiltration membrane module uses a 316L stainless steel shell and regenerated cellulose membrane material, with an adjustable molecular weight cutoff of 10kDa-1000kDa. The tangential flow ultrafiltration membrane module employs a cross-flow design, with a 120° angle between the inlet and outlet. Combined with a built-in micro-stirrer on the membrane surface, operating at 0-300 rpm, it generates a composite flow field of tangential and radial stirring; reducing concentration polarization at the membrane surface and improving purification efficiency by 60%.

[0106] The ultrafiltration membrane is fixed by a quick-tight flange to accommodate the purification needs of different molecular weight carriers, such as a 100kDa membrane for LNP and a 50kDa membrane for polymer micelles.

[0107] Furthermore, the vacuum freeze-drying module includes a freeze-drying chamber, a refrigeration unit, and a vacuum pump.

[0108] The freeze-drying chamber is made of 316L stainless steel with a transparent observation window. The freeze-drying chamber is equipped with multi-layer tray racks, which can be raised and lowered by electric push rods. The trays have a hollow design to increase the heat dissipation area and improve the freeze-drying rate. The refrigeration unit adopts a stepped temperature control program, PLC control, and the cooling rate is adjustable from 0.5-5℃ / min to avoid rapid freezing that causes the carrier to agglomerate. After freeze-drying, the moisture content of the carrier is ≤5%, and the storage life is extended.

[0109] Furthermore, the finished product collection module includes sterile collection bottles and automatic dispensing valves.

[0110] The sterile collection bottle is made of glass, with a volume of 50-200mL, and is equipped with a Luer connector. The automatic dispensing valve is a solenoid valve with a diameter of 3.0mm. The dispensing valve uses a weighing sensor to provide feedback for quantitative dispensing (adjustable from 5.0-200mL), avoiding human error. The collection bottle and the freeze-drying chamber outlet use a sterile quick connector with a silicone sealing ring (sterilizable to 121℃), meeting the sterility requirements of biopharmaceuticals.

[0111] Furthermore, the image analysis module analyzes and identifies the acquired images to promptly detect anomalies, including the following steps.

[0112] 1. Image data reception and synchronization: Multiple sets of high-definition cameras (two sets of hybrid chip flow path, one set of modification reaction chamber, one set of ultrafiltration membrane module, one set of freeze-drying chamber, and one set of pipeline connector area) acquire images in real time and transmit them to the image analysis module in real time via gigabit Ethernet; the module has a built-in timestamp synchronizer to align the image data with the corresponding sensor data with millisecond-level precision to ensure spatiotemporal matching of image and parameters.

[0113] 2. Image preprocessing: including noise reduction and correction: Denoising: Gaussian filtering is used to remove random noise from the image, and median filtering is used to eliminate particle impurities, improving the signal-to-noise ratio to over 30dB.

[0114] Correction: Lens distortion correction is performed using a pre-set checkerboard calibration plate, with radial distortion ≤0.1% and tangential distortion ≤0.05%. Brightness uniformity correction is also performed based on a standard grayscale card to avoid misjudgment caused by differences in LED light source illuminance.

[0115] 3. Region of Interest (ROI) Extraction: Based on the target observed by each camera, the Region of Interest (ROI) is automatically selected: ① Hybrid Chip Flow Path: The impact jet collision point / vortex spiral groove area is selected; ② Modified Reaction Chamber: The ten-millimeter range of the spiral guide plate outlet is selected; ③ Ultrafiltration Membrane Module: The central area of ​​the membrane surface is selected; ④ Pipeline Connector Area: The sealing surface of each quick-release connector is selected to reduce background interference.

[0116] 4. Multi-dimensional feature extraction: Features are extracted by using targeted algorithms in conjunction with the core observation scenarios of the device.

[0117] Hybrid flow pattern characteristics: The flow pattern contour is extracted by Canny edge detection algorithm, and the symmetry of the impinging jet (difference between left and right contours / total width), the flatness of the laminar interface (angle between the interface fitted line and the baseline) and the number of vortices (Hough circle detection) are calculated.

[0118] Carrier particle characteristics: The Otsu threshold segmentation algorithm is used to separate carrier particles from the fluid background, and the particle area, equivalent diameter, and particle number ratio are statistically analyzed.

[0119] Equipment status characteristics: By comparing the location of pipeline joints and the outline of contaminated areas on the membrane surface using a template matching algorithm, the area of ​​leakage liquid diffusion and the proportion of contaminated areas are extracted.

[0120] 5. Anomaly detection and classification: Based on preset thresholds and feature logic, the anomaly type is determined and classified.

[0121] Anomalies in the mixed flow pattern: symmetry of the impinging jet <90%, laminar interface angle >5°, and number of vortices <2.0.

[0122] Abnormal carrier aggregation: The proportion of particles with an equivalent diameter > 500 nm > 5% (normal ≤ 3%), or the area of ​​a single particle > 200 pixels (normal ≤ 50 pixels).

[0123] Abnormal membrane fouling: The area of ​​fouled region on the membrane surface is greater than 10% (normal ≤5%).

[0124] Abnormal pipeline leakage: The liquid diffusion area in the joint area is greater than 100 pixels (normally there is no diffusion), or liquid contour expansion is detected in three consecutive frames of images.

[0125] Abnormal freeze-drying condition: Ice crystal protrusions appear on the surface of the sample inside the freeze-drying chamber. The grayscale difference is >50 as detected by grayscale value mutation.

[0126] 6. Result Output: Anomaly results are generated, outputting the anomaly type, occurrence time, ROI image screenshot, and feature parameter values; and the output is visualized. Early warning levels are matched: Level 1 anomalies, such as flow pattern symmetry 85%-90% → Yellow warning; Level 2 anomalies, such as aggregation rate 5%-8% → Orange warning; Level 3 anomalies, such as pipeline leakage, membrane fouling >15% → Red warning. Anomaly results and warning levels are transmitted in real-time to the comprehensive analysis module via RS485 bus, simultaneously triggering a pop-up display on the touchscreen.

[0127] 7. Historical data storage and model optimization: After each batch of synthesis is completed, abnormal images, feature parameters, and processing results are automatically stored to a 1TB industrial hard drive to form an abnormal case library; the model is iterated weekly, and the newly accumulated abnormal cases are trained using the ResNet50 transfer learning model to update the judgment threshold, and the accuracy of abnormal identification is continuously optimized.

[0128] Furthermore, the data analysis module performs data analysis, including the following steps.

[0129] 1. Data Reception and Classification: Real-time reception of sensor data from each unit, classified and stored according to unit-parameter dimension: ① Raw material pretreatment unit (lipid concentration, grinding bead particle concentration, post-homogenization PDI); ② Multimodal synthesis system (four-way pump flow rate, chip inlet and outlet pressure difference, modification chamber temperature); ③ Online characterization unit (particle size CV value, Zeta potential, drug loading rate); ④ Post-processing unit (ultrafiltration transmembrane pressure difference, freeze-drying chamber vacuum degree, dispensing weight deviation), ensuring traceability of data source.

[0130] 2. Data preprocessing: including denoising and standardization; ① Denoising: Kalman filtering is used to eliminate random noise in dynamic data such as flow rate and pressure, and moving average filtering is used for static data such as temperature and concentration, reducing data fluctuation to within ±0.5%; ② Standardization: Different range parameters are mapped to the 0-1 interval to eliminate the influence of dimensional differences on subsequent analysis.

[0131] 3. Single-parameter anomaly analysis: ① Real-time threshold comparison: Compare the preprocessed data with the preset threshold range. If the flow rate deviation exceeds ±1%, the pressure fluctuation exceeds ±0.1MPa, the particle size CV is greater than 10%, or the drug loading rate is less than 80%, it is directly marked as a single-parameter anomaly; ② Trend slope warning: Calculate the change slope of ten consecutive sets of data. If a gradual anomaly occurs, it is marked as a potential anomaly in advance.

[0132] 4. Multi-parameter correlation verification: For single-parameter abnormal data, cross-validation with correlated parameters is used to eliminate false judgments: ① If the flow rate is abnormal, check whether the corresponding pipeline pressure changes simultaneously. If the flow rate decreases and the pressure increases, it is determined to be a pipeline blockage; if only the flow rate is abnormal, it is a pump problem; ② If the drug loading rate is abnormal, correlate the mixing chip pressure and modification chamber temperature data. If the pressure is normal but the temperature exceeds the tolerance, it is determined that the temperature affects the modification efficiency. The correlation verification accuracy is ≥99%.

[0133] 5. Anomaly Classification and Labeling: Anomalies are classified and labeled according to their impact: ① Level 1 anomaly: minor impact, such as flow rate deviation of 0.5%-1% → label fine-tuning warning; ② Level 2 anomaly: moderate impact, such as particle size CV of 10%-15%, drug loading rate of 75%-80% → label module switching warning; ③ Level 3 anomaly: severe impact, such as sudden pressure increase > 5.0 MPa, drug loading rate < 75% → label emergency shutdown warning.

[0134] 6. Output Results: Generate an anomaly report, including the name of the abnormal parameter, the time of occurrence, the degree of anomaly, related parameter data, and a preliminary determination of the cause; transmit the anomaly report and the classification label to the comprehensive analysis module via RS485 bus, and trigger corresponding pre-actions: Level 1 anomalies trigger automatic adjustment commands, such as adjusting pump speed and compensating for heating power; Level 2 anomalies trigger a backup module switching prompt; Level 3 anomalies trigger an emergency shutdown signal.

[0135] Furthermore, the comprehensive analysis module performs a comprehensive analysis to accurately identify anomalies and predict potential risks. This includes the following steps.

[0136] 1. Multi-source data reception and spatiotemporal alignment: Simultaneously receive real-time results from the image analysis module and the data analysis module, and use the module's built-in timestamp calibrator to align the two types of data along the time-space dimension. At the same time, retrieve matching data from the historical database of similar carriers (such as LNP) to form a real-time + historical dataset.

[0137] 2. Evidence Theory Fusion Analysis: The DS evidence theory is used to fuse the credibility of multi-source data, constructing a two-dimensional evidence body of image features and sensor parameters. Image evidence, such as carrier aggregation, accounted for 6%, with a confidence level of 0.92.

[0138] Sensor evidence, such as an 8W decrease in ultrasonic power, has a confidence level of 0.95.

[0139] The fusion confidence level is calculated using evidence combination rules. For example, the fusion confidence level of clustering + power decrease is 0.98. When the fusion confidence level is ≥0.85, it is judged as a valid anomaly. When the fusion confidence level is <0.7, it is marked as pending verification and a secondary sampling analysis is triggered. The fusion false positive rate is controlled at <0.3%.

[0140] 3. Precise Root Cause Identification: Establish an anomaly-root cause association rule base and combine it with fused data to pinpoint the root cause. By combining visual feature data from the image analysis module and sensor parameter data from the data analysis module, and through a logic of first matching with the rule base, then cross-validating, and finally supplementing with historical data, the root cause can be accurately located, avoiding decision-making biases caused by misjudgment due to single data or rule blind spots. The rule base needs to be built in advance based on equipment principles, process characteristics, and historical failure cases, covering the three-dimensional association between anomaly combinations, root causes, and verification indicators, and supporting dynamic updates.

[0141] 3.1 Root cause determination of a single anomaly: Taking pipeline leakage as an example, the determination process is as follows.

[0142] 3.1.1 Verification of Abnormal Isolation: First, confirm that only the image module captures liquid seepage in the pipe joint area, and that the corresponding pipe pressure, flow rate, temperature and other parameters of the data analysis module are all within the normal threshold. This eliminates root causes that require multiple parameters to support, such as excessive pressure causing pipe rupture or fluid corrosion of the pipe.

[0143] 3.1.2 Direct matching from the rule base: Search the rule base for entries with a single pipeline leak image and no other abnormal parameters. The root cause of this entry is assumed to be aging of the seal. Since the seal is a vulnerable part of the pipeline joint, its elasticity will decrease after long-term use, which will lead to seal failure. Moreover, such failure only manifests as local leakage and will not cause abnormalities in global parameters such as pressure and flow.

[0144] 3.1.3 Secondary Confidence Confirmation: The fusion confidence score of the pipeline leakage output by the image analysis module is 0.96, which is higher than the rule base judgment threshold of 0.9. Combined with the equipment maintenance records, such as the seal having been used continuously for six months, close to the recommended replacement cycle of eight months, the root cause is further verified to be reasonable, and finally it is directly determined to be the aging of the pipeline joint seal.

[0145] 3.2 Root cause determination of compound anomalies: Taking flow pattern asymmetry + 2% deviation of flow velocity on one side of the pump as an example, the determination process is as follows.

[0146] 3.2.1 Anomaly Correlation Analysis: First, confirm the spatiotemporal consistency of the two types of anomalies; the image shows that the flow pattern in the hybrid chip is unilaterally biased, such as the left flow stream being wider and the right flow stream being narrower, and the sensor data shows that the flow rate of the left feed pump is 2% lower than the set value, while the flow rate of the right feed pump is normal. The time difference between the two occurrences is less than ten seconds, eliminating independent anomalies that occur one after the other. It is preliminarily judged that the flow rate deviation is the cause of the flow pattern asymmetry.

[0147] 3.2.2 Root Cause Elimination: Potential root causes of flow pattern asymmetry are retrieved from the rule base, including three categories: servo pump flow calibration drift, hybrid chip blockage, and feed line blockage. These are eliminated one by one using sensor data.

[0148] Eliminate chip blockage: The pressure difference between the chip inlet and outlet is 0.2MPa, and the normal range is 0.15-0.25MPa. If the chip is blocked, the pressure difference will increase sharply, usually >0.3MPa, so this can be eliminated.

[0149] Clear the blockage in the feed line: The pressure in the left feed line is 0.3 MPa, which is the same as that in the right line. If the line is blocked, it will cause the local pressure to rise, so clear the blockage.

[0150] 3.2.3. Accurate matching of rule base: The remaining root cause of servo pump flow calibration drift is completely matched with the abnormal combination of 2% flow velocity deviation on one side pump + flow pattern asymmetry + normal pressure; because after long-term operation of the servo pump, internal gear wear will lead to a decrease in flow output accuracy. The flow velocity deviation on one side will directly lead to an imbalance in the fluid ratio on both sides during mixing, which in turn will cause flow pattern asymmetry. Finally, it is determined that the flow calibration drift of the left servo pump is the cause, and it is recommended to perform batch calibration of all servo pumps at the same time.

[0151] 3.3 Determination of the root cause of new anomalies: Taking ice crystal protrusion and temperature fluctuation in the freeze-drying chamber as examples, the determination process is as follows.

[0152] 3.3.1 Anomaly Feature Extraction: First, quantify the core features of the newly added anomaly; the image shows irregular ice crystal protrusions on the sample surface inside the freeze-drying chamber, with a protrusion height > 3.0 mm, distributed in the edge area of ​​the chamber. Sensor data shows that the temperature inside the freeze-drying chamber fluctuates between -40℃ and -35℃ (normally it should be stable at -40℃±1℃), with a fluctuation period of about five minutes. At the same time, auxiliary information such as the current freeze-drying stage (end of primary drying) and raw material batch are recorded.

[0153] 3.3.2 Historical Data Retrieval and Comparison: Screen cases containing ice crystal anomalies or temperature fluctuations from the historical fault case database, extract the features of each case, such as ice crystal morphology, temperature fluctuation range, and corresponding equipment status, and compare them with the current anomaly features; use feature dimension matching, that is, compare key dimensions such as whether the ice crystal protrusion is at the edge, whether the temperature fluctuation range is >3℃, and whether it is in the late stage of drying.

[0154] 3.3.3 Similarity threshold verification: Set a similarity threshold of ≥80%, and finally select a highly similar case: three months ago, ice crystals protruded on the edge of the freeze-drying chamber and the temperature fluctuated by ±4℃. The root cause was insufficient refrigerant in the refrigeration unit. The similarity between the two reached 85%.

[0155] 3.3.4 Root Cause Inference and Auxiliary Verification: Based on similar cases, the root cause was inferred to be insufficient refrigerant in the refrigeration unit, and verified through additional parameters; the low-pressure side pressure of the refrigeration unit was checked and showed 0.2MPa, while it should normally be ≥0.3MPa. This further confirmed that insufficient refrigerant would lead to unstable cooling capacity, which in turn would cause temperature fluctuations in the compartment. Due to the lower heat exchange efficiency, ice crystals were the first to appear in the edge area. Finally, the root cause was identified as insufficient refrigerant in the refrigeration unit, triggering a maintenance command to replenish the refrigerant.

[0156] 4. LSTM Model Risk Prediction: Call the pre-trained LSTM neural network model to perform two types of predictions. The input features are real-time fused data, equipment runtime, and raw material batch information; the output is the risk probability and remaining time.

[0157] 4.1 Equipment life prediction: For example, based on the curve of the change of servo pump speed fluctuation with running time, the remaining life of the bearing can be predicted, and the spare parts replacement prompt can be triggered in advance.

[0158] 4.2 Process drift prediction: Based on the mixing pressure-particle size correlation trend, it is predicted that the drug loading rate may drop to 78% after three days, which is lower than the threshold of 80%. It is recommended to calibrate the mixing chip temperature in advance.

[0159] 5. Dynamic Risk Level Classification and Response: Risk levels are dynamically updated based on the urgency of the anomaly and the predicted scope of risk impact. Level 1 Risk: Minor anomaly with no subsequent impact, such as flow pattern symmetry of 88% → Yellow alert, only touchscreen notification will be sent.

[0160] Level 2 Risk: Moderate anomaly + localized impact, such as agglomeration rate of 6% + drug loading rate of 79% → Orange alert, triggering automatic adjustment + standby module standby, such as increasing ultrasound power + preparing to switch to backup chip.

[0161] Level 3 Risk: Severe anomaly + global impact, such as pipeline leakage + sudden pressure drop → red alert, triggering emergency shutdown + personnel notification, cutting off the feed pump within ten seconds, and simultaneously pushing the location of maintenance personnel and abnormal images through the APP.

[0162] 6. Results Output and Closed-Loop Recording: Generate a comprehensive analysis report, including abnormal phenomena, root cause determination, risk level, countermeasures, predicted results, and related data screenshots; synchronize the report to the intelligent control center; display the fusion analysis results in real time on the touch screen, annotate the abnormal location and risk countdown on the 3D flowchart, and support operators to retrospectively view data correlation curves, such as the pressure-image feature change curve during abnormal periods.

[0163] Furthermore, the intelligent control center predicts the carrier particle size and encapsulation efficiency based on historical synthetic data. After inputting the carrier type and targeting requirements, it automatically generates initial process parameters (flow rate, pressure, and temperature). This includes the following steps.

[0164] 1. LSTM model deployment and initial process parameter generation.

[0165] 1.1 Historical Data Loading and Model Activation: The pre-trained LSTM neural network model is automatically loaded (model parameters are stored in the PLC's built-in storage area). At the same time, historical synthetic data from nearly 300 batches of similar carriers are retrieved from the data storage module, including raw material parameters, process parameters, and quality indicators. The model is then fine-tuned in real time to avoid model drift and ensure prediction accuracy.

[0166] 1.2 Process Requirements Input and Parameter Generation: Operators input the carrier type (e.g., LNP, PLGA nanoparticles), targeting requirements (e.g., Her2-targeted, untargeted), and core quality targets, such as particle size 100±5nm and drug loading ≥85%, through the process configuration interface on the touchscreen. After receiving the input information, the PLC calls the LSTM model and automatically generates initial process parameters based on the mapping relationship between carrier type, targeting requirements, process parameters, and quality indicators in historical data.

[0167] Raw material pretreatment unit: lipid concentration 8 mg / mL, polymer molecular weight 20 kDa, inorganic nanoparticle dispersibility PDI ≤ 0.2.

[0168] Multimodal synthesis system: four-pump flow rate ratio (lipid phase: polymer phase: inorganic phase: modifier phase = 3:2:1:0.5), chip inlet and outlet pressure difference 0.8 MPa, mixing chamber temperature 35℃, ultrasonic power 200W. The multimodal dynamic control synthesis model is as follows.

[0169] .

[0170] In the formula, Φ is the comprehensive control objective function, and σ is the core indicator for multi-parameter collaborative optimization; v It is the flow velocity fluctuation rate, with a target of ≤0.3%, a core parameter of servo pump PID+feedforward control, reflecting the stability of fluid transport; σ P It refers to the pressure fluctuation rate, with a target of ≤±0.05MPa, which is the dynamic response index of pipeline pressure and is associated with overload protection logic; η mix It refers to the mixing efficiency, with a target of ≥90%; η modify Modification efficiency, with a target of ≥90%, is a performance evaluation indicator for hybrid chips and affects carrier uniformity; ΔT is the temperature deviation, with a target of ≤2. ∘ C, temperature control system accuracy index, threshold for triggering backup chip switching; C resThe residual concentration of the modifying reagent is ≤5μg / mL, which is a UV-Vis spectral detection index to control the completeness of modification. W1, w2, w3, w4, w5 and w6 are the weighting coefficients of flow rate fluctuation rate, pressure fluctuation rate, mixing efficiency, modification efficiency, temperature deviation and residual concentration of modifying reagent, which are dynamically adjusted according to process priority and dynamically adjusted by LSTM timing encoder.

[0171] Parameter coupling relationships and control strategies include.

[0172] Fluid transport-pressure-temperature coupled control.

[0173] .

[0174] e i (t)=v i,set -v i (t). Constraints: When |e i Flow rate regulation is triggered when (t)|>1%; when P≥0.5MPa or ΔP chip Protection logic is activated when the pressure is >2MPa; when |T−T set |>2C Switch to backup chip; where v i (t) is the output variable of the control system, representing the actual flow velocity value of the i-th flow velocity control channel at time t. K is the target physical quantity that the entire control algorithm ultimately aims to regulate and maintain. p It is the proportional coefficient. The function of proportional control is to quickly respond to errors, causing the output to change in the direction of reducing the error. i (t) is the error signal, v i,set It sets the flow rate, v i (t) is the actual flow velocity; K i It is the integral coefficient. K d It is the differential coefficient. K f These are the feedforward coefficients. Feedforward control compensates for known disturbances or changes in the setpoint in advance to improve the system's dynamic response. τ is the integral variable, used to iterate through every time point from 0 to t during the integration process. P actual ρ is the actual pressure, the target physical quantity that the model aims to calculate and predict, representing the true pressure value of the system at a given moment. ρ is the fluid density, representing the mass of the fluid per unit volume. g is the gravitational acceleration, the acceleration caused by the Earth's gravitational pull on an object. h is the height, referring to the vertical distance of the fluid's location relative to a reference horizontal plane. V is the pipe volume, i.e., the internal space volume of the pipe or container containing the fluid. β is the fluid compressibility coefficient, representing the relative degree of volume change of the fluid under pressure. i Flow velocity refers to the speed at which a fluid flows in a pipe at a specific location or in a specific direction. actual This is the actual temperature. Tset α is the set temperature, the temperature value that the system expects to reach or maintain. α is the thermal diffusivity (or thermal conductivity), representing the relative magnitude of a material's ability to conduct and store heat. ∇•q is the heat flux density divergence, ∇ is the gradient operator, and q is the heat flux density vector. ∇•q represents the degree of spatial divergence of the heat flux density.

[0175] Hybrid-modification efficiency synergistic optimization.

[0176] .

[0177] .

[0178] In the formula, η mix ρ is the mixing efficiency, used to measure the effective mixing degree of a fluid during the mixing process. Its value ranges from 0 to 1, with 1 representing complete mixing. k is an empirical constant, related to specific mixing equipment, fluid properties, etc., used to adjust the influence of the Reynolds number on the mixing efficiency. Re is a dimensionless number characterizing the fluid flow state, used to determine whether the fluid is laminar or turbulent. It is dimensionless. ρ is the fluid density, representing the mass of the fluid per unit volume. v is the flow velocity, referring to the speed at which the fluid flows in a pipe or flow channel. h η is the hydraulic diameter, used to calculate the equivalent diameter of a circular pipe for non-circular pipes, simplifying flow analysis. μ is the fluid viscosity, reflecting the fluid's resistance to flow deformation. θ is the impingement jet chip angle; in this model, the mixing efficiency is optimally optimized when θ = 60°. modify η0 is the modification efficiency, representing the effectiveness of modifying the target material under ultrasonic treatment. Its value ranges from 0 to 1, with 1 indicating complete modification. It is dimensionless. η0 is the initial modification efficiency, i.e., the modification efficiency under conditions without ultrasonic treatment or other reference conditions. α is a constant related to ultrasonic power and the modification process, reflecting the influence coefficient of ultrasonic power on the modification efficiency. P ultra is the ultrasonic power, representing the energy output of the ultrasonic equipment. t is the ultrasonic treatment time, i.e., the duration for which the ultrasonic equipment is turned on and acts on the target substance. Δt is the ultrasonic time that needs to be extended, when the residual concentration C... res When the concentration is >5.0 μg / mL, the ultrasound time is automatically extended according to this formula. C res This refers to the residual concentration, specifically the concentration of the unmodified portion of the target substance after ultrasonic modification. C target It is the target concentration, that is, the residual concentration or modified concentration standard that is expected to be achieved.

[0179] Dynamic valve - ultrasonic-assisted linkage control. . Global constraints and security boundaries: In the formula, x is the valve core displacement, representing the distance the valve core moves in the piezoelectric ceramic valve, used to control the flow and flow rate of the fluid. u K is the proportionality coefficient between valve core displacement and voltage, reflecting the degree of influence of voltage on valve core displacement. U is the voltage, the electrical signal voltage applied to the piezoelectric ceramic valve, used to control the valve core displacement. p It is the proportional coefficient, used in PID control to adjust the pressure based on the current pressure error e. P (t) Rapid voltage adjustment. K i It is the integral coefficient, used to accumulate historical pressure errors and eliminate steady-state errors. K d It is the differential coefficient, used to predict the changing trend of pressure error and adjust the voltage in advance to reduce overshoot. P (t) represents the pressure error, i.e., the difference between the set pressure and the actual pressure. τ is the integration variable. K f This is the feedforward coefficient, used to adjust the voltage in advance based on the rate of change of flow velocity to address the impact of flow velocity variations on pressure. σ P It is a pressure fluctuation. In this control system, the pressure fluctuation σ is required to be within a certain range. P ≤±0.05MPa. P ultra It is the ultrasonic power, which is dynamically adjusted based on the mixing efficiency or modification efficiency. P base This is the baseline ultrasonic power, which is the reference ultrasonic power without considering the linkage between mixing efficiency and modification efficiency. ΔP is the adjustment amount of the ultrasonic power, used to increase or decrease the baseline ultrasonic power according to the mixing efficiency or modification efficiency.

[0180] Post-processing unit: ultrafiltration transmembrane pressure difference 0.1MPa, freeze-drying chamber temperature -45℃, freeze-drying time 18h.

[0181] The initial parameters are automatically synchronized to the initial parameter preview interface on the touch screen. Operators can fine-tune them according to the actual batch of raw materials, such as adjusting the lipid concentration to 7.5 mg / mL. After confirmation, click the parameter distribution button, and the PLC will allocate the parameters to the actuators of the corresponding unit modules.

[0182] 2. Real-time data acquisition and dynamic interaction.

[0183] 2.1 Real-time reception and processing of multi-source data: The PLC collects data from each module at a frequency of 100ms / time.

[0184] Sensor data: servo pump flow rate, chip inlet and outlet pressure difference, mixing chamber temperature, ultrafiltration transmembrane pressure difference, freeze-drying chamber vacuum, and online characterization of particle size / CV value / Zeta potential / drug loading rate.

[0185] Image data: The image acquisition module uploads high-definition images of key areas such as the hybrid chip flow path, modified reaction chamber, and ultrafiltration membrane assembly at 1.0 frames / second. The image analysis module simultaneously outputs characteristic parameters, such as flow pattern symmetry, carrier aggregation ratio, and leakage area.

[0186] Equipment status data: the operating status of each execution unit, such as the servo pump running and the piezoelectric valve voltage at 5.2V.

[0187] The PLC preprocesses the collected data: removes obvious outliers, aligns multi-source data by timestamp, ensures that image features match the pressure / flow rate data at the corresponding time, and transmits the data to the touch screen and data analysis module in real time.

[0188] 2.2 Dynamic control of process parameters: The data analysis module analyzes the preprocessed data. If a single parameter is abnormal or multiple parameters are abnormal, the abnormal results and the predicted root causes are transmitted to the PLC. The PLC combines the prediction results of the LSTM model to automatically generate control instructions.

[0189] For minor anomalies, such as a flow rate deviation of 0.8%, directly adjust the motor speed of the corresponding servo pump to correct the flow rate to the target value.

[0190] For moderate anomalies, such as a mixing chamber temperature deviating from the target by 1.5℃: control the power of the ceramic heating element of the chip holder to compensate for the temperature deviation.

[0191] Potential risks, such as the predicted decrease in vacuum level in the freeze-drying chamber after one hour: adjust the vacuum pump speed of the vacuum freeze-drying module in advance to pre-increase the vacuum level.

[0192] During the control process, the PLC feeds back the command execution results to the touch screen in real time, displaying the status flow from control in progress to control completed to parameter stability.

[0193] 3. Visual monitoring and human-computer interaction.

[0194] 3.1 Full-process visualization via touchscreen: The touchscreen uses a 15-inch capacitive screen to display information in the form of an overview page and module pages.

[0195] Overview page: The 3D dynamic flowchart intuitively displays the connection relationship between each unit module, uses different colors to mark the equipment status, and refreshes the values ​​and trend curves of core quality indicators in real time.

[0196] Module-specific page: Click on a unit module, such as the multimodal synthesis system, to enter the details page and view all parameters of that module: real-time flow rate curve of the servo pump, chip pressure difference change trend, mixing chamber temperature fluctuation graph, and real-time images uploaded by the image acquisition module, such as the impact jet status of the mixing chip flow path.

[0197] Trend Analysis Page: Supports selecting any parameters to view historical curves for the past hour / 24 hours / 7 days, and allows you to trace data changes over a specific time period using zoom and pan functions.

[0198] 3.2 Manual Intervention and Operation: Operators perform manual operation via touch screen.

[0199] Parameter modification: In the manual control interface, after entering the permission password, you can manually adjust parameters such as pump speed, pressure, and temperature. After clicking confirm, the PLC will execute immediately.

[0200] Module start / stop: Supports independent start / stop of a single module (such as pausing the raw material pretreatment unit or starting the post-processing unit) or start / stop of the entire process. Start / stop operations require secondary confirmation to avoid misoperation.

[0201] Alarm Handling: When the alarm module triggers a warning (such as a red alarm for pipeline leakage), an alarm pop-up window will appear on the touch screen, displaying the abnormal location, root cause, and suggested measures. The operator can click to confirm the handling. The PLC records the handling time and the operator, and at the same time executes preset safety actions, such as pausing the pump unit and opening the pressure relief valve.

[0202] Reference Figure 1 , Figure 2 and Figure 3 The double-layer spiral stirring paddle mechanism includes a temperature-controlled stirring and dissolving tank 1, a sealing cover 2, a drive motor 3, a lower spiral stirring paddle 7, an upper spiral stirring paddle 8, and a transmission mechanism 9.

[0203] A sealing cover 2 is detachably fixed at the upper end of the temperature-controlled stirring dissolving tank 1. A drive motor 3, an electric push rod 4, and a feed pump 5 are fixedly installed on the sealing cover 2. A feed pipe 6 is fixedly installed at the output end of the feed pump 5. The lower end of the feed pipe 6 extends below the sealing cover 2 and into the temperature-controlled stirring dissolving tank 1.

[0204] A shaft 12 is rotatably mounted below the sealing cover 2, and the output end of the drive motor 3 is coaxially and fixedly connected to the shaft 12.

[0205] A lower helical impeller 7 is detachably fixed to the lower end of shaft 12; an upper helical impeller 8 is rotatably mounted on shaft 12; a transmission mechanism 9 is located below the sealing cover 2, and the transmission mechanism 9 is connected to shaft 12 and upper helical impeller 8. Multiple helical impeller blades are fixedly mounted on both the lower helical impeller 7 and the upper helical impeller 8. An ultrasonic probe 13 is fixedly mounted at the lower end of the movable rod of the electric push rod 4.

[0206] Multiple tank wall guide plates 11 are evenly distributed around the inner circumference of the temperature-controlled stirring melting tank 1.

[0207] In this technical solution, the drive motor 3 drives the lower helical agitator 7 to rotate via the shaft 12, and the shaft 12 drives the upper helical agitator 8 to rotate via the transmission mechanism 9. The lower helical agitator 7 and the upper helical agitator 8 rotate in opposite directions.

[0208] Furthermore, the transmission mechanism 9 includes a driving bevel gear 91, a driven bevel gear 92, and a transmission bevel gear 93. The driving bevel gear 91 is coaxially fixedly mounted with the shaft 12; the driven bevel gear 92 is fixedly connected to the upper spiral agitator 8 and rotatably connected to the shaft 12; the transmission bevel gear 93 is rotatably mounted on the sealing cover 2, and the transmission bevel gear 93 meshes with both the driving bevel gear 91 and the driven bevel gear 92 for transmission.

[0209] In this technical solution, when the shaft 12 rotates, it drives the active bevel gear 91 to rotate. The active bevel gear 91 drives the transmission bevel gear 93, the transmission bevel gear 93 drives the driven bevel gear 92 to rotate, and the driven bevel gear 92 drives the upper spiral agitator 8 to rotate. The lower spiral agitator 7 and the upper spiral agitator 8 rotate in opposite directions.

[0210] This invention provides a method for preparing intelligent controlled-release nanomedicine carriers, comprising the following steps.

[0211] S1. The intelligent control center initializes and sets process parameters. Power is turned on, and the system automatically checks the communication connections of each module and the zero points of the sensors. The touchscreen displays progress, and an alarm pops up for any abnormalities until the self-test is passed. The LSTM model is loaded and fine-tuned. The operator inputs the carrier type, target requirements, and quality objectives. The PLC automatically generates initial process parameters for raw material pretreatment, synthesis, and post-treatment, and synchronizes them to each module.

[0212] S2, The raw material pretreatment unit pretreatments the raw materials.

[0213] S21, Lipid Pretreatment Channel: Lipid pretreatment is performed using temperature-controlled stirring and ultrasonic dispersion to ensure stable lipid activity and concentration.

[0214] S22, Polymer raw material pretreatment channel for polymer pretreatment. Wet grinding and two-stage filtration ensure uniform polymer particle size and absence of impurities.

[0215] S23, Inorganic Nanomaterial Pretreatment Channel: Inorganic nanomaterial pretreatment is performed. High-pressure homogenization and degassing are used to achieve uniform dispersion of inorganic nanoparticles without bubbles.

[0216] S3. The multimodal dynamic control synthesis system enables the directional assembly of intelligent carriers and performs multimodal synthesis.

[0217] S31, the precision fluid delivery module accurately delivers materials; it simultaneously delivers pre-treated raw materials and modifying reagents.

[0218] S32. Switchable multimodal mixing chip assembly for multimodal mixing. Select impingement jet / vortex / laminar mixing mode according to carrier type.

[0219] S33, the dynamic pressure feedback regulating valve regulates the pressure in real time to avoid particle size fluctuations.

[0220] S34. In-situ modification is performed in the reaction chamber. This completes the modification of the support and prevents its sedimentation.

[0221] S4. The online characterization and quality control unit performs online characterization and quality control. The dynamic light scattering (DLS) particle size detection module detects particle size to ensure the carrier particle size meets standards. The zeta potential detection module monitors the zeta potential of the carrier to ensure stability. The UV-Vis spectrophotometric drug loading rate detection module detects the drug loading rate and calculates the carrier encapsulation efficiency and free drug concentration.

[0222] S5, the post-processing integrated unit performs post-processing.

[0223] S51. The nanomedicine carrier that has passed online characterization and quality control is introduced into the fractional ultrafiltration module through a quick-release tubing; the fractional ultrafiltration module performs fractional ultrafiltration purification to remove impurities and concentrate the carrier.

[0224] S52. The ultrafiltration carrier liquid flows into the intermediate storage tank through a quick-release pipeline for intermediate storage.

[0225] S53. The carrier liquid in the intermediate storage tank is transported to the freeze-drying chamber of the vacuum freeze-drying module through a quick-release pipeline for vacuum freeze-drying and shaping to remove moisture and extend storage life.

[0226] S54. The freeze-dried solid nanomedicine carrier is introduced into the finished product collection module through a quick-release tubing (with a sterile quick connector) for aseptic dispensing and quantitative collection of the finished product.

[0227] S6. The image acquisition module acquires high-definition images of the nanomedicine carrier preparation process.

[0228] S7, the image analysis module analyzes and identifies the acquired images, promptly detecting any abnormalities.

[0229] S8, the data analysis module analyzes the data collected by each sensor and promptly detects anomalies.

[0230] S9. The comprehensive analysis module integrates the monitoring results of the image analysis module and the data analysis module, combines them with historical data, performs comprehensive analysis, accurately identifies abnormal situations, and predicts potential risks.

[0231] S10. When an abnormal situation or potential risk is detected, the alarm module will issue an alarm in a timely manner.

[0232] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for preparing a smart controlled-release nanomedicine carrier, characterized in that, Includes the following steps: S1. The intelligent control center performs initialization and process parameter setting; S2, The raw material pretreatment unit pretreatments the raw materials; S3. The multimodal dynamic control synthesis system enables the directional assembly of intelligent carriers and performs multimodal synthesis. S4. The online characterization quality control unit performs online characterization quality control. S5. Post-processing is performed by an integrated post-processing unit. S6. The image acquisition module acquires high-definition images of the nanomedicine carrier preparation process; S7, the image analysis module analyzes and identifies the acquired images, and promptly identifies abnormal situations; S8, the data analysis module analyzes the data collected by each sensor to detect anomalies in a timely manner; S9. The comprehensive analysis module integrates the monitoring results of the image analysis module and the data analysis module, combines them with historical data, performs comprehensive analysis, accurately identifies abnormal situations, and predicts potential risks. S10. When an abnormal situation or potential risk is detected, the alarm module will issue an alarm in a timely manner.

2. The method for preparing intelligent controlled-release nanomedicine carriers according to claim 1, characterized in that: Step S2 includes the following steps: S21, lipid raw material pretreatment channel for lipid pretreatment, through temperature-controlled stirring and ultrasonic dispersion pretreatment to ensure stable lipid activity and concentration; S22, the polymer raw material pretreatment channel performs polymer pretreatment, including wet grinding and two-stage filtration, to ensure that the polymer particles are uniform in size and free of impurities. S23, Inorganic nanomaterial pretreatment channel for inorganic nano pretreatment, using high pressure homogenization and degassing to achieve uniform dispersion of inorganic nanoparticles without bubbles; The lipid raw material pretreatment channel includes a temperature-controlled stirring and dissolving tank and an ultrasonic disperser; the temperature-controlled stirring and dissolving tank is equipped with a double-layer spiral stirring paddle mechanism; the double-layer spiral stirring paddle mechanism includes a temperature-controlled stirring and dissolving tank, a sealing cover, a drive motor, a lower spiral stirring paddle, an upper spiral stirring paddle, and a transmission mechanism; A temperature-controlled stirring dissolving tank has a detachable and fixed sealing cover at its upper end. A drive motor, an electric push rod, and a feed pump are fixedly mounted on the sealing cover. A feed pipe is fixedly mounted at the output end of the feed pump. A shaft is rotatably mounted below the sealing cover, and the output end of the drive motor is coaxially and fixedly connected to the shaft. A lower spiral stirring paddle is detachably and fixedly mounted at the lower end of the shaft. An upper spiral stirring paddle is rotatably mounted on the shaft. A transmission mechanism is located below the sealing cover, and the transmission mechanism is connected to the shaft and the upper spiral stirring paddle. The lower spiral stirring paddle and the upper spiral stirring paddle... Multiple spiral stirring blades are fixedly installed on the stirring paddle, and an ultrasonic probe is fixedly installed at the lower end of the movable rod of the electric push rod; multiple tank wall guide plates are evenly distributed around the inner wall of the temperature-controlled stirring and dissolving tank; the transmission mechanism includes a driving bevel gear, a driven bevel gear, and a transmission bevel gear; the driving bevel gear is fixedly installed coaxially with the shaft; the driven bevel gear is fixedly connected to the upper spiral stirring paddle, and the driven bevel gear is rotatably connected to the shaft; a transmission bevel gear is rotatably installed on the sealing cover, and the transmission bevel gear meshes with the driving bevel gear and the driven bevel gear for transmission.

3. The method for preparing intelligent controlled-release nanomedicine carriers according to claim 1, characterized in that: Step S3 includes the following steps: S31, The precision fluid delivery module performs precise material delivery; it simultaneously delivers pre-treatment raw materials and modifying reagents. S32. A switchable multimodal mixing chip assembly can be used for multimodal mixing, and the mixing mode of impingement jet, eddy current or laminar flow can be selected according to the carrier type. S33. The dynamic pressure feedback regulating valve regulates the pressure in real time to avoid particle size fluctuations. S34. In-situ modification reaction chamber is used to perform in-situ modification, complete the carrier modification and prevent carrier sedimentation.

4. The method for preparing intelligent controlled-release nanomedicine carriers according to claim 1, characterized in that: Step S4 includes the following steps: S41. The dynamic light scattering DLS particle size detection module performs particle size detection to determine whether the carrier particle size meets the standard. S42, the eta potential detection module performs zeta potential detection to monitor the zeta potential of the carrier and ensure stability; S43, the UV-Vis spectrophotometric drug loading rate detection module detects the drug loading rate and calculates the carrier encapsulation rate and free drug concentration.

5. The method for preparing intelligent controlled-release nanomedicine carriers according to claim 1, characterized in that: Step S5 includes the following steps: S51. The nanomedicine carrier that has passed online characterization and quality control is introduced into the fractional ultrafiltration module through a quick-release tubing; the fractional ultrafiltration module performs fractional ultrafiltration purification to remove impurities and concentrate the carrier. S52. The ultrafiltration carrier liquid flows into the intermediate storage tank through a quick-release pipeline for intermediate storage. S53. The carrier liquid in the intermediate storage tank is transported to the freeze-drying chamber of the vacuum freeze-drying module through a quick-release pipeline for vacuum freeze-drying and shaping to remove moisture. S54. The freeze-dried solid nanomedicine carrier is introduced into the finished product collection module through a quick-release tubing for aseptic dispensing and quantitative collection.

6. The method for preparing intelligent controlled-release nanomedicine carriers according to claim 1, characterized in that: Step S6 includes the following steps: S61, Image data reception and synchronization; S62. Image preprocessing: including noise reduction and correction; S63. Region of Interest (ROI) Extraction; S64. Multi-dimensional feature extraction: Features are extracted by using targeted algorithms in conjunction with the core observation scenarios of the device; S65. Anomaly Detection and Classification: Based on preset thresholds and feature logic, determine and classify anomaly types; S66. Output Results: Generate and visualize abnormal results; S67. Historical data storage and model optimization.

7. The method for preparing intelligent controlled-release nanomedicine carriers according to claim 1, characterized in that: Step S7 includes the following steps: S71, Data Reception and Classification; S72. Data preprocessing: including noise reduction and standardization; S73, Single-parameter anomaly analysis; S74, Multi-parameter correlation verification; S75. Anomaly Classification and Labeling; S76. Output Results: Generate an anomaly report and visualize the output.

8. The method for preparing intelligent controlled-release nanomedicine carriers according to claim 1, characterized in that: Step S9 includes the following steps: S91, Multi-source data reception and spatiotemporal alignment; S92. Evidence Theory Fusion Analysis: The DS evidence theory is used to fuse the credibility of multi-source data and construct a two-dimensional evidence body of image features and sensor parameters. S93. Accurate identification of the root cause of anomalies: Establish a rule base for the association between abnormal phenomena and root causes, and locate the root cause by combining fused data; S94, LSTM model risk prediction: Call the pre-trained LSTM neural network model to perform two types of predictions. The input features are real-time fused data, equipment runtime, and raw material batch information; the output is the risk probability and remaining time. S95. Dynamic classification and response to risk levels: dynamically update risk levels by combining the degree of urgency of the abnormality with the predicted scope of risk impact. S96, Result Output and Closed-Loop Recording.

9. The method for preparing intelligent controlled-release nanomedicine carriers according to claim 1, characterized in that: Step S10 includes the following steps: S101, LSTM model deployment and initial process parameter generation; S102, Real-time data acquisition and dynamic interaction; S102.1 Real-time reception and processing of multi-source data; S102.2 Dynamic control of process parameters; S103, Visual monitoring and human-computer interaction.

10. A smart controlled-release nanomedicine carrier preparation device, comprising: It can control a central processing unit, a raw material pretreatment unit, a multimodal dynamic control synthesis system, an online characterization and quality control unit, a post-processing integrated unit, an image acquisition module, an image analysis module, a data analysis module, a comprehensive analysis module, and an alarm module; its features are: Raw material pretreatment unit: pretreatment of raw materials, including lipid raw material pretreatment channel, polymer raw material pretreatment channel and inorganic nano-raw material pretreatment channel; Multimodal dynamic control synthesis system: realizes the directional assembly of intelligent carriers, including a precision fluid delivery module, a switchable multimodal hybrid chip assembly, a dynamic pressure feedback regulating valve, and an in-situ modification reaction chamber; Online characterization and quality control unit: includes dynamic light scattering DLS particle size detection module, Zeta potential detection module and UV-Vis spectrophotometric drug loading rate detection module; The integrated post-processing unit includes a staged ultrafiltration module, an intermediate storage tank, a vacuum freeze-drying module, and a finished product collection module. Image acquisition module: Includes a high-definition camera and LED lights, which acquire high-definition images of the nanomedicine carrier preparation process; Image analysis module: Analyzes and identifies the acquired images to promptly detect anomalies; Data analysis module: Analyzes data collected by various sensors to promptly detect anomalies; Comprehensive Analysis Module: This module integrates the monitoring results from the image analysis module and the data analysis module, combines them with historical data, and performs comprehensive analysis to accurately identify anomalies and predict potential risks. Alarm module: includes an alarm that promptly issues an alert when abnormal situations or potential risks are detected; Intelligent control center: It is network connected to the raw material pretreatment unit, multimodal dynamic regulation and synthesis system, online characterization and quality control unit, post-processing integrated unit, image acquisition module, image analysis module, data analysis module, comprehensive analysis module and alarm module.

Images