Intelligent parameter control system for coenzyme Q10 dry extrusion granulation process

Abstract

The present application relates to dry extrusion granulation control technical field, disclose a kind of coenzyme Q10 dry extrusion granulation process intelligent parameter control system, including controller, feed module, extrusion module, monitoring module, reversing module and crushing module.The present application extracts feed resistance torque rate of change to generate feedforward prediction signal, compensates the control delay caused by space transmission;Fusion outfeed side strip thickness and spectral data to calculate real-time equivalent density, perform volume compensation on spectral feature area, obtain volume compensation amorphous conversion index, exclude the interference of physical compaction density on phase change information measurement.Controller adjusts the pressure of extrusion wheel based on the feedforward signal and the conversion index, and calculates the temperature rise of the material by combining the mechanical work input, to couple the adjustment of the extrusion parameters and the cooling flow to prevent the thermal degradation of coenzyme Q10.Meanwhile, according to the conversion index, the reversing valve is operated to release or return the material.The present application realizes the cooperative closed-loop control of physical density and crystal transformation rate.

Description

Technical Field

[0001] This invention relates to the field of dry extrusion granulation control technology, specifically to an intelligent parameter control system for the dry extrusion granulation process of coenzyme Q10. Background Technology

[0002] Coenzyme Q10 is a fat-soluble macromolecule. In the preparation of solid dosage forms, it is usually necessary to transform it from a crystalline state to an amorphous state through dry extrusion technology to improve its bioavailability. In actual production, coenzyme Q10 is sensitive to mechanical stress and heat. Pressure fluctuations or heat accumulation during the extrusion process can easily lead to thermal degradation or incomplete phase transition of the active ingredient.

[0003] Existing dry extrusion control technologies mostly employ constant pressure or constant speed control modes, lacking predictive adjustment mechanisms for fluctuations in feed flow. Due to the physical displacement between extrusion molding and downstream inspection stations, relying solely on post-processing quality feedback leads to response lag in the control system. Furthermore, the compaction density of the strip directly affects the reflection intensity of online spectral monitoring, causing a complex interplay between physical filling state and chemical crystal form information, making it difficult to accurately determine the degree of amorphous transformation. When the produced strip does not meet quality standards, existing equipment typically lacks automated diversion and recirculation compensation logic, increasing raw material loss and reducing quality consistency in continuous production. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an intelligent parameter control system for the dry extrusion granulation process of Coenzyme Q10. This system solves the problems of fluctuations in the mechanical compaction and solid-state phase transition degree of the core extrusion zone due to variations in the bulk density of upstream premixed materials during continuous dry extrusion manufacturing of Coenzyme Q10; the localized temperature rise caused by the high-pressure mechanical work during extrusion leading to thermal degradation of the Coenzyme Q10 crystal structure; and the error in determining the amorphous crystal conversion rate caused by volume scattering interference from changes in the physical compaction density of the strip-shaped sheet, resulting in system closed-loop control lag and substandard product quality.

[0005] To achieve the above objectives, the first aspect of the present invention provides an intelligent parameter control system for the dry extrusion granulation process of coenzyme Q10, comprising a controller, and: a feeding module equipped with a feeding screw shaft and a torque sensor, the torque sensor collecting the propulsion resistance torque for the controller to generate a feedforward prediction signal; an extrusion module including opposing pressure rollers, the extrusion pressure of which is dynamically adjusted by the controller; a monitoring module including a spectral probe and a laser displacement sensor, located on the discharge side of the extrusion module, collecting spectral data and thickness data of the strip-shaped sheet for the controller to calculate the volume-compensated amorphous conversion index; a reversing module including a pneumatic reversing valve and a circulation channel connected to the feeding module; and a pulverizing module located below the output end of the pneumatic reversing valve; the controller, based on the feedforward prediction signal and the volume-compensated amorphous conversion index, jointly adjusts the extrusion pressure of the pressure rollers, and controls the pneumatic reversing valve to guide the strip-shaped sheet to the pulverizing module or the circulation channel according to the volume-compensated amorphous conversion index.

[0006] The innovative principle of the first aspect of the technical solution of this invention lies in: To address the issue of system control lag caused by the physical state of the material, this system employs a feedforward control approach. A torque sensor is coaxially mounted at the end of the feed drive shaft. The controller performs timestamp-based alignment and moving average filtering on the collected raw torque data to remove high-frequency mechanical vibration noise, and then calculates the rate of change of the propulsion resistance torque over time. This rate of change serves as a feedforward prediction signal. By calculating the first derivative of the torque, features are extracted when the material density changes but before it enters the pressure roller region, providing a time margin for the pressure adjustment action of the downstream hydraulic system and compensating for the control delay caused by spatial transmission.

[0007] To address the issue of physical morphology affecting the measurement of chemical phase transition information, this system employs a multi-source data spatiotemporal alignment and volume compensation algorithm. The strip-shaped sheet output by the extrusion module exhibits fluctuations in thickness and compaction density, leading to variations in molecular concentration within the probe volume for the same cross-sectional area. This interferes with the spectrometer's measurement of the integrated area of ​​the coenzyme Q10 characteristic peak. The controller integrates the strip-shaped sheet thickness measured by the laser displacement sensor, the upstream feed mass flow rate, and the pressure roller operating parameters, and calculates the real-time equivalent density of the strip-shaped sheet using the principle of mass conservation. Subsequently, the controller extracts the integrated area of ​​the spectral characteristic peaks reflecting the crystal structure and uses this equivalent density to perform volume compensation calculations on the integrated area. This maps the spectral scattering response under dynamic conditions to a constant density reference surface, eliminating optical measurement biases introduced by the volume expansion or compression of the strip-shaped sheet. The controller compares the volume-compensated characteristic peak area with the pre-stored pure crystalline reference integrated area to obtain a volume-compensated amorphous conversion index that excludes density interference, ensuring that the output phase transition index corresponds to the degree of lattice destruction induced by mechanochemical processes.

[0008] To address the issues of extrusion heating and coenzyme Q10 thermal degradation, this system employs thermodynamic coupling control logic. The controller calculates the specific mechanical energy input per unit mass of material based on comprehensive pressure commands, roller rotation speed, and feed mass flow rate. It then uses the specific heat constant and thermal conversion efficiency of coenzyme Q10 to estimate the internal temperature rise of the material. When the estimated temperature exceeds the critical threshold for thermal degradation, the controller generates a temperature suppression coupling coefficient, which is applied to the roller extrusion pressure and the feed screw shaft speed, performing a proportional reduction. This is combined with a feedforward proportional-integral algorithm to adjust the target flow rate of the cooling water circulation unit inside the roller. This mechanism reduces the heat generated by mechanical work and removes retained heat while maintaining the extrusion density per unit volume, achieving a balance between the mechanical energy required for crystal phase change and the prevention of thermal degradation.

[0009] In the quality closed-loop and material handling process, the controller extracts the trajectory length from the monitoring point to the tangent point of the pneumatic reversing valve, divides it by the linear velocity of the moving strip to obtain the transmission delay time, and adds the cylinder mechanical response time and lead time to this delay time to generate a level window time, ensuring that the rejection action and the spatial position of the non-conforming material are synchronized. For non-conforming materials diverted to the circulation channel, the controller calculates the bypass return mass flow rate based on the accumulated mass of the buffer hopper and the operating current of the extruder main motor. Since the return material has already undergone one mechanical work, the controller introduces a secondary phase change energy attenuation coefficient based on the return ratio to perform deduction compensation on the main pressure control command, preventing the mixture from absorbing excessive mechanical energy in the secondary extrusion section and causing thermal degradation.

[0010] The second aspect of this invention provides an intelligent parameter control method for the dry extrusion granulation process of coenzyme Q10, comprising the following steps: acquiring the rate of change of the propulsion resistance torque of the feed screw shaft through a controller, and using the rate of change of the propulsion resistance torque as a feedforward prediction signal; during the extrusion stage, calculating the mechanical work absorbed per unit mass of material based on the extrusion pressure, rotation speed, and feed rate of the pressure roller, and determining whether the material has the conditions for solid-state phase change; simultaneously acquiring spectral data and thickness data of the strip-shaped sheet, and calculating the volume-compensated amorphous conversion index excluding density interference; fusing the feedforward prediction signal and the volume-compensated amorphous conversion index, and performing feedforward and feedback combined pressure regulation control on the pressure roller through the controller; and controlling the pneumatic reversing valve to perform qualified product release or unqualified product return circulation based on the comparison result between the volume-compensated amorphous conversion index and the preset release threshold.

[0011] The innovative principle of the second aspect of the technical solution of this invention lies in: This control method employs a closed-loop control approach. During the feeding stage, the method uses the rate of change in propulsion resistance torque as a feedforward condition to pre-adjust the feed load impact. During the extrusion stage, it performs cumulative mechanical work evaluation to define the physical conditions for solid-state phase transition. During the discharge stage, it uses thickness data and the equivalent density calculated from the feeding conditions to perform volume compensation on the spectral area characterizing the coenzyme Q10 molecular structure, thus solving for the amorphous conversion index. Based on this, the method implements joint pressure regulation control: it uses the mass flow rate deviation to generate a feedforward base pressure command to compensate for feed disturbances, and uses the volume compensation for the amorphous conversion index deviation to generate a PID feedback pressure adjustment to correct phase transition errors. These two are then superimposed and integrated with the density dynamic compensation coefficient. This method solves the hysteresis problem of single-variable feedback control, achieving coordinated control of physical density and crystal form conversion rate.

[0012] This invention provides an intelligent parameter control system for the dry extrusion granulation process of coenzyme Q10. It has the following beneficial effects: 1. This invention generates a feedforward prediction signal by collecting the propulsion resistance torque at the end of the feed screw shaft and calculating its rate of change over time. When the bulk density of the upstream premixed material fluctuates, the system extracts the torque change characteristics before the material enters the extrusion zone and outputs a pressure adjustment command in advance. This mechanism provides a time margin for the downstream hydraulic system, compensates for the control delay caused by material spatial transmission, avoids the control lag caused by relying solely on feedback from the discharge end, and keeps the mechanical compaction operation in the extrusion zone stable.

[0013] 2. This invention calculates the real-time equivalent density of the material by fusing data on the thickness of the strip-shaped sheet collected by a laser displacement sensor, the upstream feed mass flow rate, and the operating parameters of the pressure roller. It also performs volume compensation calculations on the integrated area of ​​the characteristic peaks collected by the spectral probe. This calculation process eliminates optical volume scattering deviations introduced by the uneven thickness and physical compaction density of the strip-shaped sheet, obtaining a volume-compensated amorphous conversion index. This multi-source data compensation mechanism eliminates the interference of physical morphology fluctuations on phase transition information measurement, allowing the system's determination of the coenzyme Q10 crystal form conversion rate to directly correspond to the degree of phase transition in the molecular structure.

[0014] 3. This invention constructs a thermodynamic coupling control logic. The controller calculates the specific mechanical energy input and estimates the internal temperature rise of the material based on the extrusion pressure, roller speed, and feed mass flow rate. When the estimated temperature exceeds the critical threshold for thermal degradation, the controller generates a temperature suppression coupling coefficient to proportionally reduce the extrusion pressure and feed speed, and simultaneously increases the flow rate of the cooling water circulation unit inside the roller. This mechanism, while meeting the mechanical work required for the coenzyme Q10 crystal transformation, controls the proportion of mechanical energy converted into heat energy and accelerates the removal of retained heat, preventing excessive local temperature rise during extrusion that could lead to the thermal degradation of coenzyme Q10. Attached Figure Description

[0015] Figure 1 This is a system architecture diagram of the present invention; Figure 2 This is a flowchart of the present invention; Figure 3 This is a graph comparing the measured density of the strip sheet using the conventional control method of this invention and the method of this invention. Detailed Implementation

[0016] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] See attached document Figure 1 This invention provides a coenzyme Q10 dry extrusion intelligent parameter control system, including a feeding module, an extrusion module, a monitoring module, a reversing module, and a pulverizing module.

[0018] The feeding module is located above the extrusion module. The output end of the feeding module is connected to the input end of the extrusion module.

[0019] The feeding module contains a feeding screw shaft and a torque sensor. The torque sensor is connected to the controller.

[0020] The extrusion module includes two opposing pressure rollers. A cooling water circulation unit is integrated inside each pressure roller.

[0021] The monitoring module is installed on the discharge side of the extrusion module. The monitoring module includes a spectral probe and a laser displacement sensor.

[0022] Both the spectral probe and the laser displacement sensor are positioned facing the surface of the material extruded by the pressure roller.

[0023] The reversing module is located below the monitoring module. The reversing module includes a pneumatic reversing valve and a circulation channel.

[0024] One end of the circulation channel is connected to one output of the pneumatic directional valve. The other end of the circulation channel is connected to the input of the feeding module.

[0025] The pulverizing module is located below the other output end of the pneumatic reversing valve. The pulverizing module contains pulverizing blades and a screen.

[0026] See attached document Figure 2 , Figure 2 This is a flowchart of a coenzyme Q10 dry extrusion intelligent parameter control method according to an embodiment of the present invention. The present invention provides a coenzyme Q10 dry extrusion intelligent parameter control method, comprising the following steps: The controller performs initialization, sets the initial pressure roller pressure, initial pressure roller gap, feed screw speed, initial cooling water circulation flow rate, and target volume compensation amorphous conversion index, and controls the pneumatic reversing valve to default to the channel where the crushing module is located; After the coenzyme Q10 raw material powder is mixed with the water-soluble carrier excipient, it is fed into the feeding module. The feeding screw shaft rotates to propel the material. The torque sensor collects the propulsion resistance torque of the feeding screw shaft in real time. The controller obtains the rate of change of the propulsion resistance torque and uses it as a feedforward prediction signal. The material enters between the pressure rollers, and the pressure rollers apply mechanical extrusion force to the material. Coenzyme Q10 and the water-soluble carrier combine under force to form a solid dispersion, and the extrusion module outputs a strip-shaped sheet. As the strip passes through the discharge port, the monitoring module simultaneously collects measurement data, the spectral probe acquires the spectral data of the strip surface, and the laser displacement sensor acquires the thickness data of the strip. The controller receives spectral data and thickness data, calculates and obtains the volume-compensated amorphous conversion index, and compares the volume-compensated amorphous conversion index with the target volume-compensated amorphous conversion index. Based on the comparison results and the feedforward prediction signal, the controller calculates and adjusts the target regulating pressure of the pressure roller and the target cooling water flow rate of the cooling water circulation unit. The controller sends control commands to the reversing module based on the comparison results. When the volume compensation amorphous conversion index reaches or exceeds the target volume compensation amorphous conversion index, the pneumatic reversing valve maintains the current guiding state, and the strip sheet enters the crushing module for crushing and screening. When the volume compensation amorphous conversion index is less than the target volume compensation amorphous conversion index, the controller triggers the pneumatic reversing valve to switch the guiding state, and the strip sheet in this section enters the circulation channel and is transported back to the feeding module for re-extrusion processing.

[0027] In this embodiment, the specific process of material premixing and feedforward sensing feeding control is as follows.

[0028] Coenzyme Q10 raw material powder and water-soluble carrier excipient are physically mixed in a mixing device at a set mass ratio of 1:1 to 1:5. This mass ratio range is determined based on the steric hindrance effect of coenzyme Q10 to ensure that the carrier can provide hydrogen bonding sites and encapsulate drug molecules. As a preferred method, the water-soluble carrier excipient is one or a combination of polyvinylpyrrolidone, polyethylene glycol, or hydroxypropyl methylcellulose. Coenzyme Q10 is a lipid-soluble macromolecule, and using the above-mentioned polymers as water-soluble carriers can provide the skeletal structure and hydrogen bonding sites required to form a solid dispersion during subsequent high-pressure extrusion. For the specific structure and operating parameters of the mixing device, those skilled in the art can select a three-dimensional motion mixer or a V-type mixer according to the actual production scale. The physical mixing process is well-known in the art and will not be described in detail here.

[0029] The uniformly mixed premixed material is conveyed to the servo feeding hopper of the feeding module via a vacuum feeder or a closed-pipe pneumatic conveying system. The conveying process is maintained in a closed, dust-free environment, isolating external humidity and oxygen from entering, thus preventing the coenzyme Q10 from undergoing oxidative degradation or moisture absorption and agglomeration before extrusion.

[0030] The bottom of the servo-driven feeding hopper is equipped with a servo motor-driven feeding screw shaft. The feeding screw shaft rotates at a set feeding screw speed, pushing the premixed material into the gap between the pressure rollers of the extrusion module. Based on the overall control commands of the system, the servo motor receives the speed signal and performs volumetric quantitative conveying of the feeding amount.

[0031] Based on the general principles of powder dynamics, the resistance torque during screw conveying is positively correlated with the real-time bulk density of the material. The propulsion resistance torque is selected as a characteristic input parameter for evaluating the feeding state, reflecting the pre-compression state of the material before the pressure roller. During the propulsion of the premixed material on the feeding screw shaft, the transient fluctuations in the bulk density of the material itself cause changes in the conveying resistance experienced by the screw blades. A torque sensor is installed at the drive shaft end of the feeding screw shaft, employing a strain gauge or piezomagnetic sensing structure to continuously collect the propulsion resistance torque of the feeding screw shaft. Considering the presence of high-frequency mechanical vibration interference in industrial environments, the controller performs timestamp-based alignment processing on the multi-source time series data before calculation and applies a moving average filtering algorithm to smooth the collected raw torque data. In specific implementation, the window length of the moving average filter is preferably 5 to 15 sampling periods to retain the material density step signal while filtering out high-frequency mechanical noise. The controller acquires the processed propulsion resistance torque data, calculates the transient rate of change of the propulsion resistance torque over time to quantify the degree of density variation of the premixed material, and generates a feedforward prediction signal. The calculation formula is as follows: ; In the formula, The current propulsion resistance torque change rate is the feedforward prediction signal. The instantaneous value of propulsion resistance torque, acquired by the torque sensor at the current moment and filtered. This represents the instantaneous value of the propulsion resistance torque from the previous sampling period; This is the sampling period of the controller. Because... Determined by the system hardware clock and set to a constant strictly greater than zero, its actual value is set to a fixed parameter between 0.01s and 0.1s. Therefore, in engineering implementation, this algorithm logic avoids singularity anomalies where the denominator approaches 0. The physical meaning of this formula is that by capturing the instantaneous first derivative of torque, the system can extract abnormal characteristics in the early stages of sudden changes in material density, providing time margin for downstream closed-loop control.

[0032] The controller inputs the calculated feedforward prediction signal into the subsequent pressure regulation closed-loop control model. To avoid equipment malfunctions caused by a single numerical anomaly, a multi-dimensional weighted judgment mechanism is introduced into the judgment logic for outputting feedforward intervention commands. The controller presets a threshold for the rate of change of propulsion resistance torque and an absolute torque tolerance range. The threshold for the rate of change of propulsion resistance torque is determined statistically based on three times the standard deviation of historical steady-state operating data; the absolute torque tolerance range is set to ±5% to ±10% of the reference steady-state torque, which is the calibration baseline value obtained under no-load operation and standard feeding test conditions. When the rate of change of propulsion resistance torque exceeds the set threshold, and the instantaneous torque value deviates from the reference steady-state torque by more than the absolute torque tolerance, it indicates that there is a local density mutation in the material about to enter the pressure roller. The controller uses this feedforward prediction signal to trigger the pressure roller pressure compensation adjustment action in advance before the material actually enters the pressure roller zone, eliminating system control lag caused by the non-uniformity of the material's physical state.

[0033] In this embodiment, the specific implementation details of online monitoring of feed torque and extraction of feedforward prediction signals are as follows.

[0034] The feeding module is equipped with a servo motor-driven feeding screw shaft. To acquire data on the physical state changes of the material during conveying, a torque sensor is coaxially mounted on the drive shaft end of the feeding screw shaft. Preferably, this torque sensor employs a strain gauge sensing structure, transmitting analog electrical signals to the high-speed analog-to-digital converter interface of the main control unit via a non-contact slip ring. Based on powder dynamics principles, the propulsion resistance torque during screw conveying is positively correlated with the bulk density of the material. Based on this physical correspondence, the controller selects the propulsion resistance torque as a characteristic input parameter for evaluating the feeding state, reflecting the material's pre-compression state before entering the extrusion module.

[0035] During the propulsion of the premixed material via the feeding auger, transient fluctuations in the material's bulk density cause corresponding changes in the conveying resistance experienced by the auger blades. A torque sensor continuously collects the original propulsion resistance torque of the feeding auger. Considering the high-frequency mechanical vibration interference caused by motor rotation and gear meshing in industrial production environments, the controller performs timestamp-based alignment processing on the multi-source time-series data before extracting feature signals. To address the timestamp misalignment problem caused by inconsistent sampling frequencies from different sensors, the controller uses a zero-order hold or linear interpolation algorithm to resample the collected data, ensuring data synchronization in the time dimension. To eliminate high-frequency noise and obtain smooth data reflecting the material's state, the controller applies a moving average filtering algorithm to the aligned original propulsion resistance torque data. The specific data smoothing formula is as follows: ; In the formula, This is the sampling window length for the moving average filter; The original propulsion resistance torque is the actual output of the sensor. This is the index for the discrete time step within the sliding window. The physical purpose of this filtering process is to smooth out the glitches caused by mechanical vibration by calculating the mean within a specific time window, while preserving the low-frequency trend changes caused by sudden changes in material density. The sampling window length is set to filter out high-frequency mechanical noise while retaining the characteristic signal of material density steps. The value range is defined as 5 to 15 sampling periods. Sampling period Determined by the system hardware clock tick, its actual value is set to a fixed parameter of 0.01 seconds to 0.1 seconds.

[0036] In this embodiment, the premixed material enters the core extrusion stage after being pushed by the feeding module, and the specific implementation process of its mechanochemical induced solid dispersion formation is as follows.

[0037] The premixed material is continuously pushed by a feeding auger shaft between two opposing pressure rollers in the extrusion module. In this embodiment, the roller surfaces employ a textured or straight-knotted structure to increase the frictional gripping force on the powder material. The two rollers rotate in opposite directions at a set roller speed, biting the premixed material into the working bite angle region of the rollers. Within this spatial region, the porosity between powder particles decreases as the physical gap between the rollers shrinks, resulting in material rearrangement and initial compaction. Regarding the selection of the roller drive motor and the synchronous transmission structure, those skilled in the art can select a dual-output shaft reducer for synchronous drive based on the equipment's capacity requirements. The mechanical transmission design is well-known in the field and will not be elaborated upon here.

[0038] The material, after initial compaction, enters the core extrusion zone as the pressure roller rotates. Based on the mechanical principle of powder material undergoing deformation under pressure in a confined space, the mechanical work transmitted by the pressure roller to the material is the fundamental energy that causes changes in the microstructure within the crystal. This mechanical stress acts directly on the coenzyme Q10 crystal particles, forcing them to undergo plastic deformation and micro-fragmentation. The accumulated mechanical energy is converted into activation energy that disrupts the lattice energy within the crystal, breaking the cohesive force within the coenzyme Q10 molecule and transforming the crystal structure into an amorphous state. To quantify the degree of mechanical work done in this region, extrusion pressure, rotational speed, and feed rate, which are directly related to the work done, are selected as input parameters. The system introduces the mechanical energy input per unit mass as a basic parameter for molding evaluation, and its calculation formula is as follows: ; In the formula, Mechanical work absorbed per unit mass of material; The real-time extrusion pressure applied to the pressure roller; This refers to the outer diameter of the pressure roller; The rotational speed of the pressure roller; This is the real-time mass flow rate of the material. Among the parameters, the outer diameter... These are inherent constants of the equipment. Determined by the feed rate of the feedforward module. Real-time extrusion pressure. The value range is set to 20kN to 80kN, and the rotational speed of the pressure roller is... The rpm setting is 5 to 20 rpm, and this parameter range is determined based on the powder rheological properties test of coenzyme Q10 material. To ensure the completeness of the engineering calculations, when the equipment is in the feeding start-up / stop phase or when material blockage occurs... When the flow rate falls below the rated minimum operating flow threshold (set to 0.1 kg / h), the controller will... Forced zeroing avoids computational overflow anomalies caused by the denominator approaching 0. The physical meaning of this formula lies in assessing the energy boundary driving crystal structure destruction by calculating the mechanical energy acquired per unit mass of material in the extrusion zone. To avoid biased judgments based solely on the accumulated mechanical work, the system's logic for verifying whether the material possesses the conditions for a solid-state phase transition is based on a multi-dimensional index fusion: when the calculated absorbed mechanical work reaches the coenzyme Q10 lattice activation energy threshold, and the real-time extrusion pressure... When the absolute value of the coenzyme Q10 lattice activation energy threshold and the lower limit of the material yield strength simultaneously exceed the lower limit of the material yield strength, the material is deemed to have the physical conditions for a solid-state phase transition. As a preferred method, the aforementioned coenzyme Q10 lattice activation energy threshold and the lower limit of the material yield strength are pre-calibrated using offline differential scanning calorimetry and a powder compression tester, and stored as constants in the controller's data register.

[0039] Once the mechanical energy input meets the aforementioned multidimensional criteria, intermolecular reconstruction at the microscopic level begins. Under the high-pressure environment of the core extrusion zone, the spatial distance between coenzyme Q10 molecules and water-soluble carrier excipient molecules is compressed to the range of intermolecular van der Waals forces and hydrogen bonds. The coenzyme Q10 molecules, with their lattice constraints broken, interact with hydrogen donor or acceptor groups on the carrier polymer chains, forming a stable network of intermolecular hydrogen bonds. Through this microscopic reconstruction process, coenzyme Q10 is dispersed amorphously within the framework of the water-soluble carrier, transforming from a powder mixture into a thermodynamically metastable solid dispersion.

[0040] After the aforementioned mechanochemical induction, the formed solid dispersion needs to be demolded. The solid dispersion, having completed phase change and compaction, continues to rotate with the pressure roller, separating from the core extrusion zone and entering the release zone. As the mechanical pressure on the material gradually decreases to zero, it undergoes a volumetric elastic recovery process and is continuously discharged from the gap at the bottom of the pressure roller in the form of a strip-shaped sheet. The output strip-shaped sheet possesses a certain mechanical strength and a continuous geometric shape, providing a standardized physical carrier for subsequent online phase change monitoring and granulation.

[0041] In this embodiment, the specific implementation process of solid-state phase transition spectral monitoring and analysis after the strip sheet is formed is as follows.

[0042] After being extruded from the gap between the pressure rollers, the strip-shaped sheet is conveyed downwards along the feed chute, with an online spectral monitoring station along this conveying path. The controller is equipped with a spectrometer based on Raman scattering or near-infrared absorption principles for assessing the internal state of the material. In this embodiment, the spectral probe is fixed above the feed chute by a shock-resistant bracket, with its lens tip maintaining a set detection distance from the sheet surface. The probe emits an excitation beam to irradiate the moving material surface, receives the echo light signal, and transmits the light signal back to the host for photoelectric conversion via optical fiber. During equipment operation, the spectrometer collects raw spectral data sequences reflecting the vibrational modes of material molecules at set time intervals. To ensure spatial continuity of monitoring, this time interval parameter is determined based on the quotient of the linear velocity of the strip-shaped sheet and the physical diameter of the probe spot, preventing missed detection blind spots or overlapping sampling. The underlying hardware working principle of the spectrometer's light source excitation and photoelectric conversion module is well-known in the art and will not be described further here.

[0043] Due to variations in ambient light in industrial settings and diffuse reflection caused by the surface roughness of the strip-shaped sheet, the directly acquired raw spectral data suffers from baseline drift and multiplicative scattering interference. To remove the masking effect of physical morphology on chemical information, the controller extracts the raw spectral vector at any given time point and then applies a standard normal variable transformation algorithm for independent correction of the single-spectral data. The specific data standardization formula is as follows: ; In the formula, After standardization, at wavelength The scalar value of the spectral intensity at that location; This represents the raw spectral intensity actually output by the spectrometer. This represents the average spectral intensity of the current full-spectrum data; This represents the standard deviation of the current full-spectrum data; This is a preset bias normal value for the system. Since the background noise variance of the signal captured by the sensor under no-load or no-reflection conditions approaches zero, this bias normal value is introduced into the algorithm logic. (Its empirical value is set to 1×10) -6 This process avoids division overflow exceptions at the computational level. The physical purpose of this step is to ensure that the data distribution of each spectrum satisfies zero mean and unit variance, smoothing out baseline shifts caused by differences in material compaction between different production batches, and allowing subsequent algorithms to focus on the characteristic peak positions and intensity variations that reflect the crystal structure.

[0044] Based on the principle of quantitative analysis using molecular vibrational spectroscopy, the content of a specific crystal form of a substance is directly proportional to the area of ​​its characteristic absorption or scattering peaks. During the transformation of Coenzyme Q10 from an ordered crystalline state to an amorphous state, the degree of inhibition on the vibrations of its internal functional groups decreases. Due to the formation of intermolecular hydrogen bonds with the carrier excipients, the characteristic peaks corresponding to the ordered lattice exhibit a regular intensity decay. The controller extracts sensitive characteristic peak intervals characterizing the crystal structure of Coenzyme Q10 across the entire spectrum. As a preferred approach, this sensitive interval corresponds to the vibrational frequency band of the benzoquinone ring or isoprene side chain in the Coenzyme Q10 molecule, and its start and end wavelength ranges are predefined by comparative testing of offline standard samples. The system calculates the characteristic peak area within the set band interval through numerical integration; the discretization formula of its integration algorithm is as follows: ; In the formula, For the current moment Real-time integral area of ​​the characteristic peak region; and These are the wavelength discrete point indices for the start and end positions of the sensitive characteristic peak interval, respectively. At the current moment at wavelength Standardized spectral intensity at the location; This represents the spectral resolution constant between adjacent sampling points of the spectrometer. The technical purpose of this calculation is to quantify the content of residual ordered crystal structures within the currently monitored ribbon-like thin slice by using the value of the integral area.

[0045] After acquiring the real-time integrated area, the system converts it into a macroscopic indicator characterizing the quality of the solid dispersion. The controller retrieves the pre-cured pure crystalline reference area from the data storage unit, compares it with the real-time measurement value, and calculates to generate a solid-state phase change conversion rate indicator. ; In the formula, This represents the solid-state phase transition conversion rate calculated at the current moment. This refers to the baseline integrated area of ​​a pure crystalline coenzyme Q10 physical mixture, obtained during the offline calibration phase of the device, without mechanical or chemical extrusion, within the same wavelength range. Since the characteristic peak area of ​​a pure crystalline substance is a strictly positive physical calibration constant, the division operation here avoids the singularity problem of a zero denominator. To prevent distortion caused by localized compositional fluctuations, the controller incorporates mean verification logic within a sliding time window. When multiple consecutive sampling periods... When the mean value is higher than the set acceptable conversion rate threshold, and the variance of the numerical fluctuation in each period is lower than the set stability tolerance, the controller determines that the currently produced strip-shaped sheet has achieved sufficient mechanochemical phase transition. In this embodiment, the acceptable conversion rate threshold is pre-calibrated based on offline in vitro drug dissolution testing experiments and stored in the controller as a fixed parameter set within the range of 90% to 95%; the aforementioned stability tolerance is set to a fixed percentage of less than or equal to 2% to ensure the continuous consistency of product conversion quality.

[0046] In this embodiment, the specific implementation process of thickness monitoring and dynamic compensation of compaction density of the strip sheet is as follows.

[0047] After being discharged from the gap between the pressure rollers and undergoing volume recovery, the strip-shaped sheet enters the thickness measurement station along the guiding mechanism. Laser displacement sensors are installed opposite each other on the upper and lower sides of the conveying path. Preferably, the main optical axes of the upper and lower sensors are collinear and perpendicular to the surface of the material strip. During equipment operation, the upper and lower sensors synchronously emit laser beams towards the surface of the strip-shaped sheet and receive the reflected light signals. Based on the principle of optical triangulation, the system obtains the distance from the probe to the sheet surface. The internal photoelectric conversion and distance calculation of the sensors are well-known technologies in the field and will not be described in detail here. The controller calculates the dynamic thickness of the sheet based on the distance data, using the following formula: ; In the formula, For the current moment Real-time thickness of the strip-shaped sheet; This is a fixed reference distance between the end faces of the upper and lower sensor probes; and These are the actual distances from the probe to the thin film surface measured by the upper and lower sensors at the current moment, respectively. Fixed reference distance. This is a constant calibrated using standard gauge blocks during the equipment installation phase. This thickness parameter quantifies the expansion state of the solid dispersion after it leaves the extrusion zone, providing a geometric basis for assessing compaction density.

[0048] After thickness data acquisition, a density assessment model is constructed based on the equipment's operating conditions. In actual continuous production, there is a fixed distance between the feeding and thickness measurement stations. Directly mixing and calculating data from the same absolute moment would lead to material misalignment. The controller calculates the time delay of material transport based on the path length between the sensor and the pressure roller and the average linear velocity of the material movement. The upstream feed mass flow rate and pressure roller speed data are then aligned using this delay before being matched with the downstream thickness value in the same time slice. Based on the principle of mass conservation, the controller calculates the real-time equivalent density of the material. ; In the formula, The calculated density of the strip at the current moment; The feed mass flow rate is matched after delayed translation; The outer diameter constant of the pressure roller; This is the rotational speed matched after delayed translation; The width of the strip sheet being formed; This is a preset bias constant. Among the parameters, The clearance is determined by the side sealing plates at both ends of the pressure roller. This is due to material breakage or measurement abnormalities. Approaching zero, the system introduces a value of 1×10. -5 Preset bias constant This step avoids computational anomalies caused by a denominator of zero. It couples geometric quantities with operating conditions, converting them into indicators that reflect the internal tightness.

[0049] Based on the difference between the obtained density index and the target value, the system implements closed-loop compensation control. To avoid actuator oscillations caused by measurement noise or sudden changes in local thickness, the controller's decision logic relies on the mean value within the sliding time window. The system applies a moving average filtering algorithm to the continuously acquired real-time density to calculate the average density. When this average density exceeds the allowable tolerance dead zone, the controller triggers the compensation algorithm to calculate the dynamic compensation coefficient for the hydraulic actuator. ; In the formula, This is the generated extrusion pressure compensation coefficient; The target compaction density is preset; The average density calculated within the sliding time window; This is a proportional adjustment gain coefficient. In this embodiment, based on the material characteristics of coenzyme Q10, The value range is set to 1.1 g / cm³. 3 Up to 1.3 g / cm 3 This value is based on previous offline tablet compression experiments. A dimensionless constant between 0.1 and 0.3 is set. As a preferred method, the sliding time window length is set to 3 to 10 seconds, and the tolerance dead zone is set to ±2% of the target density. These two parameters are set based on offline testing of the hydraulic mechanism's mechanical response cycle. Theoretically, the target density is always greater than zero; the division operation here avoids singularity anomalies. This step converts the relative density error into the control ratio of the actuator.

[0050] By applying the aforementioned compensation coefficient, the hydraulic system achieves dynamic adaptation of the extrusion pressure. The controller sends command frames to the hydraulic servo execution unit, and the hydraulic system multiplies the current base extrusion pressure setting by the compensation coefficient to update the actual output pressure of the pressure roller hydraulic cylinder. Based on proportional compensation logic, when the density is lower than the target value, the system increases the hydraulic extrusion pressure to compress residual pores; when the density is higher than the target value, the system reduces the extrusion pressure to alleviate over-compaction. Through the closed loop of thickness measurement and density control, the produced strip sheets maintain a continuous and consistent macroscopic density.

[0051] In this embodiment, the dynamic calculation process of volume compensation and true amorphous conversion index for the strip-shaped sheet is as follows.

[0052] During the conveying process of the strip sheet in the feed chute, the online spectral monitoring station and the laser displacement thickness measurement station are physically positioned one behind the other. To ensure the correspondence between the microscopic chemical information and macroscopic physical information of the same batch of materials, the controller executes data spatiotemporal alignment logic at the underlying level. The controller extracts the physical distance between the main optical axes of the sensors at the two stations, divides this distance value by the real-time linear velocity of the strip sheet, and calculates the physical transmission delay time of the same material block between the two stations. To ensure closed-loop data processing, this linear velocity parameter is calculated by the controller based on the product of the physical outer diameter of the equipment's pressure roller and the real-time rotational speed at the current moment. Test data collected from the upstream station is shifted by this delay time and enters the data buffer queue, where it is paired with the data collected by the downstream station at the current moment. Through this synchronization mechanism, the system binds the integral area and equivalent density of characteristic peaks belonging to the same physical slice into coupled data pairs.

[0053] After completing the spatiotemporal alignment of the data, the controller initiates the volume scattering compensation algorithm. Based on the principle of powder optical analysis, the intensity of the characteristic peak response acquired by the optical probe is limited by the number of molecules within the probe volume. Under conditions of fluctuating compaction density, an increase in material density leads to an increase in the concentration of crystal molecules per unit probe volume, causing a positive response deviation in the integral area of ​​the characteristic peak as the compaction density increases. To eliminate the interference of physical density changes on microstructure analysis, in this embodiment, the controller introduces a reference density parameter to perform volume normalization compensation on the real-time integral area. The calculation formula is as follows: ; In the formula, This represents the area of ​​the characteristic peak after volume compensation at the current moment. The integral area of ​​the real-time characteristic peak after spatiotemporal alignment processing; The reference compaction density of pure crystalline materials when the equipment is calibrated offline to the reference area; The real-time equivalent density of the strip-shaped thin sheet after spatiotemporal alignment; This is the bias constant for the area of ​​the characteristic peak. As a preferred method, The value range is set to 0.8 g / cm³. 3 Up to 1.0 g / cm 3 This value was simultaneously measured using a powder tap density tester during the offline spectral calibration stage and then solidified in the controller. The system input value is 1×10. -6 bias constant This avoids the controller triggering a division overflow exception when reading the no-load state and the density drops to zero. The technical purpose of this calculation step is to map the spectral response under dynamic operating conditions to a unified constant density reference surface, filtering out optical measurement biases introduced by volume expansion or compression.

[0054] After obtaining the density-normalized characteristic peak area, the controller substitutes it into the conversion rate evaluation model to calculate the true phase transition index, excluding physical state interference. The controller then calls the pure crystalline reference parameters in the data register to calculate the volume-compensated amorphous conversion index. ; In the formula, The volume-compensated amorphous conversion index is calculated at the current moment. This represents the baseline integrated area of ​​the pure crystalline coenzyme Q10 physical mixture obtained during the offline calibration phase of the equipment. Since the calibration area of ​​pure crystalline substances is a non-zero physical constant, this formula avoids the singularity problem of a zero denominator at the underlying logic level. The physical meaning of this compensation model is that, through the fusion of multi-dimensional parameters, it ensures that the final output amorphous conversion rate is only related to the degree of lattice disruption induced by mechanochemical processes, independent of the macroscopic geometry and compact state of the material.

[0055] Based on the decoupled true conversion index, the controller performs the final quality judgment and material diversion closed loop. The controller retrieves the set release threshold parameters and processes continuously generated data. The data stream is compared in real time. When the index value is consistently higher than the release threshold for multiple consecutive calculation cycles, and the fluctuation variance is within the set tolerance dead zone, the controller determines that the solid dispersion in the current segment meets the qualified phase change requirements and outputs a release level command. When the index falls below the release threshold, the controller determines that the current material has a phase change blind zone or a defect in the carrier coating, triggering the pneumatic diversion flap to remove the unqualified strip-shaped flakes in that segment to the waste recycling channel. In this embodiment, the release threshold is set to 95%, and the parameter is pre-established based on the in vitro dissolution consistency evaluation requirements of solid drug formulations. As a preferred method, the tolerance dead zone is set to ±1% to 3% of the index mean, and this parameter boundary is calibrated offline based on the statistical distribution of historical qualified batches. By integrating physical density and chemical spectrum for multi-dimensional compensation judgment, the system eliminates the judgment distortion caused by simply relying on spectral area. The solenoid valve drive and mechanical linkage structure of the pneumatic diversion flap are well-known technologies in the field and will not be described in detail here.

[0056] In this embodiment, the specific implementation process of the feedforward and feedback combined pressure regulation closed-loop control law of the pressure roller hydraulic actuator is as follows.

[0057] During extrusion molding, fluctuations in the mass flow rate of the upstream feeding module directly alter the material filling density entering the pressure roller gap. There is a physical time lag between material extrusion molding and downstream spectral monitoring; relying solely on downstream monitoring results for feedback adjustment leads to system response delays. To intervene before physical disturbances contact the pressure rollers, the controller constructs a feedforward control model based on mass flow rate deviation. The controller extracts the current feed mass flow rate entering the pressure roller gap, a value derived from historical data measured by the upstream feeder before the transmission delay and aligned via a data queue. The system calculates basic feedforward pressure adjustment commands based on upstream load changes, compensating in advance for the mechanical work required by the pressure rollers. ; In the formula, For the current moment The calculated feedforward base pressure command; The baseline maintenance pressure calibrated for offline no-load testing of the equipment; This is the feedforward proportional gain coefficient; The actual feed mass flow rate delivered to the pressure roller at the current moment after translation and alignment; The preset rated reference mass flow rate for the controller. In this embodiment, The parameter is set to a dimensionless value range of 0.5 to 1.5, and its specific value is pre-calibrated based on the compressive rheological properties of the offline extruded material. This calculation step offsets the interference of feed rate fluctuations on compaction density.

[0058] To address the internal rheological disturbances of materials that cannot be completely eliminated by feedforward control, the controller introduces a volume-compensated amorphous conversion index for closed-loop feedback correction. Based on proportional-integral-derivative feedback control theory, the system calculates the conversion index error value. This error value is equal to the preset target conversion index. The volume-compensated amorphous conversion index at the current moment, output from the previous steps. The difference. Combining the current error value, historical cumulative trend, and transient rate of change, the controller runs a discretized calculation formula at the lower level to calculate the feedback compensation pressure: ; In the formula, This is the generated feedback pressure adjustment amount; , and These are the proportional gain, integral gain, and differential gain constants, respectively. This is an index for the discrete time series from the initial time to the current time. In this embodiment, The parameters are set to fixed values ​​within the range of 96% to 98%; the above three control gain constants are tuned offline based on the critical proportionality method and fixed in the storage unit; This represents the volume compensation amorphous conversion index error value at the current sampling time; Indicates the previous sampling period (i.e., time). The volume compensation amorphous conversion index error value recorded in the control system; Indicates from the initial running time ( Accumulate to the current sampling time ( The discrete summation operator is used to implement the numerical accumulation calculation of integral terms in the underlying control logic; Indicates the index in discrete time series The volume-compensated amorphous conversion index error value is recorded and stored in the control system's buffer queue; The sampling period represents the bottom-level control step size of the control system.

[0059] because Controlled by the underlying hardware clock frequency of the equipment, its value is a strictly positive physical constant, thus avoiding overflow anomalies caused by the denominator being zero in the differential term at the computational architecture level. To prevent the system from experiencing integral term saturation divergence due to continuous feeding deviations, the controller incorporates anti-integral saturation logic at the discrete computation level. When the comprehensive pressure adjustment command reaches the physical safety upper limit threshold of the hydraulic cylinder, the system forcibly freezes the integral accumulation calculation of the current cycle until the aforementioned conversion exponent error value is reached. When a reverse zero-point crossing occurs, the integral accumulation is restored to maintain the dynamic recovery capability of the actuator.

[0060] After obtaining the feedforward and feedback calculation results, the controller couples them with the density dynamic compensation coefficient to generate the final hydraulic servo execution command. To prevent sudden changes in the command caused by superposition calculations from leading to hydraulic cylinder overload, a hardware limiting algorithm is connected in series at the command output end. The joint pressure regulation calculation formula is as follows: ; In the formula, To comprehensively issue the final output pressure command to the hydraulic system; Forward pressure command; For feedback pressure adjustment; These are the dynamic compensation coefficients generated based on the density gradient deviation. The controller... Boundary checks are performed. When the value exceeds the upper and lower limit threshold ranges permitted by the equipment's hydraulic system safety, the underlying logic forcibly truncates it to the boundary extreme value parameter. In this embodiment, the aforementioned upper and lower limit threshold ranges are pre-calibrated based on the hydraulic cylinder's safety pressure resistance extreme value and the minimum forming pressure requirement of the material. After limiting, the controller converts the digital command into an analog current signal or a fieldbus data frame and sends it to the hydraulic servo proportional relief valve of the pressure roller. The oil circuit pressure regulation, pressure holding logic, and electro-hydraulic signal conversion principle of the hydraulic system are well-known technologies in the field and will not be elaborated here. With the help of this joint closed-loop control architecture, the equipment can maintain a continuous and stable mechanochemical output field under conditions of external feeding disturbances and internal phase change fluctuations.

[0061] In this embodiment, to avoid thermal degradation of the material during overload extrusion, the thermodynamic coupling control of pressure and feed flow rate is implemented as follows.

[0062] During the extrusion molding stage, solid dispersions undergo mechanical shearing and physical compression, converting some mechanical energy into heat. Components such as Coenzyme Q10 exhibit thermosensitive properties; elevated temperatures can easily trigger thermal degradation and deactivation of the crystal structure. Based on the principle of mechanical work-heat equivalent conversion, the controller constructs a thermodynamic conversion estimation model for the extrusion process. The controller extracts the currently issued comprehensive output pressure command, the real-time rotational speed of the pressure roller, and the actual feed mass flow rate to calculate the specific mechanical energy input per unit mass of material under the current operating conditions. ; In the formula, For the current moment The estimated material specific mechanical energy; To output pressure commands in a comprehensive manner; This refers to the real-time rotational speed; The torque conversion constant is determined by the combined mechanical transmission structure of the equipment and the geometry of the pressure roller; This represents the actual feed mass flow rate; This is a preset bias constant for the specific mechanical energy of the material. As a preferred method, The value is 1×10 -4 This design avoids division overflow anomalies caused by the denominator approaching zero due to feed interruption at the computational level.

[0063] After obtaining the specific mechanical energy, and combining it with the law of conservation of energy, the system calculates the transient temperature rise inside the material: ; In the formula, This is the estimated transient temperature rise; The physical conversion efficiency coefficient from mechanical energy to thermal energy; This is the specific heat capacity constant of the physical mixture of coenzyme Q10. Among the parameters, and Based on prior offline testing and calibration of the adiabatic calorimeter, the controller directly uses this estimated transient temperature rise to simplify the underlying computational architecture. The ambient temperature of the extrusion chamber is collected in real time by the built-in temperature sensor of the equipment. Perform arithmetic summation to obtain the current estimated temperature, which includes the background ambient heat. To avoid false triggering caused by noise from a single temperature peak, the system performs multi-dimensional joint judgment based on the current estimated temperature and the rate of temperature change. The controller extracts the estimated temperature of the current calculation cycle and performs a difference calculation with the estimated temperature of the previous cycle, and divides this temperature difference by the system sampling cycle to obtain a discretized transient heating rate. When the estimated temperature exceeds the thermal degradation critical threshold for multiple sampling cycles, and the above-mentioned discretized heating rate is greater than zero, the controller determines that the current pressure regulation command has a physical risk of causing thermal degradation and triggers thermodynamic coupling intervention. In this embodiment, the thermal degradation critical threshold is set in the range of 45°C to 50°C, and the parameters are established based on the melting point of coenzyme Q10 and thermogravimetric analysis data. The technical purpose of this calculation step is to establish a physical mapping between mechanochemistry and thermodynamics, forming a temperature early warning mechanism.

[0064] After determining the risk of thermal degradation, simply reducing the pressure of the hydraulic actuator would disrupt the amorphous transformation process within the material. To reduce the total heat generation of the system while maintaining the phase change shear force, the controller initiates a pressure and feed flow rate coupled reduction algorithm, simultaneously activating the active heat exchange closed loop of the cooling water circulation unit. Based on the principles of thermodynamics and mass conservation, the controller calculates the temperature suppression coupling coefficient: ; In the formula, The generated temperature-suppressed coupling coefficient; Thermodynamically adjustable gain parameters; This is the estimated current temperature obtained from the previous steps; This is the preset critical threshold for thermal degradation. Because... The physical constant is set to be greater than zero, so there is no risk of a singularity (denominator equal to zero) in the underlying logic. To ensure that the system does not interfere with normal production under safe operating conditions, the underlying logic limits operations to the following conditions when the estimated temperature is below a threshold: The parameter is constant at 1 and does not participate in the calculation. This is a preferred method. The value range is set to 0.2 to 0.6, and this range is determined through engineering testing and calibration of the servo motor's temperature control lag response time. After obtaining this dimensionless coefficient, the controller synchronously performs coupled correction on the pressure command and the feed flow command: ; ; In the formula, The execution pressure command is calculated based on thermodynamic safety. This is to synchronize and match the safe feed flow rate command. By proportionally reducing the pressure input and material throughput, the technical purpose of this calculation step is to reduce the heat generated by a single mechanical operation, maintain the extrusion density boundary conditions per unit volume of material, and ensure that the crystal phase transformation and thermal degradation prevention are in a balanced state.

[0065] While performing the aforementioned physical reduction of pressure and flow, the controller extracts the estimated current temperature. Compared with the preset thermal degradation critical threshold The difference is used to dynamically adjust the target flow rate of the cooling water circulation unit through a feedforward proportional-integral algorithm: ; In the formula, The target cooling water flow rate command generated at the current moment; The basic maintenance cooling water flow rate is set for system initialization; and These are the preset proportional and integral gain constants for the cooling water circuit. When the estimated temperature exceeds the critical threshold, the controller will... The signal is converted into an analog signal and sent to the electric regulating valve in the cooling water pipeline to proportionally increase the flow rate of the cooling medium. The physical meaning of this closed-loop mechanism is that it actively removes the heat trapped inside the pressure roller by using forced water cooling, and in conjunction with the synchronous reduction of mechanical heat sources, it achieves accelerated thermodynamic equilibrium of the extrusion system, ensuring that the expected goals of preventing crystal phase transformation and thermal degradation are achieved.

[0066] After completing the calculation, the controller sends a safety execution command to the corresponding hardware driver unit. To prevent mechanical shock to the electromechanical system caused by transient jumps, the controller connects a low-pass filter and a rate-of-change limiting algorithm in series at the end of the command. The system judges the flow difference between the previous and subsequent calculation cycles. If the calculated difference exceeds the acceleration boundary allowed by the drive inverter, the controller performs ramp limiting processing on the output frame based on the maximum angular acceleration constant. In this embodiment, the maximum angular acceleration constant is configured offline based on the rotor inertia of the feeder servo motor and the maximum peak torque allowed by the system. After smoothing processing, Converted into an analog current signal to drive the hydraulic proportional relief valve. The signal is converted into fieldbus messages to drive the servo frequency converter motor of the upstream feeder. The vector control logic and digital low-pass filtering algorithm of the servo frequency converter motor are well-known technologies in this field and will not be elaborated here. With the help of this pressure and flow coordinated control closed loop, the equipment avoids the degradation of chemical components of coenzyme Q10 caused by mechanical heat accumulation when dealing with compaction conditions with high density requirements.

[0067] In this embodiment, the physical structure arrangement of the pneumatic directional valve and the specific implementation process of the material rejection triggering logic are as follows.

[0068] After spectral and displacement monitoring, the solid dispersion strip-shaped flakes are conveyed to the discharge end. To achieve physical isolation between qualified and unqualified products, the equipment is equipped with a pneumatic reversing valve mechanism at the end of the feed chute. This mechanism includes an actuator flap, a double-acting cylinder, and a two-position five-way solenoid reversing valve. The actuator flap is hinged to the chute bifurcation and connected to the cylinder piston rod via a mechanical linkage. The control coil of the solenoid reversing valve is electrically connected to the digital output port of the controller. Under normal operating conditions, the cylinder remains retracted, the actuator flap blocks the waste recovery channel, and the material slides down to the qualified product collection container by gravity. When a rejection level command is received from the controller, the solenoid reversing valve reverses, high-pressure gas drives the cylinder piston to extend, and the actuator flap cuts into the main flow channel, guiding the strip-shaped flakes in a specific section to the waste recovery channel. For the cylinder sealing structure, the selection of the air source pressure regulating and filtration device, and the pneumatic circuit connection, those skilled in the art can refer to standard pneumatic automation atlases, which are well-known technologies in the field and will not be elaborated here.

[0069] When the preceding quality judgment module outputs a non-conforming pulse signal, the corresponding defective material is still below the monitoring station and has not yet reached the pneumatic reversing valve. To ensure that the physical rejection action corresponds to the spatial position of the defective material, the controller establishes a spatiotemporal translation tracking queue for rejection commands at the bottom layer. Based on the kinematic transmission principle, the controller extracts the physical trajectory length between the center of the spectral monitoring main optical axis and the tangent point of the flip-plate action, and calculates the transmission delay time of the rejection command by combining it with the linear velocity of the material conveying. ; In the formula, The transmission delay time of the elimination instruction calculated at the current moment; The trajectory length from the monitoring point to the tangent point of the reversing valve, which is fixed in the physical spatial layout of the equipment; The real-time linear velocity of the strip at the current moment; This is a preset velocity offset constant. As a preferred method, The value is 1×10 -5 This value avoids division overflow exceptions triggered when the linear velocity drops to zero due to equipment downtime during low-level calculations. After obtaining the delay time, the controller uses the initial rejection timestamp generated by the quality assessment module. With the above The variables are arithmetically summed to generate the theoretical action timestamp of the directional valve. The data is then pushed into a first-in-first-out (FIFO) digital queue to await execution. The technical purpose of this step is to eliminate the spatial difference between the monitoring point and the execution point, and to achieve dynamic synchronization between the material flow status and the control commands.

[0070] There is an inherent mechanical response time difference between cylinder charging and solenoid valve coil energization. If commands are issued strictly according to the theoretical action timestamp, it will lead to a delay, causing some defective materials to be missed or qualified materials to be mistakenly rejected. Based on these hardware delay characteristics, the controller introduces a compensation constant and pre- and post-safety redundancy to calculate the actual action time boundary of the directional valve. To simplify the underlying calculation logic, the controller directly uses the aforementioned theoretical action timestamp... Subtract the inherent mechanical response time And setting a safety lead time for spatial span The actual start time of issuing the high-level command can be calculated. Meanwhile, to ensure that the flapper opens before the defective material arrives at the head end and closes after it passes the tail end, the controller calculates the duration for which the solenoid valve remains open by adding twice the safety lead time and the mechanical response time to the passage time of the defective material itself. ; In the formula, This refers to the inherent mechanical response time of the combined pneumatic circuit and linkage mechanism; The safety lead time for the set spatial span; The duration for which the solenoid valve remains open; This refers to the cumulative total length of the physically sliced ​​strips that are continuously determined to be substandard. In this embodiment, The parameters are established and stored in registers based on offline calibration of the flipping action process using a high frame rate visual sensor; as a preferred method, The settings are in the 50-100 millisecond range to cover the displacement slippage of the material during its sliding process in the chute. The controller... Always output drive level, maintain The signal is canceled after a set time, and the cylinder resets under reverse air pressure. By constructing an execution window that includes mechanical delay compensation and redundancy tolerance, the system establishes action redundancy boundaries when eliminating materials in the phase change blind zone, maintaining a closed-loop quality system throughout the continuous manufacturing process.

[0071] In this embodiment, the specific implementation process of the online bypass circulation and secondary phase change mechanism for unqualified tablets is as follows.

[0072] After the substandard strip-shaped flakes are guided into the waste recycling channel by the pneumatic reversing valve, the controller activates the online bypass circulation mechanism to recover the effective active ingredients. The strip-shaped flakes entering the bypass channel have an irregular physical shape, and direct recirculation would cause bridging and blockage of the feed throat. The equipment is equipped with a low-shear pulverizer with built-in staggered toothed rollers in the middle of the bypass channel. The strip-shaped flakes undergo mechanical cutting during the fall, transforming into granular material and improving the flowability of subsequent conveying. The reduced-size granular material falls under gravity into a buffer hopper at the bottom equipped with a weighing sensor. For the adjustment of the physical gap of the pulverizing toothed rollers and the calibration of the bridge circuit of the weighing sensor, those skilled in the art can refer to the standard solid pulverizing equipment manual, which is well-known in the field and will not be elaborated here.

[0073] After a certain amount of recirculated material accumulates in the buffer silo, it needs to be remixed into the main feed stream in a certain proportion. Since the recirculated material has already undergone one compaction operation, its physical density and matrix melting state differ from the initial raw material. A fixed-proportion unidirectional recirculation would disrupt the steady-state rheological field within the main extrusion chamber. Based on the principles of mass conservation and system steady-state continuity, the controller constructs a multi-dimensional dynamic recirculation model that includes dynamic material level margin and main unit load redundancy. The transient current of the main drive motor is affected by physical friction and cutting, accompanied by high-frequency noise. Directly introducing this noise into the calculation would cause oscillations in the recirculation command. The controller performs a moving average digital filter on the real-time acquired transient current to obtain a smooth and effective operating current. Given that the effective operating current of the motor directly maps to the main extruder's work load, and the accumulated mass in the silo reflects the bypass blockage risk, the controller extracts the current actual accumulated mass in the buffer silo and the aforementioned effective operating current to calculate the target mass flow rate of the bypass recirculation at the current moment. ; In the formula, For the current moment The calculated target mass flow rate for bypass recirculation; The preset rated reference main feed mass flow rate for the controller; This is the maximum allowable basic reflux ratio coefficient for the system; The accumulated mass in the buffer silo is fed back in real time by the weighing sensor; The maximum safe loading capacity for the physical design of the buffer silo; The upper limit of the rated safe current of the main drive motor; This is the effective operating current of the motor after filtering at the current moment; and These are the dimensionless weighting coefficients corresponding to the material level margin and the main unit load redundancy, respectively. This is a preset current bias constant. As a preferred method, The value is 1×10 -4 This parameter avoids the abnormal division overflow caused by the denominator approaching zero due to no-load or current acquisition disconnection. In this embodiment, The value range is set to 0.05 to 0.15, with a dimensionless weighting coefficient. and The sum equals 1, with the specific value pre-tuned based on offline rheological hybrid testing. To prevent abnormally negative values ​​in the calculation results under host overload conditions, the controller performs non-negative boundary truncation on the above load redundancy term, when the effective operating current of the motor... Greater than or equal to the upper limit of the rated safe current At that time, this item is forcibly reset to zero. The technical purpose of this calculation step is to dynamically accelerate the emptying of the buffer hopper through multi-dimensional weighted judgment, while ensuring that the main extruder does not experience current overload, thus avoiding sudden increases or decreases in backflow caused by relying on a single extreme material level. After obtaining the target mass flow rate, the controller will... The driving frequency is converted and sent to the servo rotary feeder at the bottom of the buffer silo, and the valve body speed is adjusted to realize the physical quantitative feeding of particulate materials.

[0074] The refluxed particulate material has absorbed some mechanical energy during previous processing, and its internal amorphous conversion index has changed. When this material mixes with fresh raw materials and re-enters the pressure roller gap, maintaining the original pressure closed-loop command will cause a local excess of specific mechanical energy input, triggering secondary thermal degradation of coenzyme Q10. Based on the principle of thermodynamic energy superposition and compensation, the controller introduces a reflux ratio feedforward compensation algorithm before synthesizing the final execution pressure. The controller extracts the real-time main feed flow rate and the aforementioned reflux target flow rate, and calculates the compensated execution pressure command: ; In the formula, The actual pressure command issued after secondary phase change energy compensation; The comprehensive pressure command calculated by the preceding closed-loop module; This is the preset secondary phase transition energy attenuation coefficient; The actual bypass return quality flow rate injected at the current moment; Real-time feed mass flow rate of fresh basic raw materials; This is a preset flow rate bias constant. In this embodiment, The value range is set to 0.1 to 0.3. This coefficient is calibrated based on offline analysis data of the residual phase change enthalpy of primary compacted materials using differential scanning calorimetry. As a preferred method, The value is 1×10 -5This avoids the zero-point overflow anomaly in the denominator triggered by the dual interruption of virgin and recycled material flow. By extracting the dynamic mass ratio of recycled material in the total feed, this calculation step proportionally reduces the total mechanical work input to the system. Physically, this means that by proportionally deducting the redundant phase change energy carried by the recycled material, the mixture absorbs the total mechanical energy in the extrusion zone to align with the target crystal transformation threshold, maintaining thermodynamic balance in the continuous manufacturing process. After completing the command compensation, the controller will... It is converted into an analog drive signal and executed to complete the physical cycle and energy closed-loop control of unqualified tablet compression.

[0075] In this embodiment, the specific implementation process of the frequency conversion oscillation crushing and screening mechanism is as follows.

[0076] Qualified strip-shaped flakes enter the granulation section via a discharge chute. To achieve the target particle size distribution for solid dosage forms, the equipment is equipped with an integrated vibrating pulverizer and sieve unit with a variable frequency drive. This mechanism includes a crushing chamber with alternating moving and fixed teeth, an eccentric vibrating motor, and a multi-layer woven metal screen. Within the crushing chamber, the strip-shaped flakes are subjected to mechanical shearing and compression, breaking down into particles of mixed sizes. The eccentric vibrating motor drives the screen to generate compound vibrations, promoting the separation of the mixed particles through the screen according to their size. For the adjustment of the eccentric block angle, the screen tensioning structure, and the motor bearing lubrication components, those skilled in the art can refer to standard vibrating screening machinery manuals, as these are well-known technologies in the field and will not be elaborated upon here.

[0077] Relying on a fixed excitation frequency to handle fluctuations in the front-end output flow rate can lead to screen clogging or material friction heating. Given that the discharge mass flow rate determines the transient load on the screen, and the material's brittleness index affects its degree of fragmentation within the crushing chamber, an adaptive adjustment model for the excitation frequency is constructed based on vibration dynamics and the principle of particle penetration probability. The controller extracts the actual discharge mass flow rate of the extruder at the current moment and the preset material brittleness index to calculate the target operating frequency of the variable frequency excitation motor. ; In the formula, The target excitation frequency is calculated at the current moment; The preset fundamental excitation frequency; The frequency adjustment gain coefficient set for the system; Provides real-time mass flow rate output to the front-end extruder; The rated baseline discharge mass flow rate; Offline measurement of the brittleness index of coenzyme Q10 solid dispersion; The brittleness constant is determined based on this. and These are the dimensionless weighting coefficients corresponding to flow fluctuations and brittleness deviations, respectively. This is the preset flow rate bias constant; This is a preset brittleness bias constant. As a preferred method, and The value is 1×10 -5 This parameter combination avoids denominator zero-point overflow anomalies triggered by missing flow benchmarks or basic constants during equipment commissioning in the underlying calculations. In this embodiment, and The sum of these values ​​equals 1, and they are all set within the range of 0.4 to 0.6. The value range is set to 5 to 15 Hz, with the specific value determined based on offline orthogonal experiments of the material's angle of repose and sieve penetration rate. To prevent the calculated target excitation frequency from exceeding the inverter's drive limit, the controller incorporates specific measures in its underlying logic. The upper and lower boundary limits are implemented, ensuring that the vibration frequency is not lower than the minimum frequency required for the equipment to maintain basic screening and not higher than the rated maximum frequency of the motor. After extracting the dynamic material flow load and physical properties of the upstream and downstream, this calculation step achieves dynamic matching between the excitation work and the real-time operating conditions. The technical purpose is to increase the vibration frequency to accelerate screening when the material flow rate increases or the brittleness decreases, and vice versa, thereby avoiding over-crushing caused by constant power output while ensuring that the material does not undergo secondary thermal degradation.

[0078] After achieving the aforementioned boundary limiting, the controller converts the target operating frequency into an analog signal and sends it to the frequency converter to drive the excitation motor. Mechanical structures possess inherent resonance bands within specific frequency ranges; continuous operation within this range can lead to fatigue damage to the equipment structure. Conventional boundary clamping keeps the operating frequency close to the resonance critical point, but the risk of forced vibration amplification remains. Therefore, the controller incorporates resonance band avoidance logic with a safety margin into its underlying algorithm. The controller extracts the factory-preset lower limit value of the mechanical resonance frequency. With upper limit The value is then compared with the target excitation frequency. If the target excitation frequency falls within the preset resonance frequency range, the controller executes a proximity clamping algorithm. The controller calculates the absolute difference between the target excitation frequency and the upper and lower resonance limits, introducing a safety offset bandwidth. If the target frequency is close to the lower limit, the actual frequency command will be forcibly written to [the specified value]. If it is close to the upper limit, then force write to ; In this embodiment, The value range is set to 2 to 3 Hz, and the specific value is extracted offline based on the attenuation envelope of the whole machine sweep frequency vibration test. Through multi-dimensional variable dynamic matching and underlying electrical limiting, this step establishes a dynamic balance between improving the screening rate and reducing the heat generated by secondary mechanical work, thus maintaining the uniformity of the particle size distribution of the final solid material product.

[0079] In this embodiment, the specific implementation process of the qualitative and quantitative evaluation mechanism for the final morphology of dry-pressed solid dispersion particles is as follows.

[0080] The solid dispersion particles, after being screened by frequency conversion vibration, fall into the finished product collection container. These particles exhibit an irregular polyhedral structure, increasing the physical specific surface area of ​​the formulation. To assess the particle size uniformity, the controller extracts real-time detection data from an online laser diffractometer and calculates the particle size distribution span index. ; In the formula, The real-time particle size distribution span index is calculated at the current moment; , and These are the equivalent particle diameters corresponding to the current cumulative distribution reaching 90%, 10%, and 50%, respectively. This is a preset particle size bias constant. As a preferred method, The value is 1×10 -6 This avoids division overflow anomalies triggered by median particle size feedback being zero due to equipment initialization or test optical path obstruction. In this embodiment, The upper limit of the threshold is set to 1.5, a value pre-tuned based on orthogonal data from offline particle group repose angle and flowability tests. The purpose of this calculation step is to quantify the physical size concentration of particulate materials, avoiding variations in fill weight during subsequent tableting or capsule filling processes caused by a broad particle size distribution. For the laser diffraction particle size analyzer optical path calibration and ultrasonic dispersion components, those skilled in the art can refer to standard particle testing instrument manuals, as these are well-known techniques in the field and will not be elaborated upon here.

[0081] Besides macroscopic physical dimensions, the core indicator for solid dispersion particles lies in the physical dispersion state of coenzyme Q10 within the matrix material. Coenzyme Q10 needs to transform from its initial crystalline structure to an amorphous state to improve bioavailability. The controller, in conjunction with an online X-ray diffractometer, extracts the diffraction energy spectrum data of the material crystals and calculates the real-time relative crystallinity within the particles. ; In the formula, This represents the relative crystallinity within the particle, calculated at the current moment. The integral area of ​​the crystal characteristic peaks extracted from the diffraction pattern; The total integral area of ​​the diffraction pattern; This is a preset area offset constant. As a preferred method, The value is 1×10 -5This avoids triggering zero-point overflow anomalies in the denominator under conditions of low baseline noise and undetected amorphous halos. In this embodiment, the relative crystallinity threshold for the qualified finished product is set to no more than 5%, which is calibrated based on in vitro dissolution kinetic curves and accelerated stability test data. The technical significance of this calculation step is to quantitatively assess the degree of damage to the coenzyme Q10 crystal structure caused by the initial mechanical stress and phase transition work, and to confirm that the phase transition results at the molecular level of the formulation have achieved the expected results.

[0082] After verifying the microcrystalline form, the hydrodynamic properties of the formulation need to be established. Dry granulation relies on mechanical compaction to impart a specific microporous structure to the particles. This porosity maps to the capillary permeation efficiency of the solid formulation in the gastrointestinal dissolution medium. The controller extracts feedback data from the tap density meter and the true density analyzer to calculate the effective porosity of the particle structure. ; In the formula, The effective porosity of the particles is calculated at the current moment; This refers to the real-time measured particle compaction density. True density of the particle skeleton determined by helium displacement method; This is a preset density bias constant. As a preferred method, The value is 1×10 -4 To avoid the anomaly of zero true density readings triggering zero-point overflow in the denominator due to insufficient material in the testing chamber or sensor disconnection, this embodiment controls the effective porosity within the range of 15% to 30%. This range is established based on orthogonal experiments on particle friability and tableting properties. Too low a porosity leads to particle densification, causing dissolution inhibition; too high a porosity reduces particle mechanical strength, leading to packaging breakage.

[0083] After acquiring the aforementioned microscopic and macroscopic data, given the spatial span between the physical installation stations of the laser diffractometer, X-ray diffractometer, and true density analyzer, and the physical differences in the sampling and calculation lag times of each instrument, direct comparison of transient data could lead to batch misjudgment. Since the material after frequency-controlled oscillating sieving has been transformed into discrete particle form and no longer exhibits the physical linear velocity of ribbon-like sheets, the controller extracts the physical pipeline trajectory distance between each detection station. Combined with the particle apparent velocity calculated based on real-time mass flow rate and pipeline cross-sectional area, the controller calculates the operational transmission delay time between data sources. Based on this delay time, the controller establishes a timestamp alignment cache queue at the bottom layer, dynamically offsetting and aligning the span index, relative crystallinity, and effective porosity of the same physical batch of material. After alignment, the controller performs a logical AND operation on the three morphological indicators and preset release standards. The logical AND gate outputs a high-level release signal only when all three indicators of the same batch of material fall within their respective threshold ranges. All indicators met the standards, and the particle flow was determined to be a finished product and transferred to the downstream storage tank, completing the physical morphology verification and quality closed loop of continuous solid dispersion manufacturing.

[0084] In one embodiment, Figure 3 This is a schematic diagram illustrating the changes in the measured compaction density of a strip-shaped sheet recorded by the system described in this invention under a specific scenario: responding to a sudden, localized impact disturbance from high-density agglomerated material in the upstream feeding module. The diagram is a two-dimensional coordinate graph, with the horizontal axis representing the continuous operating time of the equipment and the vertical axis representing the equivalent compaction density, reflecting the microscopic porosity characteristics within the strip-shaped sheet. .

[0085] Reference Figure 3 The diagram contains the following element: a horizontal line segment marked as the target preset density. This line segment represents a preset compliance target threshold (calibration parameter) for the compaction density of a strip of material. For example, 1.2g / cm 3 This is used to ensure the effective porosity of the final solid dosage form. Within the optimal dissolution range. A dashed line, marked for the traditional PID control method. This curve represents the equivalent compaction density of the strip in a system that does not employ the multi-source spatiotemporal alignment and feedforward compensation method described in this invention. The process of change. Before time T1, the value of the dashed line stabilizes near the target preset density threshold. After time T1, due to the lag feedback relying solely on downstream thickness measurement data, the extrusion pressure fails to respond in advance, and its value rises rapidly, exceeding the tolerance threshold defined by the target density for a considerable period (the highest peak exceeding 1.35 g / cm³). 3After undergoing multiple damped oscillations, it slowly recovers to within the threshold dead zone. A solid line marks the feedforward and feedback combined control method of this invention. This curve represents the equivalent compaction density of the strip in the system employing the method described in this invention. The change process. Throughout the entire time axis, the values ​​of the solid line are effectively clamped within the extremely narrow tolerance band allowed by the target preset density. Near time T1, the values ​​of the curve show only a very small perturbation (e.g., the peak value is limited to 1.23 g / cm³). 3 (Internal), relying on the comprehensive output pressure commands issued by the system in real time With dynamic compensation coefficient The combined effect of these factors then quickly converged and returned to stability.

[0086] Figure 3 The data also marks two specific time points: Time point T0: represents the instantaneous value of the propulsion resistance torque collected by the torque sensor inside the feeding module. A significant step change occurs, and the rate of change of propulsion drag torque calculated by the controller... The physical warning time for exceeding a set threshold (corresponding to abnormal movement of high-density materials). The method described in this invention initiates feature analysis after time T0 and calculates the feedforward base pressure command based on the mass flow rate difference. Time point T1: This represents the moment when the high-density local premixed material captured by the feature at time T0 actually reaches the extrusion module and is bitten into the working area by the pressure roller after being propelled by the physical displacement of the feed screw shaft.

[0087] The figure illustrates that, when addressing the technical challenge of a sudden impact from high-density agglomerated material upstream, the traditional control method (shown by the dashed line) suffers from transient over-compaction and formulation porosity collapse (significantly excessive compaction density) after the impact load actually reaches the pressure roller (after T1) due to a lack of predictability and physicorheological feedforward mechanisms. In contrast, the method of this invention (shown by the solid line) provides early warning by capturing the instantaneous first derivative of the feed torque at upstream time T0 and performs predictive pressure regulation and coordinated control based on the underlying timestamp tracking queue. This ensures that the pressure roller's extrusion pressure has already adaptively reduced when the impact load actually reaches the core extrusion zone (at time T1) and afterward, resulting in a significantly reduced equivalent compaction density of the strip sheet. It always stays near the reference plane defined by the target calibration parameters, thus solving the technical problem that existing technologies cannot effectively cope with physical flow variations that lead to product batch scrapping in continuous manufacturing.

Claims

1. An intelligent parameter control system for the dry extrusion granulation process of coenzyme Q10, characterized in that, Including the controller, and: The feeding module is equipped with a feeding screw shaft and a torque sensor. The torque sensor collects the propulsion resistance torque and provides the controller with a feedforward prediction signal. The extrusion module includes pressure rollers arranged opposite each other, the extrusion force of which is dynamically adjusted by the controller; The monitoring module, including a spectral probe and a laser displacement sensor, is located on the discharge side of the extrusion module to collect spectral and thickness data of the strip sheet for the controller to calculate the volume-compensated amorphous conversion index. The reversing module includes a pneumatic reversing valve and a circulation channel connected to the feeding module; The pulverizing module is located below the output end of the pneumatic reversing valve; The controller adjusts the extrusion pressure of the pressure roller based on the feedforward prediction signal and the volume-compensated amorphous conversion index, and controls the pneumatic reversing valve to guide the strip sheet to the crushing module or the circulation channel according to the volume-compensated amorphous conversion index.

2. The intelligent parameter control system for the coenzyme Q10 dry extrusion granulation process according to claim 1, characterized in that, The feeding module also includes a servo feeding hopper located at the bottom of the feeding screw shaft; The torque sensor is coaxially mounted on the drive shaft end of the feed screw shaft; The controller acquires the feedforward prediction signal through the following process: The raw torque data from the torque sensor is acquired and resampling based on timestamp alignment is performed. The resampled data is smoothed using a moving average filtering algorithm, and the transient rate of change of the smoothed propulsion resistance torque over time is calculated.

3. The intelligent parameter control system for the dry extrusion granulation process of coenzyme Q10 according to claim 1, characterized in that, The intelligent parameter control system also includes a feed chute with a forked opening, which is located on the discharge side of the extrusion module to receive the strip sheet. The spectral probe is fixed above the feed chute by a shockproof bracket; The laser displacement sensor includes an upper sensor and a lower sensor, which are installed opposite each other on the upper and lower sides of the feed chute, and the probe end faces of the upper sensor and the lower sensor maintain a fixed reference distance.

4. The intelligent parameter control system for the coenzyme Q10 dry extrusion granulation process according to claim 1, characterized in that, The controller performs the following procedure to calculate the volume-compensated amorphous conversion index: The real-time thickness of the strip is calculated based on the distance data collected by the upper sensor and the lower sensor, as well as the fixed reference distance. The real-time equivalent density of the strip sheet is calculated based on the real-time thickness, the real-time mass flow rate converted from the rotational speed of the feed screw shaft, and the operating parameters of the pressure roller. Extract the real-time integral area of ​​the spectral data in the preset characteristic peak range, and use the real-time equivalent density to perform volume normalization compensation on the real-time integral area to obtain the volume-compensated characteristic peak area. The volume-compensated amorphous conversion index is calculated based on the area of ​​the volume-compensated characteristic peak and the pre-stored pure crystalline reference parameters.

5. The intelligent parameter control system for the coenzyme Q10 dry extrusion granulation process according to claim 1, characterized in that, The controller adjusts the pressing force of the pressure roller through the following logic: The deviation between the real-time mass flow rate and the preset rated mass flow rate is calculated, and the feedforward prediction signal is introduced to perform advance correction on the deviation. The feedforward base pressure command is generated by combining the feedforward gain coefficient. The deviation between the preset target conversion index and the volume-compensated amorphous conversion index is calculated, and the feedback pressure adjustment amount is generated through PID calculation. A dynamic compensation coefficient is generated based on the difference between the real-time equivalent density and the preset target compaction density. The feedforward base pressure command and the feedback pressure adjustment amount are superimposed, and the superimposed result is multiplied by the dynamic compensation coefficient to generate the final pressure command and send it to the extrusion module.

6. The intelligent parameter control system for the coenzyme Q10 dry extrusion granulation process according to claim 1, characterized in that, The pressure roller integrates a cooling water circulation unit, and the controller performs thermodynamic coupling control of the extrusion process. The specific mechanical energy input of the material is calculated based on the final pressure command, the rotational speed of the pressure roller, and the real-time mass flow rate, and the transient temperature rise of the material is estimated. When the estimated temperature exceeds the preset thermal degradation critical threshold, a temperature suppression coupling coefficient is generated, and the final pressure command and the feed screw shaft speed command of the feeding module are reduced proportionally according to the temperature suppression coupling coefficient, while the target cooling water flow rate of the cooling water circulation unit is increased.

7. The intelligent parameter control system for the coenzyme Q10 dry extrusion granulation process according to claim 1, characterized in that, The pneumatic reversing valve includes an actuating flap, a double-acting cylinder, and a solenoid reversing valve; the actuating flap is hinged to the bifurcation of the guide chute and connected to the double-acting cylinder via a mechanical linkage. The controller calculates the transmission delay time of the rejection command based on the physical trajectory length between the spectral probe and the execution flap and the real-time linear velocity of the strip. When the volume-compensated amorphous conversion index is lower than the preset release threshold, the controller drives the electromagnetic reversing valve to operate, so that the execution flap introduces the unqualified material into the circulation channel.

8. The intelligent parameter control system for the coenzyme Q10 dry extrusion granulation process according to claim 1, characterized in that, The circulation channel is equipped with a low-shear shredder and a buffer hopper with a weighing sensor. The controller calculates the return mass flow rate in the circulation channel based on the accumulated mass fed back in real time from the buffer silo and the operating current of the main motor of the extrusion module. Based on the mass ratio of the reflux mass flow rate in the total feed, a preset secondary phase change energy attenuation coefficient is introduced to reduce the compensation for the command to adjust the extrusion pressure of the pressure roller.

9. The intelligent parameter control system for the coenzyme Q10 dry extrusion granulation process according to claim 1, characterized in that, The crushing module is equipped with a crushing chamber, an eccentric vibrating motor, and multiple layers of metal screens. The controller calculates the target operating frequency of the eccentric vibrating motor based on the real-time output mass flow rate derived from the rotational speed of the pressure roller and a preset material brittleness index, and executes resonance zone avoidance logic: If the target operating frequency falls within the preset mechanical resonance frequency range, the target operating frequency will be clamped to the safe frequency band outside the edge of the mechanical resonance frequency range.

10. A control method for an intelligent parameter control system based on the coenzyme Q10 dry extrusion granulation process according to any one of claims 1-9, characterized in that, Includes the following steps: The rate of change of the propulsion resistance torque of the feed screw shaft is obtained by the controller, and the rate of change of the propulsion resistance torque is used as a feedforward prediction signal. During the extrusion stage, the mechanical work absorbed per unit mass of the material is calculated based on the extrusion pressure, rotation speed, and feed rate of the pressure rollers to determine whether the material has the conditions for solid-state phase change. Spectral and thickness data of the strip-shaped thin sheet were collected simultaneously, and the volume-compensated amorphous conversion index, excluding density interference, was calculated. By integrating the feedforward prediction signal with the volume compensation amorphous conversion index, the controller performs feedforward and feedback joint pressure regulation control on the pressure roller extrusion pressure. Based on the comparison result between the volume compensation amorphous conversion index and the preset release threshold, the pneumatic reversing valve is controlled to perform qualified product release or unqualified product return circulation.

Images