Automatic control method and system for constant-temperature bottle hanging type freeze dryer

Abstract

The invention relates to the technical field of freeze dryer control, and discloses an automatic control method and system for a constant-temperature bottle hanging type freeze dryer. The method comprises the following steps: acquiring multi-dimensional parameter data such as temperature gradient, vacuum degree fluctuation and material ice crystal form in the freeze dryer, performing real-time synchronous calibration through a multi-source data fusion engine, and generating an environment state matrix; and constructing a freeze-drying stage dynamic division model based on the matrix, accurately identifying the switching critical points of the pre-freezing stage, the primary drying stage and the secondary drying stage, and outputting and activating a corresponding temperature control strategy and a vacuum adjustment strategy according to the model. The temperature control strategy comprises a gradient cooling curve and a heating rate constraint, and the vacuum adjustment strategy comprises a pressure threshold sequence and a leakage compensation mechanism. The deviation between the ice crystal form of the material and the preset template is monitored through the closed-loop feedback unit, control strategy execution parameters are dynamically corrected, refined intelligent control over the freeze-drying process is achieved, and the stability of a freeze-dried product is guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of freeze-dryer control, in particular to an automatic control method and system for a constant-temperature bottle-hanging freeze-dryer. BACKGROUND

[0002] Freeze-drying technology is widely used in medicine, biological products, food processing and other fields due to its advantages of maximizing the preservation of active ingredients and extending the shelf life. The control precision of the constant-temperature bottle-hanging freeze-dryer, as the core equipment of this technology, is directly related to the quality of the final product. Currently, the control method of traditional freeze-dryers relies on pre-set fixed programs and lacks dynamic adaptation to the complex internal environment of the equipment, resulting in many problems that need to be solved.

[0003] In the parameter collection link, existing technologies often only focus on single or a few indicators such as temperature and vacuum degree, ignoring key parameters such as temperature gradient distribution, vacuum degree instantaneous fluctuation and material ice crystal morphology change. This one-sided parameter collection method cannot fully reflect the real environment state inside the freeze-dryer, making the subsequent control decision lack complete data support. For example, uneven temperature gradient will cause differences in freezing rate of materials in different bottles, affecting the uniformity of ice crystal formation, and ultimately leading to inconsistent sublimation rate of water in the drying stage, resulting in incomplete drying or over-drying of some materials.

[0004] In terms of stage division, traditional methods usually define the switching time of pre-freezing, primary drying and secondary drying according to fixed time nodes or single parameter thresholds. However, the physicochemical properties of different materials differ significantly, and even the same material may change in ice crystal growth and sublimation characteristics due to factors such as initial state and environmental disturbance during different batches of processing. The fixed stage division method cannot adapt to such dynamic changes, often resulting in premature or late stage switching. If the pre-freezing stage does not reach a low enough temperature, the water in the material cannot completely freeze to form stable ice crystals, leading to defects such as bubbling and shrinkage in the drying stage; if the primary drying stage ends prematurely, residual free water will affect the subsequent adsorbed water removal effect, and vice versa, increasing energy consumption and processing time.

[0005] In the control strategy execution layer, the existing temperature control mostly adopts a single constant temperature or linear temperature rising and falling mode, without considering the specific requirements of different freeze-drying stages for temperature change rate, which may easily cause damage to the material structure; the vacuum regulation strategy lacks effective leakage compensation mechanism, when a small leakage occurs in the equipment, the vacuum degree cannot be maintained in the set range, directly affecting the water sublimation efficiency. At the same time, the feedback regulation mechanism of the traditional control method is not perfect, which can only be adjusted passively after obvious quality problems occur, and cannot actively correct the control parameters according to the real-time changes of the material state, resulting in insufficient stability and reliability of the entire freeze-drying process, which is difficult to meet the freeze-drying requirements of high-quality materials. SUMMARY

[0006] The purpose of the present application is to provide an automatic control method and system for a constant-temperature hanging bottle type freeze dryer to solve the problems raised in the background art.

[0007] To achieve the above-mentioned purpose, the present application provides an automatic control method for a constant-temperature hanging bottle type freeze dryer, which comprises: Collecting multi-dimensional parameter data of the internal environment of the freeze dryer, the multi-dimensional parameter data including temperature gradient data, vacuum degree fluctuation data and material ice crystal morphology data; Calibrating the multi-dimensional parameter data in real time through a multi-source data fusion engine to generate an environmental state matrix after calibration; Based on the environmental state matrix after calibration, a freeze-drying stage dynamic division model is constructed, which is used to identify the switching critical points of the pre-freezing stage, the primary drying stage and the secondary drying stage; According to the output of the freeze-drying stage dynamic division model, the corresponding temperature control strategy and vacuum regulation strategy are activated, the temperature control strategy including gradient temperature drop curve and temperature rise rate constraint, and the vacuum regulation strategy including pressure threshold sequence and leakage compensation mechanism; Through a closed-loop feedback unit, the deviation of the material ice crystal morphology data from the preset morphology template is monitored in real time, and the execution parameters of the temperature control strategy and the vacuum regulation strategy are dynamically corrected.

[0008] Preferably, when the multi-source data fusion engine performs data synchronization calibration, a space-time alignment algorithm is used to eliminate the time delay difference between the temperature gradient data and the vacuum degree fluctuation data, and the periodic characteristics of the material ice crystal morphology data are extracted through frequency domain decomposition; If the correlation between the temperature gradient data and the vacuum degree fluctuation data is lower than the dynamic threshold, a redundant data elimination mechanism is triggered, and the effective data segment of the high-frequency sampling point is retained; The effective data segment is input into an environmental state matrix generator to output a three-dimensional environmental state matrix with time stamp markers.

[0009] Preferably, the freeze-drying stage dynamic division model identifies the switching critical point from the pre-freezing stage to the primary drying stage through an ice crystal phase transition detector, which makes judgments based on the light transmittance mutation and surface tension decay rate of the ice crystal morphology data of the material; When the ice crystal phase transition detector outputs a phase switching signal, the temperature control strategy of the primary drying stage is activated, and the gradient cooling curve of the temperature control strategy is adaptively matched with the maximum allowed temperature rise slope according to the type of the material; During the duration of the primary drying stage, the cavity pressure is gradually reduced through a pressure threshold sequence of the vacuum adjustment strategy, and the vacuum fluctuation caused by the sublimation of the material is dynamically compensated according to a leakage compensation mechanism.

[0010] Preferably, the closed-loop feedback unit starts a residual moisture monitoring mode in the secondary drying stage, obtains the real-time dielectric constant of the material through a dielectric constant sensor, and compares it with the dielectric constant threshold of the preset morphology template; If the real-time dielectric constant continuously exceeds the dielectric constant threshold, a drying termination instruction is generated, and an emergency pressure relief protocol of the vacuum adjustment strategy is triggered; The drying termination instruction is sent to the temperature control strategy executor synchronously, and the subsequent execution of the gradient cooling curve is forcibly interrupted.

[0011] Preferably, after the gradient cooling curve is interrupted, the temperature control strategy executor starts a cooling rate buffering mechanism to linearly decrease the power of the heating plate until shutdown; The decreasing slope of the cooling rate buffering mechanism is dynamically adjusted according to the thermal conductivity coefficient of the material, and the adjustment process is constrained by the current pressure value of the vacuum adjustment strategy; When the power of the heating plate decreases below a safety threshold, a cavity sealing self-checking process is triggered, which is completed by comparing the difference between the historical compensation amount of the leakage compensation mechanism and the theoretical leakage amount.

[0012] Preferably, if the difference exceeds a fault tolerance upper limit, the cavity sealing self-checking process generates a sealing failure alarm and records an abnormal pressure fluctuation interval; The sealing failure alarm triggers the stage rollback operation of the freeze-drying stage dynamic division model, forcibly resets the current stage to the primary drying stage, and reloads the corresponding temperature control strategy and vacuum adjustment strategy; During the execution of the stage rollback operation, the dielectric constant monitoring function of the closed-loop feedback unit is disabled until the pressure fluctuation interval returns to the range of the theoretical leakage amount.

[0013] Preferably, after the stage rollback operation is completed, the effectiveness of the rollback is verified through the real-time light transmittance of the ice crystal morphology data of the material; If the light transmittance does not reach the preset standard of the first drying stage, an emergency drying protocol is started, which runs in a constant temperature and constant pressure mode until the light transmittance reaches the standard; The execution time length of the emergency drying protocol is limited by the remaining buffer time length of the cooling rate buffer mechanism, and after the timeout, the system shutdown instruction is triggered.

[0014] Preferably, after the system shutdown instruction is activated, the mechanical locking device of the bottle hanging support is started to prevent the displacement of the material bottle due to pressure sudden change; The locking state of the mechanical locking device is verified by the gravity sensor, and if the weight distribution of the material bottle is detected to be abnormal, the pressure relief operation of the vacuum regulation strategy is delayed; During the delay pressure relief, the temperature control strategy executor maintains the minimum heating power to avoid the sudden rise of the material temperature.

[0015] Preferably, the maintenance time length of the minimum heating power is proportional to the abnormal duration of the gravity sensor, and the proportional coefficient is dynamically configured according to the heat sensitivity level of the material; When the gravity sensor returns to normal reading, the mechanical locking device is preferentially released, and the final pressure relief of the vacuum regulation strategy is executed; After the final pressure relief is completed, a freeze-drying process end signal is generated, and the data acquisition channel of the multi-source data fusion engine is closed.

[0016] Preferably, the present application also includes an automatic control system of a constant temperature bottle hanging freeze dryer, which comprises a memory, a processor and a computer program stored in the memory and running on the processor, and the processor implements the steps of the automatic control method of the constant temperature bottle hanging freeze dryer when executing the computer program.

[0017] Compared with the prior art, the present application has the following beneficial effects: Through comprehensive collection of multi-dimensional parameter data, the limitation of single parameter monitoring in traditional control is broken, and the dynamic changes of the internal environment of the freeze dryer and the state characteristics of the material can be completely captured. The introduction of temperature gradient data can clearly reflect the temperature distribution difference of different regions in the device, providing direction for subsequent precise temperature control; real-time acquisition of vacuum degree fluctuation data can timely discover abnormal changes of the vacuum system; and material ice crystal morphology data as a key indicator directly reflecting the freeze-drying effect can intuitively show the progress of the freeze-drying process. The cooperative collection of multi-dimensional parameters builds a comprehensive freeze-drying process monitoring system, providing rich and comprehensive basic information for subsequent control decisions.

[0018] The application of multi-source data fusion engine realizes the real-time synchronous calibration of different types of parameter data. Due to the differences in measurement principle and response speed of different sensors, the directly collected data may have deviations or be out of sync. Through fusion calibration, these errors can be effectively eliminated to generate a more accurate environmental state matrix. This matrix can accurately depict the internal environment of the freeze dryer, providing a reliable data basis for accurate phase division in the freeze-drying stage and avoiding control decision errors caused by inaccurate data.

[0019] The construction of the freeze-drying stage dynamic division model breaks free from the constraints of the traditional fixed phase division method and can accurately identify the switching critical point of different freeze-drying stages based on the real-time environmental state matrix. This model fully considers the synergistic effects of temperature, vacuum degree, and material state, and can adaptively adjust the phase switching timing based on the changes in ice crystal morphology and the dynamic fluctuations of environmental parameters, ensuring that each freeze-drying stage is carried out under optimal conditions. For the pre-freezing stage, it can ensure that the material is completely frozen to form a stable ice crystal structure; for the primary drying stage, it can switch to secondary drying in time before the ice crystals completely sublimate; for the secondary drying stage, it can accurately control the removal of adsorbed water, making the entire freeze-drying process smoother and more reasonable.

[0020] For the exclusive temperature control strategy and vacuum regulation strategy activated in different freeze-drying stages, the control method is accurately matched with the requirements of the stage. The gradient temperature curve in the temperature control strategy can gradually reduce the temperature according to the needs of the pre-freezing stage, avoiding damage to the material structure caused by rapid temperature drop; the temperature rise rate constraint can control the temperature rise speed in the drying stage, ensuring the smooth progress of the water sublimation and resolution process. The pressure threshold sequence in the vacuum regulation strategy can set a reasonable pressure range according to the vacuum degree requirements of different stages to ensure the water sublimation efficiency; the leakage compensation mechanism can respond to small leaks in the vacuum system in time to maintain the stability of the vacuum degree and avoid the adverse effects of vacuum degree fluctuations on the freeze-drying process.

[0021] The setting of the closed-loop feedback unit forms a complete control loop of "collection-calibration-decision-execution-monitoring-correction". By monitoring the deviation of the material ice crystal morphology data from the preset morphology template in real time, problems that occur during the freeze-drying process can be detected in time, and the execution parameters of the temperature control strategy and the vacuum regulation strategy can be dynamically corrected. This dynamic correction mechanism can quickly respond to various changes in the freeze-drying process, adjust the control parameters in time, and ensure that the freeze-drying process always proceeds towards the preset target, effectively avoiding freeze-drying effect deviations caused by external disturbances or internal parameter drifts, and ensuring the quality stability of the final freeze-dried product. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 Figure 1 is a freeze-drying process temperature and vacuum degree trend graph for a constant-temperature hanging bottle freeze dryer; Figure 2 A flowchart illustrating the dynamic division of the freeze-drying stage and the control of primary drying; Figure 3 A flowchart for the self-test and phased rollback of sealing performance; Figure 4 A comparison chart showing the theoretical leakage rate and the upper limit of tolerance for freeze dryers with different chamber volumes. Detailed Implementation

[0023] The technical solutions of 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.

[0024] Please see Figure 1 This invention provides an automated control method for a constant-temperature hanging-bottle freeze dryer. The method includes: achieving precise control through integrated multi-dimensional parameter acquisition, data fusion, stage identification, and dynamic adjustment. Multi-dimensional parameter data of the freeze dryer's internal environment are collected, including temperature gradient data, vacuum fluctuation data, and material ice crystal morphology data. A multi-source data fusion engine performs real-time synchronous calibration of the multi-dimensional parameter data, generating a calibrated environmental state matrix. Based on the calibrated environmental state matrix, a dynamic segmentation model for freeze-drying stages is constructed, which identifies the switching critical points between the pre-freezing stage, the primary drying stage, and the secondary drying stage. The corresponding temperature control strategy and vacuum regulation strategy are activated according to the output of the dynamic segmentation model. The temperature control strategy includes a gradient cooling curve and a heating rate constraint, while the vacuum regulation strategy includes a pressure threshold sequence and a leakage compensation mechanism. A closed-loop feedback unit monitors the deviation between the material ice crystal morphology data and a preset morphology template in real time, dynamically correcting the execution parameters of the temperature control strategy and the vacuum regulation strategy.

[0025] Example 1: When performing data synchronization calibration, the multi-source data fusion engine employs a spatiotemporal alignment algorithm to eliminate the acquisition delay difference between temperature gradient data and vacuum fluctuation data. This algorithm achieves temporal consistency of multi-sensor data through timestamp interpolation and spatial coordinate mapping. Its core lies in establishing a unified time reference coordinate system, mapping sensor readings distributed at different locations within the freeze dryer cavity to the same spatiotemporal framework. Temperature gradient data originates from a high-precision thermocouple array arranged on the heating plate, material tray, and cavity wall. These thermocouples record temperature changes at millisecond intervals, forming a three-dimensional temperature field. Vacuum fluctuation data is acquired by a capacitive thin-film vacuum gauge installed near the cavity's evacuation port; the measured data reflects the dynamic pressure within the system. Due to the inherent delay between the thermocouple's thermal inertial response and the vacuum gauge's electrical signal response, the acquisition delay difference manifests as a time axis offset in the data sequence. The spatiotemporal alignment algorithm compensates for this offset using a bidirectional interpolation method. Specifically, it performs forward prediction interpolation on the temperature gradient data sequence and backward correction interpolation on the vacuum fluctuation data sequence, aligning the effective timestamps of both to the synchronization point defined by the system's master clock. Spatial coordinate mapping transforms the coordinates based on the physical position of the sensor within the cavity, assigning each data point three-dimensional spatial coordinates to form a data point cloud with spatiotemporal attributes.

[0026] The periodic characteristics of ice crystal morphology data in the material are extracted through frequency domain decomposition. This data is acquired using a multimodal sensor system comprised of a near-infrared spectrometer and a high-speed camera. The near-infrared spectrometer monitors changes in the absorption rate of ice crystals at specific wavelengths, while the high-speed camera captures the macroscopic morphological evolution of the ice crystals. The frequency domain decomposition process uses Fast Fourier Transform (FFT) to convert the ice crystal morphology signal from the time domain to the frequency domain, identifying the periodic oscillation modes inherent in the signal. Ice crystal growth and sublimation processes generate characteristic frequency components, which are directly related to the physicochemical changes during the freeze-drying stage. The spectral data processed by FFT undergoes bandpass filtering to retain characteristic frequency bands related to ice crystal phase transitions and filter out high-frequency interference caused by equipment vibration and environmental noise. The periodic feature extraction module calculates characteristic parameters such as the dominant frequency amplitude, harmonic components, and spectral entropy, which constitute the frequency domain fingerprint of the ice crystal's dynamic behavior. The frequency domain decomposition results, along with spatiotemporally aligned temperature gradient data and vacuum fluctuation data, are fed into a feature fusion layer, providing multi-dimensional feature inputs for subsequent correlation analysis. A redundant data removal mechanism is triggered when the correlation between temperature gradient data and vacuum fluctuation data falls below a dynamic threshold. The correlation calculation employs an improved grey relational analysis method, which can handle small sample data and has no specific requirements on data distribution type. The dynamic threshold is adaptively adjusted based on the statistical characteristics of a historical data window, which contains multi-dimensional parameter data from the most recent N sampling periods. The dynamic threshold calculation involves the mean, variance, and autocorrelation function of the data within the window. A correlation below the dynamic threshold indicates a misalignment between temperature and pressure field changes, potentially stemming from anomalies such as sensor drift, local hotspots, or vacuum leaks. After the redundant data removal mechanism is activated, a sliding window analysis is performed on the original sampled data stream, with the window size matched to the sensor sampling frequency. The redundant data removal mechanism calculates the local outlier for each data point; data segments with outliers exceeding a set threshold are marked as redundant data. The selection of valid data segments for high-frequency sampling points is based on the information entropy criterion, retaining data segments with higher entropy values ​​to ensure information integrity. The removal process employs a multi-verification strategy, combining temporal domain consistency and spatial domain continuity checks to avoid mistakenly deleting valid data.

[0027] The valid data segments are input to the environmental state matrix generator, a dedicated data processing module based on tensor operations. Its architecture includes a data normalization layer, a dimension alignment layer, and a matrix synthesis layer. The data normalization layer performs Z-score normalization on temperature gradient data, vacuum fluctuation data, and material ice crystal morphology data to eliminate the influence of dimensional differences. The dimension alignment layer uses filling and pruning operations to ensure that multi-source data have the same spatiotemporal dimensions. Filling employs a mirror-symmetric algorithm, and pruning is based on important region detection. The matrix synthesis layer organizes the normalized and aligned data into a three-dimensional environmental state matrix. The three dimensions of the matrix correspond to time series, spatial location, and parameter type, respectively. High-precision clock signals are used for timestamp marking, and each matrix block carries the absolute time information of the generation moment. The data structure design of the three-dimensional environmental state matrix supports fast slice querying and batch processing. The matrix elements are arranged according to the principle of spatiotemporal locality, with adjacent storage units corresponding to data at adjacent spatiotemporal points. When the environmental state matrix generator outputs the matrix, it attaches an integrity check code. The check code is generated using a cyclic redundancy check algorithm to ensure the reliability of data transmission.

[0028] The calibration process of the multi-source data fusion engine adopts a pipelined architecture, with temperature gradient data, vacuum fluctuation data, and material ice crystal morphology data entering the preprocessing channel in parallel. The preprocessing channel includes signal conditioning circuitry and digital filters. The signal conditioning circuitry amplifies and isolates the analog sensor signals, while the digital filters employ finite impulse response (FIR) filters to eliminate power frequency interference. The preprocessed data is then sent to the core processing unit, which is equipped with a multi-core digital signal processor (DSP). The DSP executes the spatiotemporal alignment algorithm and the frequency domain decomposition algorithm in parallel. The time delay difference elimination module of the spatiotemporal alignment algorithm integrates a Kalman predictor. The Kalman predictor builds a state-space model based on historical sensor data to predict data values ​​at future moments, thereby reducing calibration errors. The periodic feature extraction module of the frequency domain decomposition includes a spectrum analysis coprocessor, which specializes in handling fast Fourier transform and inverse transform operations. A redundant data removal mechanism runs on a separate coprocessor, which calculates data correlation in real time and monitors dynamic threshold changes. The environmental state matrix generator directly accesses the processor cache, which stores the latest calibration data from several seconds prior for matrix construction. The entire multi-source data fusion engine employs a double buffering mechanism to ensure uninterrupted data acquisition and processing.

[0029] The hardware implementation of the multi-source data fusion engine is based on a field-programmable gate array (FPGA) platform. The FPGA chip is configured as a multi-channel data acquisition and processing system. The temperature gradient data channel includes a 24-bit analog-to-digital converter (ADC) and an isolation amplifier. The ADC sampling rate is programmable to adapt to the needs of different freeze-drying stages. The vacuum fluctuation data channel employs a differential input structure and a software-programmable gain amplifier to enhance the accuracy of small-signal measurements. The material ice crystal morphology data channel integrates a gigabit Ethernet interface and an image processing unit, which calculates ice crystal morphology characteristic parameters in real time. The FPGA internally incorporates dedicated logic circuitry to implement a spatiotemporal alignment algorithm, including a timestamp management module and a coordinate transformation engine. The frequency domain decomposition function is implemented using a Fast Fourier Transform (FFT) proprietary core, which supports configurable points and window functions. The environmental state matrix generator is mapped to the FPGA's block memory resources. The block memory is organized as a multi-port random access memory architecture, supporting high-speed matrix operations. The multi-source data fusion engine communicates with the host computer via an industrial Ethernet interface, uploading the calibrated environmental state matrix and system state information.

[0030] The multi-source data fusion engine runs a real-time operating system at the software level, which schedules data acquisition, calibration calculation, and matrix generation tasks. Data acquisition tasks have the highest priority, and their cycles are strictly synchronized with the sensor sampling cycles. Calibration calculation tasks include spatiotemporal alignment threads, frequency domain decomposition threads, and correlation analysis threads, which employ a priority inheritance protocol to avoid priority inversion. Matrix generation tasks run in the background, triggered when cached data reaches a set capacity. The multi-source data fusion engine's configuration parameters are stored in non-volatile memory, including sensor calibration coefficients, alignment algorithm parameters, and dynamic threshold calculation formulas. Upon system startup, configuration parameters are automatically loaded, and a self-test process is executed to verify the responsiveness of each sensor channel and the functional integrity of the algorithm modules. During operation, the multi-source data fusion engine continuously monitors its own status. Upon detecting hardware failures or algorithm anomalies, an alarm signal is triggered and transmitted to the system monitoring center via a digital output interface.

[0031] The calibration accuracy of the multi-source data fusion engine is affected by environmental factors, including electromagnetic interference (EMI), temperature drift, and mechanical vibration. EMI suppression employs shielded twisted-pair cable transmission of sensor signals, with a magnetic ring filter added to the signal line path. Temperature drift compensation uses a digital temperature sensor to monitor the circuit board temperature; the sensor readings are used to correct the amplifier bias voltage. Mechanical vibration effects are eliminated by installing a vibration-damping base and flexible connectors. The multi-source data fusion engine periodically performs an online self-calibration process, which uses a built-in standard signal source to verify the gain and bias parameters of each channel. The standard signal source generates precise voltage and frequency signals, which are switched to each input channel via a multiplexer. The self-calibration results automatically update the sensor calibration coefficients, maintaining the long-term measurement accuracy of the multi-source data fusion engine.

[0032] Example 2: See Figure 2The freeze-drying stage dynamic segmentation model identifies the critical point of transition from the pre-freezing stage to the primary drying stage using an ice crystal phase transition detector. The ice crystal phase transition detector operates on an environmental state matrix generated by a multi-source data fusion engine. This matrix includes temperature gradient data, vacuum fluctuation data, and material ice crystal morphology data, all processed through spatiotemporal alignment and frequency domain decomposition. The ice crystal phase transition detector integrates a transmittance analysis module and a surface tension calculation unit. The transmittance analysis module processes the raw light intensity signal from an optical sensor. The optical sensor uses a laser light source of a specific wavelength to penetrate the material layer, and a photodetector receives the transmitted light and converts it into an electrical signal. A transmittance abrupt change detection algorithm calculates the difference in transmittance between adjacent sampling periods in real time. When the difference exceeds a preset abrupt change threshold, a phase transition flag is triggered. The surface tension decay rate calculation unit uses a sequence of ice crystal edge images captured by a high-speed camera. The image processing algorithm identifies the ice crystal contour and calculates the curvature change per unit time. The rate of curvature change has a quantitative relationship with the surface tension decay. The ice crystal phase transition detector inputs the transmittance abrupt change signal and the surface tension decay rate signal into the decision logic unit. The decision logic unit uses a fuzzy inference mechanism to comprehensively determine the probability of a phase transition occurring. When the probability value reaches a confidence threshold, a phase switching signal is generated. This phase switching signal is transmitted to the temperature control strategy actuator via a fieldbus. The temperature control strategy actuator includes a gradient cooling curve generator and a heating slope constraint. The gradient cooling curve generator retrieves thermodynamic parameters from a preset parameter library based on the material type. These thermodynamic parameters include specific heat capacity, latent heat, and glass transition temperature. The gradient cooling curve generator constructs a temperature-time curve using a piecewise linearization method, with the curve slope set in a gradually changing region near the eutectic point of the material. The heating slope constraint limits the rate of increase in heating power using a limiting algorithm. The limiting value is dynamically set according to the thermal stability classification of the material; thermosensitive materials correspond to a lower upper limit for the heating slope. The temperature control strategy actuator outputs a pulse width modulation signal to drive a solid-state relay. The solid-state relay controls the power supply to the heating plate, achieving precise tracking of the gradient cooling curve. The temperature control strategy actuator has a built-in self-tuning proportional-integral-derivative controller, which adjusts the duty cycle of the pulse width modulation signal in real time according to the temperature deviation.

[0033] During the drying phase, the vacuum regulation strategy is activated, and its pressure threshold sequence is executed by the pressure controller. The pressure threshold sequence adopts a stepped descent pattern, with each pressure step corresponding to a sublimation rate range. The pressure controller adjusts the pumping rate of the vacuum pump via an electro-proportional valve, which receives analog voltage signals and converts them into precise pneumatic pressure control. A leakage compensation mechanism, as an auxiliary module of the vacuum regulation strategy, calculates the compensation amount by monitoring the deviation between the actual pressure and the set pressure. The compensation amount is calculated using an incremental proportional-integral algorithm, and the algorithm output is superimposed on the control signal of the electro-proportional valve. The leakage compensation mechanism includes a dead-zone compensation function to eliminate control errors caused by valve hysteresis. The vacuum regulation strategy and temperature control strategy are coupled; the switching timing of the pressure threshold sequence is synchronized with the inflection point of the gradient cooling curve. This synchronization mechanism is achieved through the time trigger signal of the system master clock.

[0034] The dynamic segmentation model for freeze-drying stages undergoes multiple verifications during stage switching. The phase switching signal generated by the ice crystal phase transition detector needs to pass a consistency check. The consistency check module compares the time difference between the transmittance abrupt change point and the surface tension decay inflection point. If the time difference exceeds the allowable range, a re-detection process is initiated. The re-detection process increases the sampling frequency and extends the observation window. Within the observation window, a weighted average algorithm is used to improve the signal-to-noise ratio. Verified phase switching signals activate event logging, which records a snapshot of the environmental state matrix at the switching critical point and the original sensor data. The dynamic segmentation model for freeze-drying stages maintains a stage state machine, which manages the state transitions between the pre-freezing stage, the primary drying stage, and the secondary drying stage. State transition conditions include time constraints and physical parameter constraints. Time constraints prevent premature stage switching, while physical parameter constraints ensure the phase transition process is fully completed. The transmittance abrupt change detection of the material's ice crystal morphology data involves complex optical compensation. The optical compensation algorithm corrects the optical path deviation caused by the randomness of ice crystal orientation. The transmittance sensor is equipped with multi-angle receiving probes, which collect the intensity of transmitted light from different directions. Transmittance calculation employs a vector synthesis method, which eliminates the influence of anisotropy on the measurement results. The surface tension decay rate calculation incorporates a temperature compensation coefficient, which is dynamically corrected based on material temperature gradient data. The surface tension calculation unit integrates an edge detection optimization algorithm, which adapts to the blurring of contours during ice crystal growth. The ice crystal phase transition detector performs baseline calibration periodically, acquiring reference transmittance and reference surface tension values ​​during the initial pre-freezing stage.

[0035] The gradient cooling curve generation takes into account the influence of material loading, which is obtained through a weighing sensor. The weighing sensor data is used in the curve parameter calculation. The gradient cooling curve generator has safety protection logic that monitors the difference between the material temperature and the chamber temperature; if the difference exceeds a limit, it triggers curve smoothing adjustment. The heating slope constraint is linked to the vacuum regulation strategy; this linkage mechanism temporarily relaxes the slope constraint when the pressure changes rapidly. The pressure threshold sequence is generated based on a sublimation kinetic model, which takes material thickness and porosity parameters as input. The pressure controller has an overshoot suppression function, which reduces the proportional valve opening in advance at pressure switching points. The leakage compensation mechanism's learning function records historical compensation data, which is used to optimize the parameter settings of the compensation algorithm.

[0036] The dynamic partitioning model for freeze-drying stages relies on a high-performance embedded platform equipped with a multi-core processor and a real-time operating system. The ice crystal phase transition detector runs as an independent task on a dedicated kernel, ensuring real-time phase transition identification. The temperature control strategy actuator and vacuum regulation strategy controller each have their own processing cores, exchanging data via shared memory. The system employs a time-triggered architecture, ensuring the accuracy of the control cycle. The dynamic partitioning model includes a watchdog timer to monitor timeouts during stage switching. Model parameters are stored in a ferroelectric memory, ensuring parameter retention even after power failure. The dynamic partitioning model supports a remote configuration interface, allowing operators to adjust the sensitivity parameters for stage switching. The optical path of the ice crystal phase transition detector is kept clean, maintained through a periodic purging mechanism. This mechanism sprays filtered and drying gas to prevent water vapor condensation on the optical window. The image acquisition system of the surface tension calculation unit is equipped with an autofocus mechanism that tracks changes in the distance between the ice crystals and the lens. The gradient cooling curve generator incorporates a curve smoothing filter to eliminate abrupt changes in the temperature setpoint. The pressure threshold sequence switching employs a gradual transition method to prevent material splashing caused by sudden pressure changes. The pressure sensor used in the leakage compensation mechanism is a high-stability model, which reduces compensation errors caused by drift. The dynamic partitioning model throughout the freeze-drying stage undergoes robustness testing under extreme conditions to verify the model's fault tolerance under sensor failure and external interference.

[0037] Example 3: The closed-loop feedback unit initiates a residual moisture monitoring mode during the secondary drying stage. This mode acquires the real-time dielectric constant of the material using a dielectric constant sensor. The sensor employs a parallel-plate capacitor structure, with the parallel-plate electrodes embedded in the material tray and in direct contact with the material. Dielectric constant measurement is based on the capacitance-to-dielectric constant conversion principle. The sensor applies a high-frequency AC electric field and measures the dielectric loss tangent of the material. The real-time dielectric constant data is amplified and filtered by a signal conditioning circuit, and the conditioned signal is sampled and digitized by an analog-to-digital converter. The dielectric constant threshold of the preset morphology template is derived from a material drying endpoint database, which stores the ideal dielectric constant range for different materials. The real-time dielectric constant is compared with the dielectric constant threshold of the preset morphology template using a sliding window mean comparison algorithm. The window size is dynamically adjusted according to the material drying characteristics. When the real-time dielectric constant continuously exceeds the dielectric constant threshold, a drying termination command is generated. This command is transmitted to the vacuum regulation strategy actuator via a digital output interface. Upon receiving the drying termination command, the vacuum regulation strategy actuator triggers an emergency pressure relief protocol, which releases chamber pressure by rapidly opening a pressure relief valve.

[0038] The drying termination command is synchronously sent to the temperature control strategy executor, which forcibly interrupts the subsequent execution of the gradient cooling curve. The interruption point of the gradient cooling curve records the current temperature setpoint and the remaining curve segment length, and the interruption point data is stored in the interruption status register. After the gradient cooling curve is interrupted, the temperature control strategy executor initiates a cooling rate buffering mechanism, which reduces the heating plate power linearly until shutdown. The linear decreasing function is expressed as:

[0039] in: The power of the heating plate at time t is expressed in watts. This represents the initial heating plate power when the gradient cooling curve is interrupted, and the unit is watts. The decreasing slope of the cooling rate buffer mechanism is expressed in watts per second. This indicates the elapsed time from the start of cooling, in seconds. The decreasing slope of the cooling rate buffer mechanism is dynamically adjusted based on the material's thermal conductivity coefficient, which is obtained from the material property table. The adjustment process is constrained by the current pressure value of the vacuum regulation strategy, which is read in real time by a vacuum sensor. The power output of the cooling rate buffer mechanism drives the heating plate through a pulse width modulator, which adjusts the duty cycle according to the power setpoint.

[0040] When the heating plate power drops below a safety threshold, a chamber sealing self-test is triggered. The safety threshold is set as a percentage of the heating plate's maximum power. The chamber sealing self-test is performed by comparing the difference between the historical compensation amount and the theoretical leakage amount of the leakage compensation mechanism. The historical compensation amount records the total compensation amount of the leakage compensation mechanism within the most recent operating cycle. The theoretical leakage amount is calculated based on the ideal gas law, with chamber volume, gas molecular weight, and temperature parameters input during calculation. The difference is calculated using an absolute value comparison method, and the result is compared with the upper tolerance limit. During the chamber sealing self-test, other non-critical operations are suspended to ensure the accuracy of the self-test. The self-test result is recorded in the system log, which includes a timestamp, the difference size, and the upper tolerance limit. An alarm signal is triggered when the chamber sealing self-test detects an anomaly, alerting the operator via an audible and visual alarm. The dielectric constant sensor is calibrated before each freeze-drying cycle using a standard dielectric constant sample. The standard dielectric constant sample is placed in a specific location on the material tray, and the dielectric constant sensor measures the sample reading and compares it to the standard value. The calibration coefficients are obtained through least-squares fitting, and the fitted coefficients are stored in non-volatile memory. A temperature compensation algorithm is used during real-time dielectric constant measurement to correct the dielectric constant reading based on the material temperature. The dielectric constant sensor has a self-diagnostic function that detects electrode short-circuit or open-circuit faults. The dielectric constant threshold of the preset morphology template has an adaptive adjustment mechanism that optimizes the threshold range based on historical drying data.

[0041] The emergency pressure relief protocol incorporates multiple safety measures. The first level is the slow-opening control of the pressure relief valve, with the opening speed increasing exponentially. The second level is pressure release rate monitoring; when the pressure release rate exceeds a set value, flow limiting is triggered. The third level is chamber pressure equalization checking, ensuring uniform material stress during pressure relief. The emergency pressure relief protocol is linked to a mechanical locking device, which automatically activates at the start of pressure relief. The vacuum regulation strategy actuator has pressure relief timeout protection; if pressure relief is not completed within the predetermined time, a backup pressure relief channel is activated. The temperature control strategy actuator's cooling rate buffering mechanism integrates a power soft-drop function, automatically switching to natural cooling mode when power approaches zero. The decreasing slope adjustment of the cooling rate buffering mechanism considers the material's heat capacity characteristics, with heat capacity data participating in the slope calculation. The cooling rate buffering mechanism also has an interruption recovery function, allowing the cooling program to continue after an unexpected power outage. The current pressure value of the vacuum regulation strategy is verified by redundant sensors; consistency checks of the redundant sensor data ensure the reliability of the pressure readings. The surface temperature of the material is monitored in real time during the cooling process, and the surface temperature of the material is measured non-contactly using an infrared thermometer.

[0042] The differential analysis of the cavity sealing self-test process employs a statistical process control (SPC) method, which establishes historical control limits for the differential values. Historical compensation values ​​for the leakage compensation mechanism are stored in a circular buffer, which holds compensation data from the most recent few freeze-drying cycles. Theoretical leakage calculation incorporates real-time temperature correction to ensure the accuracy of the results. The cavity sealing self-test process includes a retry mechanism that automatically repeats the test process if the initial self-test fails. The fault tolerance limit for the self-test process is set based on the cavity design parameters and dynamically adjusted over time. The results of the cavity sealing self-test process are categorized into multiple levels, each corresponding to a different processing strategy. The residual moisture monitoring mode of the closed-loop feedback unit includes data validity verification, checking whether the trend of the dielectric constant reading conforms to the drying process. When the dielectric constant sensor reading is abnormal, it automatically switches to a standby judgment mode, which infers the degree of drying based on the temperature change rate. The generation of the drying termination command requires meeting a continuous exceedance condition, which requires the dielectric constant to exceed a threshold for multiple consecutive sampling cycles. The execution progress of the emergency pressure relief protocol is displayed in real time on the HMI, which shows the current pressure value and remaining pressure relief time. The status information of the temperature control strategy actuator is transmitted to the monitoring system via fieldbus, and the monitoring system records detailed parameters of the cooling process.

[0043] The cooling rate buffering mechanism relies on a precise timer, which uses a temperature-compensated crystal oscillator as the time base. Heating plate power detection is achieved through a Hall current sensor, which measures the actual operating current of the heating element. The pressure sampling rate of the chamber sealing self-check process is higher than in normal operation mode; this high-pressure sampling rate ensures the accuracy of the difference calculation. The historical compensation record of the leakage compensation mechanism includes time stamps, which are used to analyze the trend of compensation changes. The theoretical leakage calculation considers the influence of gas composition changes, which are periodically analyzed using a mass spectrometer. The execution cycle of the entire closed-loop feedback unit is synchronized with the main control cycle of the freeze dryer, which is uniformly scheduled by the system clock. The installation position of the dielectric constant sensor is optimized to ensure that the measurement results represent the overall condition of the material. The dielectric constant threshold of the preset morphology template has a seasonal adjustment factor to compensate for the influence of environmental temperature and humidity changes. The pressure relief valve of the emergency pressure relief protocol uses a fast-opening, slow-closing type, ensuring a smooth and controllable pressure relief process. The cooling rate buffering mechanism of the temperature control strategy actuator has a minimum power protection feature to prevent over-cooling of the heating plate. The difference calculation in the cavity sealing self-test process uses floating-point arithmetic, which improves the calculation accuracy for small differences.

[0044] Example 4: See Figure 3When the cavity sealing self-check process detects a difference exceeding the tolerance limit, a sealing failure alarm is generated. The difference is calculated based on the absolute deviation between the historical compensation amount and the theoretical leakage amount of the leakage compensation mechanism. The tolerance limit is set as a fixed value according to the cavity design specifications and material fatigue characteristics. The sealing failure alarm triggers a stage rollback operation in the freeze-drying stage dynamic partitioning model. The stage rollback operation forcibly modifies the current stage flag of the system to the code of the primary drying stage and clears all temporary parameters of the secondary drying stage. The freeze-drying stage dynamic partitioning model reloads the temperature control strategy and vacuum regulation strategy corresponding to the primary drying stage. The temperature control strategy reads the starting point parameter of the gradient cooling curve from the configuration file, and the vacuum regulation strategy initializes the index pointer of the pressure threshold sequence. During the stage rollback operation, the dielectric constant monitoring function of the closed-loop feedback unit is disabled. The disabling operation is achieved by setting the function prohibition bit. The power supply of the dielectric constant sensor remains normal, but the data is not processed. When the pressure fluctuation range recovers to the theoretical leakage range, the closed-loop feedback unit is re-enabled. The recovery condition is determined by the pressure stability of multiple consecutive sampling cycles. After the stage rollback operation is completed, the effectiveness of the rollback is verified by the real-time transmittance of the material's ice crystal morphology data. The real-time transmittance data is compared with the preset standard transmittance range of the primary drying stage. If the transmittance does not meet the preset standard of the primary drying stage, an emergency drying protocol is activated. The emergency drying protocol operates in a constant temperature and pressure mode, with fixed heating plate temperature and chamber pressure values. The execution duration of the emergency drying protocol is limited by the remaining buffer time of the cooling rate buffer mechanism, which is calculated based on the initial settings of the cooling rate buffer mechanism. After the timeout, the emergency drying protocol is forcibly terminated and a system shutdown command is triggered. The system shutdown command is sent to each execution unit through the main controller.

[0045] The seal failure alarm generation mechanism includes a multi-level verification process. The first level checks the persistence of the difference, requiring the difference to exceed the tolerance limit for a set number of consecutive times. The second level verifies the consistency of sensor readings, eliminating false alarms caused by a single sensor failure. The seal failure alarm record includes a timestamp, the difference magnitude, and the current pressure value, and is stored in non-volatile memory for subsequent analysis. The stage rollback operation requires backing up the current system state, including temperature setpoints, pressure parameters, and control variables. During the rollback, prompts are displayed, informing the operator of system state changes via the human-machine interface. The stage rollback operation of the freeze-drying stage dynamic partitioning model involves resetting control parameters to their initial values ​​for the first drying stage. The gradient cooling curve of the temperature control strategy restarts from the starting point, with the starting temperature fine-tuned based on the material state during rollback. The pressure threshold sequence of the vacuum regulation strategy resets the pointer, and pressure control begins from the first pressure step of the first drying stage. After the stage rollback operation is completed, a system stability check is performed, monitoring the changing trends of key parameters. After the stability check is passed, the automatic control mode is restored, and the automatic control mode continues to execute the standardized process of one drying stage.

[0046] The isothermal and pressure-controlled mode parameters of the emergency drying protocol are preset based on material characteristics. The isothermal value is lower than the glass transition temperature of the material, and the pressure value is higher than the sublimation pressure of ice. Operation of the isothermal and pressure-controlled mode requires disabling the conventional control algorithm, which is replaced by a simplified controller that directly outputs fixed values. The emergency drying protocol has priority management, with priority higher than other background tasks to ensure real-time response. The remaining buffer time calculation of the cooling rate buffer mechanism considers the duration of the emergency drying protocol, and the remaining buffer time is dynamically updated and displayed on the monitoring interface. The timeout termination mechanism uses a hardware watchdog timer, which operates independently of the main controller. The dielectric constant monitoring function of the closed-loop feedback unit continues data acquisition during periods when it is disabled; the acquired data is marked as invalid but does not participate in control decisions. Pressure fluctuation range monitoring uses a sliding window statistical method to calculate the mean and variance of the pressure values. The theoretical leakage calculation incorporates real-time temperature compensation, which corrects the theoretical leakage value according to gas laws. The conditions for re-enabling the closed-loop feedback unit include pressure stability and temperature stability; these dual stability requirements avoid frequent switching. Re-enabling the operation requires administrator confirmation, which can be achieved through password verification or a physical button.

[0047] The standard value for verifying the effectiveness of the transmittance rollback is derived from a historical database, which stores the typical transmittance range of similar materials during a single drying stage. The transmittance sensor undergoes periodic optical calibration, using a standard transmittance plate to adjust the sensor gain. The emergency drying protocol's constant temperature and pressure values ​​have safety boundaries to prevent temperature or pressure from exceeding the material's tolerance range. The remaining buffer time for the cooling rate buffer mechanism is displayed on the main interface, which provides manual adjustment of the buffer time. System shutdown commands require recording the shutdown reason, which is categorized and stored for easy fault analysis. The rollback depth for stage rollback operations is configurable, specifying the specific sub-stage to which the rollback will occur. The temperature control strategy's fine-tuning algorithm adjusts based on the degree of dryness before rollback, estimated using the completed drying time. The vacuum regulation strategy's pointer reset considers the remaining moisture content of the material, inferred through indirect measurement. The emergency drying protocol's timeout period can be set in stages, corresponding to different material safety levels. The execution of system shutdown commands includes an ordered shutdown sequence, ensuring the equipment's safe operating status. The fault tolerance limit for seal failure alarms is dynamically adjusted based on the wear model of the cavity seals, according to the equipment's usage time. The completion signal of the stage rollback operation triggers event logging, which includes parameter comparisons before and after the rollback. The execution progress of the emergency drying protocol is displayed in real-time as a percentage progress bar. The remaining buffer time alarm for the cooling rate buffer mechanism has an early warning mechanism to notify the operator in advance. The generation of system shutdown commands requires multiple conditions to be met, preventing accidental shutdowns. Referring to Table 1, the following reference data is used for calculating the difference in the cavity sealing self-test process.

[0048] Table 1: Comparison Table of Sealing Performance Self-Inspection Parameters

[0049] The theoretical leakage rate is calculated based on standard temperature and the condition of brand-new seals. The tolerance limit percentage represents the maximum percentage allowed to exceed the theoretical value. The detection cycle is determined based on the cavity volume, with larger cavities using longer detection cycles. The difference calculation compares the actual leakage rate with the theoretical value; the actual leakage rate is extrapolated from the pressure change rate. The triggering of seal failure alarms also considers leakage trends, which are determined by the difference changes over multiple consecutive cycles. The success rate of stage rollback operations is recorded in a statistical report used for system maintenance decisions. Transmittance verification uses a multi-point measurement and averaging method to avoid localized phenomena affecting overall judgment. Activating the emergency drying protocol requires recording the current state of the material, including temperature and pressure readings. The remaining buffer time protection of the cooling rate buffer mechanism has a minimum guarantee time to ensure basic safe cooling. The execution sequence of system shutdown commands has been optimized to avoid pressure shocks. The entire stage rollback and emergency handling process is fully automated, minimizing the need for human intervention.

[0050] See Figure 4 This graph is a core data visualization result of the self-inspection of cavity sealing in the automated control system of a constant-temperature hanging-bottle freeze dryer. The graph uses cavity volume as the horizontal axis and leakage rate as the vertical axis, presenting two key indicators: theoretical leakage rate and the upper tolerance limit. The technical significance of this graph lies in providing a quantitative basis for the cavity sealing self-inspection process: when the actual leakage rate exceeds the corresponding cavity's upper tolerance limit, the system will trigger a sealing failure alarm, thereby initiating emergency mechanisms such as stage rollback. Simultaneously, it provides an intuitive reference for equipment selection and maintenance. Engineers can anticipate the difficulty of leakage control based on the cavity volume, rationally set monitoring parameters, ensure the stability of the vacuum environment during freeze-drying, and ultimately ensure the retention of active ingredients and the extension of shelf life during the freeze-drying stage.

[0051] Example 5: After the system shutdown command is activated, the mechanical locking device of the bottle rack is started. The mechanical locking device adopts an electromagnetically driven pin mechanism, and the pin is precisely embedded in the positioning hole of the material bottle tray to achieve mechanical interlocking. Taking the freeze-drying process of 500 ml vials as an example, the bottle rack is configured with sixteen bottle positions, and a miniature gravity sensor is installed below each bottle position. After the electromagnetic coil of the mechanical locking device is energized, it generates a strong magnetic field, which drives the stainless steel pin to extend within 200 milliseconds. The end of the pin adopts a conical design to ensure self-alignment with the positioning hole. The locking status of the mechanical locking device is verified by a gravity sensor array. The gravity sensor adopts the strain gauge principle to measure the real-time weight distribution of each bottle position. An abnormal weight distribution of the material bottle is defined as the weight difference between adjacent bottle positions exceeding a set threshold, which is calculated based on the nominal weight of the bottled material. When an abnormal weight distribution of the material bottle is detected, the pressure relief operation of the vacuum adjustment strategy is delayed. The delay time can be configured within the range of 5-30 seconds.

[0052] During delayed depressurization, the temperature control strategy actuator maintains a minimum heating power, which is set according to the material's thermal stability classification. For thermosensitive biological agents, the minimum heating power is maintained at 5% of the rated power, while for chemical agents, it can be maintained at 10% of the rated power. The duration of maintaining the minimum heating power is proportional to the duration of the gravity sensor's abnormal reading, with the proportionality coefficient dynamically configured according to the material's thermosensitivity level. The proportionality coefficient is 0.8 for thermosensitive class A materials and 1.2 for thermosensitive class B materials. When the gravity sensor returns to normal readings, the mechanical locking device is released first, using a slow-release mode where the electromagnetic coil current linearly decreases to zero within 3 seconds. After the mechanical locking device is completely released, the final depressurization of the vacuum regulation strategy is executed. The final depressurization rate is controlled by a proportional valve, and the depressurization curve uses an S-shaped acceleration / deceleration algorithm. After final depressurization, a freeze-drying process end signal is generated and transmitted to the central monitoring system via fieldbus. The data acquisition channels of the multi-source data fusion engine are sequentially shut down, following sensor type priority. The temperature sensor channel stops sampling first, followed by the vacuum sensor channel, and the material ice crystal morphology monitoring channel stops last. Before each channel closes, a data buffer refresh operation is performed to ensure the complete storage of the last batch of data.

[0053] The mechanical locking device incorporates multiple safety designs. The electromagnetic latch is equipped with a mechanical interlock switch that detects the latch's actual position. A buffer damper is installed at the end of the latch's movement trajectory, using a hydraulic shock absorber to reduce impact noise. Data acquisition from the gravity sensor array employs synchronous sampling technology to eliminate time differences in data from different bottle positions. Each gravity sensor is equipped with a temperature compensation circuit to correct zero-point drift caused by ambient temperature changes. The weight distribution anomaly detection algorithm uses a sliding window mean comparison, with the window size adaptively adjusted based on the freeze dryer's vibration characteristics. The timing accuracy for anomaly duration reaches millisecond levels, using a temperature-controlled crystal oscillator as the time base. During delayed depressurization, the vacuum system maintains a minimum pumping rate, achieved by controlling the vacuum pump speed via a frequency converter. The minimum heating power output of the temperature control strategy actuator uses pulse width modulation (PWM), with a PWM frequency set to 20kHz to avoid audible noise. The proportional coefficient configuration interface includes access control, distinguishing between operator and engineer levels. The thermal sensitivity database is stored in encrypted memory, which uses the AES algorithm to protect critical parameters. The criteria for judging the recovery of gravity sensor data to normal include stability requirements, which stipulate that the data in multiple consecutive sampling periods are within the normal range.

[0054] The release process of the mechanical locking device involves multiple verification stages. The first stage verifies the demagnetizing current curve of the electromagnetic coil, which should conform to a preset attenuation model. The second stage verifies the microswitch signal at the retracted pin position, which must remain stable for at least 500 milliseconds. The third stage verifies the consistency of gravity sensor readings, ensuring that the weight data of all bottle positions returns to the initial distribution pattern. The depressurization rate curve for the final depressurization operation is optimized, taking into account the chamber volume and the pressure difference of the material bottles. Pressure monitoring points are distributed at different locations within the chamber during the depressurization process, and the data from these points is used to verify pressure uniformity. The generation of the freeze-drying process end signal requires a complete condition chain, including confirmation of the mechanical locking status, normal gravity sensor data, and a final depressurization completion marker. The channel shutdown procedure of the multi-source data fusion engine has timeout protection to prevent system hang due to sensor failure. The data buffer employs a dual-backup storage strategy to ensure data integrity during unexpected power outages. After receiving the signal that the freeze-drying process has ended, the central monitoring system automatically generates a freeze-drying batch report, which includes key process parameter curves and event logs.

[0055] The calibration cycle of the gravity sensor array is synchronized with the production batch; automatic calibration of the gravity sensors is performed before the start of each new batch. During the automatic calibration process, standard weights are loaded onto each bottle position, with the weight of the standard weights matching the nominal weight of the material bottle. The calibration data is fitted as a quadratic polynomial curve, and the polynomial coefficients are stored in the calibration memory of each sensor. During the delayed pressure relief period, the system status indicator light flashes yellow, and the brightness of the indicator light automatically adjusts according to the ambient light level. The minimum power maintenance mode of the temperature control strategy actuator has anti-condensation protection, which is achieved by controlling the temperature of the cavity wall. The electromagnetic coil winding of the mechanical locking device uses high-temperature resistant enameled wire with an insulation class of H (180℃). The surface of the pin material is hardened to improve the pin's wear resistance and corrosion resistance. The gravity sensor anomaly detection algorithm has a learning function that records the typical weight distribution patterns of different materials. A safety factor is incorporated into the calculation of the delayed pressure relief time, and this safety factor is dynamically adjusted based on the historical anomaly handling success rate. The thermal sensitivity classification of the proportional coefficient is based on pharmacopoeia standards, which clearly define the thermal stability requirements for various materials. The pressure release path for the final depressurization operation is optimized using computational fluid dynamics (CFD), ensuring smooth, vortex-free airflow. The pressure relief valve's sealing surface employs a hard metal seal structure, guaranteeing zero leakage upon final closure. The channel closure command from the multi-source data fusion engine is transmitted via optocoupler isolation, enhancing anti-interference capabilities. The freeze-drying process completion signal triggers an automatic printing function, automatically printing out batch reports and quality certificates. After entering standby mode, the system executes a comprehensive self-test program covering all critical actuators and sensors.

[0056] The moving parts of the mechanical locking device are regularly lubricated using high-vacuum compatible grease. The gravity sensor's signal cable uses shielded twisted-pair cabling to reduce electromagnetic interference. Pressure monitoring data during delayed depressurization is recorded via high-frequency sampling, which is used for subsequent process analysis. The power output accuracy of the temperature control strategy actuator reaches 0.1%, guaranteed by closed-loop current feedback. The dynamic configuration interface for the proportional coefficient provides a verification function to prevent parameter settings from exceeding safe ranges. The logic for determining when the gravity sensor returns to normal readings includes de-jitter processing to avoid misjudgments caused by momentary interference. System state switching after the mechanical locking device is released requires operator confirmation via a touchscreen dialog box. The pressure change rate during the final depressurization process is displayed in real time to help determine if the depressurization process is normal. After the multi-source data fusion engine's data acquisition channel is closed, a data integrity report is generated, listing the checksums of all stored data. After the freeze-drying process ends, a chamber cleaning procedure is automatically initiated to prepare for the next batch, including vacuum cleaning and high-temperature sterilization.

[0057] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0058] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An automated control method for a constant temperature bottle-hanging freeze dryer, characterized by, The method comprises the following steps: Collecting multi-dimensional parameter data of the internal environment of the freeze dryer, including temperature gradient data, vacuum fluctuation data, and material ice crystal morphology data; Real-time synchronous calibration of multi-dimensional parameter data through a multi-source data fusion engine to generate a calibrated environment state matrix; Based on the calibrated environment state matrix, a freeze-drying stage dynamic division model is constructed, which is used to identify the switching critical point of the pre-freezing stage, the primary drying stage, and the secondary drying stage; According to the output of the freeze-drying stage dynamic division model, the corresponding temperature control strategy and vacuum regulation strategy are activated, the temperature control strategy includes gradient cooling curve and temperature rising rate constraint, and the vacuum regulation strategy includes pressure threshold sequence and leakage compensation mechanism; Through a closed-loop feedback unit, the deviation of the material ice crystal morphology data from the preset morphology template is monitored in real time, and the execution parameters of the temperature control strategy and the vacuum regulation strategy are dynamically corrected.

2. The method of claim 1, wherein When the multi-source data fusion engine performs data synchronous calibration, a space-time alignment algorithm is used to eliminate the time delay difference between the temperature gradient data and the vacuum fluctuation data, and the periodic characteristics of the material ice crystal morphology data are extracted through frequency domain decomposition; If the correlation between the temperature gradient data and the vacuum fluctuation data is lower than the dynamic threshold, a redundant data elimination mechanism is triggered to retain the effective data segment of the high-frequency sampling points; The effective data segment is input into the environment state matrix generator to output a three-dimensional environment state matrix with a time stamp.

3. The method of automated control of a constant temperature bottle-hung freeze dryer according to claim 2, characterized in that The freeze-drying stage dynamic division model identifies the switching critical point of the pre-freezing stage to the primary drying stage through an ice crystal phase change detector, which judges based on the abrupt change of the light transmittance of the material ice crystal morphology data and the surface tension decay rate; When the ice crystal phase change detector outputs a phase switching signal, the temperature control strategy of the primary drying stage is activated, and the gradient cooling curve of the temperature control strategy is adaptively matched with the maximum allowed temperature rising slope according to the material type; During the continuous period of the primary drying stage, the cavity pressure is gradually reduced through the pressure threshold sequence of the vacuum regulation strategy, and the vacuum fluctuation caused by the sublimation of the material is dynamically compensated according to the leakage compensation mechanism.

4. The method of claim 3, wherein The closed-loop feedback unit starts a residual moisture monitoring mode in the secondary drying stage, obtains the real-time dielectric constant of the material through a dielectric constant sensor, and compares it with the dielectric constant threshold of the preset morphology template; If the real-time dielectric constant continuously exceeds the dielectric constant threshold, a drying termination instruction is generated, and an emergency pressure relief protocol of the vacuum regulation strategy is triggered; The drying termination instruction is sent to the temperature control strategy executor synchronously to forcibly interrupt the subsequent execution of the gradient cooling curve.

5. The method of automated control of a constant temperature bottle-hung freeze dryer according to claim 4, characterized in that After the gradient cooling curve is interrupted, the temperature control strategy executor starts a cooling rate buffer mechanism to linearly decrease the power of the heating plate until shutdown; The decreasing slope of the cooling rate buffer mechanism is dynamically adjusted according to the thermal conductivity of the material, and the adjustment process is constrained by the current pressure value of the vacuum regulation strategy; When the power of the heating plate decreases below the safety threshold, a cavity sealing self-checking process is triggered, which is completed by comparing the difference between the historical compensation amount of the leakage compensation mechanism and the theoretical leakage amount.

6. The method of automated control of a constant temperature bottle-hung freeze dryer according to claim 5, characterized in that If the difference exceeds the fault tolerance upper limit, a sealing failure alarm is generated and the abnormal pressure fluctuation range is recorded; The sealing failure alarm triggers a stage rollback operation of the dynamic stage division model, forcibly resets the current stage to a primary drying stage, and reloads the corresponding temperature control strategy and vacuum regulation strategy; During the stage rollback operation, the dielectric constant monitoring function of the closed-loop feedback unit is disabled until the pressure fluctuation range returns to the theoretical leakage range.

7. The method of automated control of a constant temperature bottle-hung freeze dryer according to claim 6, characterized in that After the stage rollback operation is completed, the effectiveness of the rollback is verified by real-time light transmittance of the material ice crystal morphology data; If the light transmittance does not meet the preset standard of the primary drying stage, an emergency drying protocol is started, which runs in a constant temperature and constant pressure mode until the light transmittance meets the standard; The execution time of the emergency drying protocol is limited by the remaining buffer time of the cooling rate buffer mechanism, and after the timeout, it is forcibly terminated and triggers a system shutdown instruction.

8. The method of automated control of a constant temperature bottle-hung freeze dryer according to claim 7, characterized in that After the system shutdown instruction is activated, the mechanical locking device of the bottle hanging bracket is started to prevent the material bottle from being displaced due to pressure surge; The locking state of the mechanical locking device is verified by the gravity sensor, and if the material bottle weight distribution is abnormal, the pressure relief operation of the vacuum regulation strategy is delayed; During the delay pressure relief, the temperature control strategy executor maintains the minimum heating power to avoid sudden temperature rise of the material.

9. The method of automated control of a constant temperature bottle-hung freeze dryer according to claim 8, characterized in that The maintenance time of the minimum heating power is proportional to the abnormal duration of the gravity sensor, and the proportional coefficient is dynamically configured according to the heat sensitivity level of the material; When the gravity sensor returns to normal reading, the mechanical locking device is preferentially released, and the final pressure relief of the vacuum regulation strategy is executed; After the final pressure relief is completed, a freeze-drying process end signal is generated, and the data acquisition channel of the multi-source data fusion engine is closed.

10. An automated control system for a constant temperature bottle-hanging freeze dryer, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, The processor, when executing the computer program, realizes the steps of the automatic control method of the constant temperature hanging bottle type freeze dryer in any one of claims 1 to 9.

Images