Microcrystal freezing resonance chamber

Abstract

The invention relates to the technical field of fresh food freezing, and particularly discloses a microcrystal freezing resonance chamber which is used for solving the problems that a large seafood thick block passes through a critical zone at 0-5 DEG C slowly, nucleation is sparse, thick ice crystals are caused, the temperature is uneven, dropping loss and cracking are caused, a glaze layer is lost, a parameter window is narrow, energy consumption is high, sanitary interlocking is difficult to control, and online monitoring is insufficient. Comprising a pre-cooling unit, a jet flow humidity control unit, a micro snow nucleation unit, an ultrasonic unit, an electric field unit, a magnetic control unit, a pile conveying unit, an online detection unit, a prediction control unit, a cleaning linkage unit and a stable cooling unit. Through cooperation of an electric field, a magnetic field and ultrasound and phase coupling with CO2 micro snow nucleation, under BP-ANN predictive control closed-loop parameter adjustment, a critical zone is quickly and uniformly crossed, the multipoint nucleation density of ice crystals is improved, drop loss and dry loss of products are reduced, recrystallization and cracking are inhibited, color and texture are stabilized, energy consumption is reduced, and the quality consistency of frozen finished products is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fresh frozen technology, and particularly relates to a microcrystal freezing resonance chamber. BACKGROUND

[0002] Large seafood such as tuna, salmon, prawns and crabs are thick and unevenly distributed with fascia fat. The traditional air cooling, plate freezing or ammonia / fluorine direct expansion in the 0 to-5℃ critical zone is slow, which leads to sparse primary nucleation, prolonged phase transition platform, water migration and freezing concentration, and generates coarse ice crystals to pierce muscle fibers, and then causes thawing juice loss, texture softening and dark color; at the same time, the temperature is discrete due to the difference in heat conduction between the center and the surface, and the recrystallization and surface dryness occur at the same time, which increases energy consumption and quality fluctuation; in order to improve the nucleation density and the crossing speed, although deep cooling high convection, plate freezing pressure connection, microwave / radio frequency, pulsed electric field, static magnetic field and liquid CO2 have been tried, due to the thickness and the heterogeneity of fat / connective tissue / peel and the gradient of heat conduction, dielectric, acoustic attenuation, the problems such as shallow microwave / radio frequency penetration and uncontrollable hot spots, low ultrasonic coupling efficiency, and CO2 micro snow frost heating resistance often occur; parameter out-of-bounds can easily cause muscle fiber / connective tissue softening, and appear erosion, strip and oil loose, and further cause thawing drip loss, flavor loss and color darkening, and the display window is narrow; and the center-surface temperature difference and the stacking shielding lead to uneven cooling, high salt and high protein biological membrane, and the hygiene dead angle of spray and air duct, CO2, noise and electromagnetic exposure are limited and the maintenance difficulty is increased; microcrystal freezing advocates nucleation impact after precooling, promotes multi-point nucleation, speeds up the crossing of the critical zone, and inhibits recrystallization and dryness by deep cooling and stable structure, but it still needs to coordinate the convection and humidity, micro snow particle size and flux, low intensity ultrasonic coupling, magnetic field arrangement and heat dissipation, linear speed stacking matching, NIR (near infrared spectrum) evaluation and MPC (model predictive linkage) linkage, and CIP (cleaning in place / online cleaning) ventilation interlocking, and lacks adaptive automatic control mechanism, and the critical zone crossing time target and temperature uniformity limit value are not determined according to the thickness and fat rate of seafood products. SUMMARY

[0003] The technical problem to be solved by the present application is to provide a microcrystal freezing resonance chamber, which is coupled with CO2 micro snow nucleation phase by electric field, magnetic field and ultrasonic coordination, and under the closed loop parameter adjustment of BP-ANN prediction control, quickly and uniformly crosses the critical zone, improves the multi-point nucleation density, reduces the product drip loss and dryness, inhibits recrystallization and cracking, stabilizes the color and texture, reduces the energy consumption, and improves the consistency of frozen product quality.

[0004] To achieve the above-mentioned purpose, the present application provides the following technical solutions: A kind of microcrystalline frozen resonance chamber, including precooling unit, jet wet control unit, micro snow nucleation unit, ultrasonic unit, electric field unit, magnetic control unit, transport stack unit, online detection unit, predictive control unit, cleaning interlock unit and stable cooling unit;Transport stack unit is placed in the center area of the two parallel electrode plates in sample chamber, electric field unit makes the upper high-voltage electrode plate, the lower ground plate and applies 0.5~3.0kV·cm -1 Electrostatic field;Magnetic control unit includes a pair of Helmholtz coil pairs located on the side wall of sample chamber, forms 0.5~2.0mT uniform static magnetic field in central area;The face of ultrasonic unit's vibrator is pasted to the side wall of sample chamber and the bottom sound coupling window is towards tray, and is excited with 20%~40% duty cycle pulse to nucleate with electromagnetic field cooperatively;Precooling unit reduces surface layer to-0.5~ -2.5℃ under the constraint of partition temperature and humidity, and triggers micro snow nucleation unit to spray D50 50~150 μm CO2 micro snow in 30~120 s with rewarming sharp; -1 And 8~15 m·s -1 Fast cross 0~-5 ℃ critical region with 80%~95% or 70%~85% relative humidity strategy;Online detection unit continuously collects fiber temperature, infrared thermal image, wind speed, humidity, carbon dioxide concentration, acoustic cavitation and near infrared spectrum;Predictive control unit predicts and controls rolling optimization linear speed, air volume, humidity, carbon dioxide flux, ultrasonic duty cycle, electric field strength and magnetic field strength based on BP-ANN model, so that temperature uniformity≤2.0℃ and drop loss and recrystallization are inhibited;Cleaning interlock unit implements shutdown and in-situ cleaning for CO2, noise and electromagnetic out-of-limit;Stable cooling unit stabilizes and cools the microcrystalline structure with low temperature and short stay and micro disturbance, and inhibits recrystallization.

[0005] As a further scheme of the application, the precooling unit includes a low-temperature evaporator, a fan, a guide air duct, a rectifier louver, an infrared thermal imager, and a fiber multi-point temperature probe. The variable frequency convection fan sends air to the low-temperature evaporator through the guide air duct. The low-temperature evaporator is connected with the rectifier louver in sequence at the air outlet side and forms a parallel air belt above the transport stack unit. The infrared thermal imager views the product surface, and the fiber multi-point temperature probe is arranged near the surface and the core of the product. The low-temperature evaporator is adjusted by the fan at a frequency of 8~12 m·s -1 The temperature of the product carried by the transport stack unit is gradually reduced and uniformly distributed by the partitioned convection and the relative humidity of 80%~95% or 70%~85%. The surface temperature is reduced to-0.5~ -2.5℃ under the condition of temperature uniformity≤2.0℃. The monitoring array composed of the infrared thermal imager and the fiber multi-point temperature probe calculates the temperature change rate of the 3~5 s sliding window. When ≥60% of the measuring points change from negative to positive and ≥0.05℃·s -1 And the rewarming sharp lasts for ≥5 s, the trigger instruction is output to the micro snow nucleation unit.

[0006] As a further scheme of the present application, in the precooling unit, the relative humidity of the precooling unit is adaptively selected by the prediction control unit according to the dew point margin and quality risk: when the dew point margin calculated by the infrared thermal imager and the temperature and humidity sensor, the dew point margin being equal to the surface temperature minus the dew point temperature, the dew point margin ≥ 3-5 ℃, and the drop loss risk predicted by the online detection unit > 3% or the glaze layer needs to be protected, 80%-95% RH is maintained; when the dew point margin ≤ 2 ℃ or the temperature uniformity > 2.0 ℃ or the frost trend appears, the frost trend criterion is that the relative humidity rising rate ≥ 1%·10 s -1 , and the infrared cold spot area ratio ≥ 10%, switch to 70%-85% RH; and set ≥ 60 s hysteresis, and after the dew point margin restores ≥ 3 ℃ and the frost criterion is removed, switch back to 80%-95% RH.

[0007] As a further scheme of the present application, the micro-snow nucleation unit includes and is connected in sequence a liquid stabilizing tank, a vacuum heat insulation liquid supply pipeline, a main electromagnetic valve, a partitioned cut-off valve, a Coriolis mass flowmeter, an atomization cavity, a back pressure regulating valve and a nozzle, and a laser particle size online monitor is arranged to measure D50 of the spray cone; after the prediction control unit is triggered, the main electromagnetic valve is opened for 30-120 s according to the ignition-platform-fall three-section flux curve, the Coriolis mass flowmeter closes loop to stabilize the mass flow of the single nozzle at 5-20 g·s -1 , the laser particle size online monitor outputs D50 in real time and feeds back to the prediction control unit to control D50 at 50-150 μm through the linkage of the back pressure regulating valve, the partitioned cut-off valve is out of phase and is cut according to the stacking shielding condition to evenly distribute the coverage, and is cooperated with 20%-40% duty pulse of the ultrasonic unit, 0.5-3.0 kV·cm -1 of the electric field unit and 0.5-2.0 mT of the magnetic control unit to induce multi-point nucleation; when CO2 ≥ 5000 ppm lasts for ≥ 5 s or the particle size deviation > |10%| lasts for ≥ 2 s, the flux is automatically reduced to 50% of the initial or the spraying is terminated and is reset according to the hysteresis logic.

[0008] As a further scheme of the present application, the ultrasonic unit, the electric field unit and the magnetic control unit are phase coupled controlled by a unified time base issued by the prediction control unit: within the spraying window of the micro-snow nucleation unit, 0.5-3.0 kV·cm -1 of the electric field is applied between two parallel electrode plates in the sample chamber for orientation pretreatment for 0.3-1.0 s, then the ultrasonic unit generates 20-40 kHz pulse train with 20%-40% duty, the electric field maintains 60%-80% time overlap, the magnetic control unit generates 0.5-2.0 mT uniform static magnetic field outside the sample chamber side wall through a pair of Helmholtz coils and changes at ≤ 0.2 mT·5 s -1 , and the magnetic field direction is orthogonal to the ultrasonic propagation direction; The acoustic cavitation intensity and temperature uniformity indices of the online detection unit, along with the infrared thermal gradient and near-infrared dripping prediction, form a feedback loop and are adjusted according to the following coupling law in a closed loop: when the acoustic cavitation intensity exceeds the set value and lasts for ≥2s, the ultrasonic duty cycle is reduced by 5%–10%, and the electric field strength is increased by 0.2–0.5 kV·cm. -1 When the temperature uniformity is >2.0℃ and lasts for ≥5s, the magnetic field strength is increased by 0.2–0.5mT and the ultrasonic frequency is decreased by 2–5kHz; when the predicted droplet loss is >3%, the electric field preprocessing window is expanded to 0.6–1.2s and the ultrasonic duty cycle is reduced to 20%–30%; the variation of each parameter is limited to an electric field ≤0.5kV·cm every 5s. -1 The system features a magnetic field ≤ 0.2 mT, an ultrasonic duty cycle ≤ 10%, and automatically pauses ultrasound when the CO2 concentration is ≥ 5000 ppm and lasts for ≥ 5 seconds, while reducing the electric field to 0.5 kV·cm. -1 .

[0009] As a further embodiment of the present invention, the jet flow humidity control unit includes a second fan, an air supply box, a guide duct, a second rectifier louver, a zoned air supply outlet, an air volume regulating valve, a zoned return air damper, a condenser dehumidification coil, a hot gas reheat valve, a bypass ventilation damper, an ultrapure water mist generator, a dew point mirror sensor, a temperature and humidity sensor, a wind speed sensor, an infrared thermal imager observation window, a feed air curtain, and an air knife. The second fan is connected to the guide duct via the air supply box, and the guide duct is equipped with a second rectifier louver and a zoned air supply outlet at its front end. Each zone's air supply outlet has a pre-installed airflow regulating valve, and the return air side is equipped with a zoned return air damper. The condenser dehumidification coil and the bypass ventilation damper are connected in parallel to the air supply path. The hot gas reheat valve and the ultrapure water micro-mist generator are connected in series after the condenser dehumidification coil. The dew point mirror sensor, temperature and humidity sensor, and wind speed sensor are respectively installed at the preset positions of the supply and return air. The feed air curtain and the air knife are located at the air supply inlet to form an isolation air curtain. Under the control of the predictive control unit, the fan frequency is 10-60Hz and the outlet air velocity is 8-15m·s. -1 Based on the dew point margin and CO2 injection events, the condenser dehumidification coil, hot gas reheat valve, bypass ventilation damper and ultrapure water micro-mist device are coordinated to form a relative humidity range of 80% to 95% or 70% to 85%, and the temperature uniformity is maintained at ≤2.0℃ by cooperating with the air volume regulating valve and the zoned return air damper.

[0010] As a further aspect of the present invention, when the online detection unit determines that the CO2 concentration reaches or exceeds 5000 ppm and lasts for at least 5 seconds, or the acoustic cavitation intensity exceeds the limit, or the output of the electric field unit and the magnetic control unit exceeds the limit, the cleaning interlock unit, within 0.2 seconds, sequentially issues the cleaning interlock by: instructing the micro-snow nucleation unit to close the main solenoid valve and the partition shut-off valve, and to place the back pressure regulating valve in the pressure relief position, and to use the air knife of the jet wet control unit to dry blow the nozzle for 3-10 seconds to remove residual micro-snow and frost; instructing the jet wet control unit to raise the fan speed to no less than 50 Hz and fully open the feed air curtain door and the partition return air door to establish a fresh air channel, and at the same time start the ultrapure water micro-mist generator to flush the atomization chamber with a micro-atomization of 1.0-1.5 liters per minute per nozzle for 2-4 minutes. The outlet area and the inner wall of the sample chamber are then conditioned by a condensation dehumidification coil in conjunction with a bypass ventilation door to keep the supply air dew point below 5°C for 5-10 minutes to condense and collect the dissolved and entrained contaminants. Subsequently, the hot air reheat valve is opened to raise the supply air above the dew point and maintain it for 10-15 minutes to complete the drying. During this process, an infrared thermal imager, a fiber optic multi-point temperature probe, a temperature and humidity sensor, and a dew point mirror sensor are used together as release criteria. After the CO2 concentration is not higher than 1000ppm, the relative humidity is not higher than 85%, and the equipment temperature is not higher than 30°C and stabilizes for 60 seconds, the cleaning interlock unit is released, the jet humidity control unit returns to the normal humidity range, and the stacking unit moves the in-place product out of the cleaning area at the guaranteed baseline speed. The entire process timestamp and threshold status are then transmitted back to the predictive control unit for archiving.

[0011] As a further aspect of this invention, the stabilizing cooling unit employs a three-stage temperature gradient plus perturbation stabilizing cooling method: the first stage is a residence at -38℃ for 4–8 minutes, the second stage is a residence at -40℃ for 6–12 minutes, and the third stage is a residence at -42℃ for 3–6 minutes; throughout the process, temperature perturbation amplitude is superimposed at 0.3–0.6℃, oscillation period at 120–240s, and airflow micro-pulse amplitude at 10%–15%, single pulse duration at 10–20s, interval at 2–4 minutes, with relative humidity at 65%–80%; the stabilizing cooling unit calculates a recrystallization risk index based on the near-infrared spectrum of the online detection unit and the surface and core temperatures, temperature uniformity, and frost thickness measured by the infrared thermal imager, and the recrystallization risk index is equal to The unfrozen water ratio is multiplied by the relative temperature margin, the temperature fluctuation intensity, and the proportion of time the sample spends in the active region of -15 to -5℃. The relative temperature margin is the proportion of the current temperature above the glass transition temperature to the difference between the melting point and the glass transition temperature. If the temperature is not above the glass transition temperature, the relative temperature margin is zero. The temperature fluctuation intensity is the product of the temperature fluctuation amplitude and the zero-crossing frequency within the near time window. When the recrystallization risk index is >1.0 and lasts for ≥5s, the stabilization cooling unit automatically extends the second stage by 2-4 minutes and increases the air volume by 10%, while reducing the temperature perturbation amplitude by 0.1℃. Stabilization cooling is released and the sample is switched to the out-of-pack mode when the recrystallization risk index is <0.8 and the temperature difference between the core and the near surface is ≤0.8℃.

[0012] As a further aspect of this invention, before the stabilization cooling unit begins, the online detection unit generates a batch profile vector of large seafood products. This batch profile vector includes thickness, fat percentage, glaze ratio, near-infrared feature values, conveyor linear velocity, and jet parameters. The prediction and control unit searches for similar batch templates in the historical parameter database with a nearest neighbor count of 5 and outputs the temperature level setting, residence time, temperature perturbation amplitude, and airflow pulse amplitude as the initial strategy. During execution, exploratory fine-tuning is performed within the safety boundary with a probability of 5%, with the temperature level adjustment not exceeding ±0.5℃ and the single-segment residence time slightly adjusted. Adjustments should not exceed ±10%, airflow pulse amplitude should not exceed ±5%, and relative humidity should not exceed ±3%, with a single adjustment interval of not less than 60 seconds. The objective function is weighted by dripping prediction, temperature uniformity, and unit energy consumption for real-time evaluation. If the objective function decreases, the current parameters are fixed and written into the historical parameter library to increase the weight. If it does not decrease, it is rolled back to the previous parameter set. Throughout the process, the CO2 concentration is constrained to be no higher than 5000ppm and the temperature uniformity to be no higher than 2.0℃. When the core and near-surface temperature difference is no higher than 0.8℃ and the dripping prediction is no higher than 3%, the stable cooling is considered complete.

[0013] As a further aspect of this invention, in the prediction and control unit, the BP-ANN model employs process quantity branching and near-infrared branching for input encoding. Each of the process quantity branching and near-infrared branching undergoes two fully connected layers for extraction. After branch fusion, a temporal modeling layer is introduced to capture the dynamics of the critical region. A multi-task output head is used at the top, simultaneously regressing droplet loss, hardness, recrystallization risk index, and the target time of the critical region. A physical constraint gating layer is connected between the fusion and output layers to perform key monotonic relationships and boundary projections. A physical constraint loss is added to the total loss, resulting in the following total loss function: ; In the formula: For the total loss function, For the number of tasks, , For task indexing, For the first The loss weight of each task, For the first The target variable for each task For the first Mean squared error of each task objective variable The weight of the physical constraint penalty term. As a physical consistency penalty item, for violations , , and The degree of monotonicity is measured, and the positive deviations of each constraint are aggregated using a squared penalty. , , , These are the model's prediction functions for dripping loss, hardness, recrystallization risk index, and critical zone target time, respectively. , , These are relative humidity, dew point margin, and temperature uniformity index, respectively. Dew point margin is equal to the measured surface temperature minus the dew point temperature, and temperature uniformity index is equal to the standard deviation of the temperature at each measuring point. The bypass branch outputs heteroscedasticity and supports Monte Carlo estimation confidence intervals; a missing measurement mask and time decay interpolation layer are added to the input side; a sensitivity linearization layer is connected in parallel at the output to generate gradients for linear velocity, air volume, relative humidity, CO2 flux, ultrasonic duty cycle, and magnetic field strength for predictive control; at the same time, a drift correction layer and a soft limiting layer are arranged, the former performing statistical alignment across batches and product types, and the latter mapping the suggested settings to the device safety domain.

[0014] The technical advantages of this invention's microcrystalline cryogenic resonance chamber are as follows: Through a pre-cooling unit with zoned convection and dew point adaptive humidity control, the surface of large seafood products or shellfish products is stably cooled to -0.5 to -2.5°C. A precise temperature recovery trigger on the micro-snow nucleation unit then sprays CO2 with a D50 of 50 to 150 μm within 30 to 120 seconds. This is combined with an ultrasonic unit providing 20 to 40 kHz pulses with 20% to 40% duty cycle, and an electric field unit providing 0.5 to 3.0 kV·cm. -1 Coupled with a unified time base of 0.5–2.0 mT by the magnetron control unit, the nucleation density is increased, the critical region of 0–-5℃ is rapidly traversed, and the temperature uniformity is ≤2.0℃; the jet humidity control unit operates according to a two-stage strategy of 80%–95%RH and 70%–85%RH at 8–15 m·s. -1 To suppress evaporation and drying loss and frost thermal resistance, the online detection unit and BP-ANN predictive control closed-loop rolling optimization optimize linear speed, air volume, humidity, CO2 flux, and ultrasonic, electric field, and magnetic field strength. It takes into account particle size and flux closed-loop and spray zone wheel shear compensation stacking obstruction to reduce dripping, softening, and cracking, reduce recrystallization, and control the core-near-surface temperature difference to ≤0.8℃. At the same time, the cleaning interlock unit interlocks and shuts down and performs CIP cleaning within 0.2s for CO2, noise, and electromagnetic over-limit. This achieves simultaneous optimization of hygiene and safety, batch consistency, energy consumption, and yield. Attached Figure Description

[0015] Figure 1 This is a system diagram of a microcrystalline cryogenic resonance chamber according to the present invention; Figure 2 This is a diagram showing the composition of the precooling unit of the present invention; Figure 3 This is a compositional diagram of the micro-snow nucleation unit of the present invention; Figure 4 This is a diagram showing the composition of the jet flow humidity control unit of the present invention. Detailed Implementation

[0016] The technical solutions of this invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described content is only a part of this invention, not all of it. Based on the content of this invention, all other solutions obtained by those skilled in the art without inventive effort are within the scope of protection of this invention.

[0017] like Figure 1 As shown, the present invention proposes a microcrystalline cryogenic resonance chamber, comprising a precooling unit, a jet humidity control unit, a micro-snow nucleation unit, an ultrasonic unit, an electric field unit, a magnetocontrol unit, a transport unit, an online detection unit, a predictive control unit, a cleaning interlock unit, and a stable cooling unit. The transport unit places the product in the center of two parallel electrode plates within the sample chamber. The electric field unit applies 0.5–3.0 kV·cm⁻¹ to the high-voltage electrode plate above and the grounding plate below. -1 The electrostatic field and magnetron unit include a pair of Helmholtz coils located outside the sample chamber sidewall, forming a uniform static magnetic field of 0.5–2.0 mT in the central region; the ultrasonic unit's transducer surface is attached to the sample chamber sidewall and the bottom acoustic coupling window faces the tray, and it is excited with 20%–40% duty pulses to nucleate in coordination with the electromagnetic field; the pre-cooling unit, under zoned temperature and humidity constraints, lowers the surface temperature to -0.5–-2.5℃ and triggers the micro-snow nucleation unit with a temperature recovery tip to spray CO2 micro-snow with a D50 of 50–150 μm in 30–120 s; the jet humidity control unit sprays CO2 micro-snow with a flow rate of 8–15 m·s. -1 It rapidly traverses the 0 to -5℃ critical zone using a relative humidity strategy of 80%–95% or 70%–85%; the online detection unit continuously collects fiber optic temperature, infrared thermal image, wind speed, humidity, carbon dioxide concentration, acoustic cavitation, and near-infrared spectrum; the predictive control unit predictively controls and optimizes linear velocity, air volume, humidity, carbon dioxide flux, ultrasonic duty cycle, electric field strength, and magnetic field strength based on the BP-ANN model to ensure temperature uniformity ≤2.0℃ and suppress dripping and recrystallization; the cleaning interlock unit shuts down and cleans on-site for CO2, noise, and electromagnetic intrusion; the stabilizing cooling unit solidifies the microcrystalline structure and suppresses recrystallization through low-temperature short-term descent and micro-perturbation stabilizing cooling.

[0018] The product is precisely placed at the center of two parallel electrode plates in the sample chamber using the stacking unit, and an electric field unit applies 0.5–3.0 kV·cm between the high-voltage electrode plate and the ground plane. -1 The electrostatic field and magnetocontrol unit use a Helmholtz coil to create a uniform static magnetic field of 0.5–2.0 mT in the central region. This magnetic field works in synergy with the transducer of the ultrasonic unit, which, through an acoustic coupling window, directs pulses toward the tray with a 20%–40% duty cycle. The pre-cooling unit lowers the surface temperature to -0.5–-2.5°C, and the micro-snow nucleation unit, triggered by a temperature recovery criterion, sprays CO2 micro-snow with a D50 of 50–150 μm. The jet humidification unit sprays CO2 micro-snow at a flow rate of 8–15 m / s.-1 Under the same unified time base, the electro-magnetic-acoustic-mass four-field coupled control is achieved, which rapidly traverses the critical zone of 0 to -5℃ using a segmented relative humidity strategy. The online detection unit uses fiber optic temperature, infrared thermal imaging, wind speed, humidity, carbon dioxide concentration, acoustic cavitation, and near-infrared spectrum as states. The predictive control unit optimizes linear velocity, air volume, humidity, carbon dioxide flux, ultrasonic duty cycle, electric field strength, and magnetic field strength based on the BP-ANN model. Compared with existing technologies that rely solely on air cooling or plate freezing, this solution improves the initial nucleation density and temperature uniformity (≤2.0℃) through temperature rebound triggering and multivariate closed-loop control, reduces dripping, and inhibits recrystallization. The cleaning interlock unit links shutdown and on-site cleaning when CO2, noise, and electromagnetic limits are exceeded. The stabilization unit solidifies the microcrystalline structure through low-temperature short-term residence and micro-perturbation stabilization, improving the industrial-grade freezing quality of large seafood products or shell-on shrimp and crabs. The stabilization unit is scheduled by the predictive control unit to ensure smooth temperature trajectory and improved product consistency, and linear velocity and air volume operate stably in coordination.

[0019] It should be noted that, as Figure 2 As shown, the precooling unit includes a low-temperature evaporator, a fan, a guide duct, rectifier louvers, an infrared thermal imager, and fiber optic multi-point temperature probes. The variable frequency convection fan supplies air to the low-temperature evaporator through the guide duct. The rectifier louvers are sequentially connected to the air outlet side of the low-temperature evaporator, forming a parallel airflow above the stacking unit. The infrared thermal imager provides a top-down view of the product surface, and the fiber optic multi-point temperature probes are positioned near the product surface and core. The low-temperature evaporator achieves an evaporation temperature of -15 to -25°C at 8–12 m / s thanks to the frequency modulation of the fan and the coordination of the rectifier louvers. -1 Zoned convection is implemented to maintain a relative humidity of 80%–95% or 70%–85%. Gradual cooling and convection homogenization are carried out on the products in the transport unit to reduce the surface temperature to -0.5 to -2.5℃ under conditions of temperature uniformity ≤2.0℃. A monitoring array consisting of an infrared thermal imager and fiber optic multi-point temperature probes calculates the temperature change rate within a 3–5s sliding window. When ≥60% of the measuring points show a change from negative to positive and ≥0.05℃·s, the temperature change rate is recorded. -1 And when the temperature rises for ≥5 seconds, a trigger command is output to the micro-snow nucleation unit.

[0020] Unlike traditional methods that rely on experience to set precooling time or a single temperature threshold, the precooling unit proposed in this invention uses a low-temperature evaporator, a fan, a guide duct, and rectifier louvers to form a parallel airflow. Combined with an infrared thermal imager and a fiber optic multi-point temperature probe, it achieves precooling at evaporation temperatures of -15 to -25°C and temperatures of 8 to 12 m / s. -1 Zoned convection, closed-loop control at 80%–95% RH or 70%–85% RH; and, under the constraint of temperature uniformity ≤2.0℃, reducing the surface temperature to -0.5 to -2.5℃, with "≥60% of measuring points turning from negative to positive and ≥0.05℃·s". -1The system uses a sustained warming peak of ≥5 seconds as the trigger criterion to link the micro-snow nucleation unit, ensuring precise nucleation timing, avoiding overcooling delays and repeated triggering, improving initial nucleation density and temperature consistency, suppressing dryness loss and cracking, reducing dripping, shortening critical zone dwell time, and reducing energy consumption, resulting in more stable freezing quality. Simultaneously, a dual humidity segment switching strategy is introduced to adaptively balance dew point margin and frost trend, with the determination achieved by monitoring the temperature change rate of the array's 3-5 second sliding window. Compared to systems that only monitor a single surface temperature, this invention employs multi-point fusion of near-surface and core temperatures, combined with coordinated tuning of linear velocity and airflow, to avoid localized hot and cold spots caused by stacking obstruction, improving synchronous cooling of the thick core and surface layers and batch-to-batch consistency.

[0021] It should be noted that in the pre-cooling unit, the relative humidity is adaptively selected by the predictive control unit based on dew point margin and quality risk: when the dew point margin calculated by the infrared thermal imager and temperature and humidity sensor is equal to the surface temperature minus the dew point temperature, the dew point margin is ≥3~5℃, and the dripping risk predicted by the online detection unit is >3% or the glaze layer needs to be protected, the RH is maintained at 80%~95%; when the dew point margin is ≤2℃ or the temperature uniformity is >2.0℃ or a frosting trend appears, the frosting trend criterion is a relative humidity rise rate ≥1%·10s. -1 Furthermore, if the infrared cold spot area accounts for ≥10%, switch to 70%~85%RH; and set a ≥60s hysteresis, switch back to 80%~95%RH after the dew point margin recovers to ≥3℃ and the frost criterion is removed.

[0022] Compared to traditional pre-cooling methods that rely solely on fixed humidity or single-point temperature settings, this invention uses "dew point margin = surface temperature - dew point temperature" as the core criterion. It combines infrared thermal imaging and temperature and humidity sensors to calculate the dew point margin in real time, and introduces quality risk quantification and trend characteristics (drip loss prediction > 3%, relative humidity rise rate ≥ 1%·10s). -1 Dual-range adaptive switching and ≥60s hysteresis control (infrared cold spot area ratio ≥10%): When the dew point margin is ≥3~5℃ or glaze protection is required, maintain 80%~95%RH; when the dew point margin is ≤2℃ or frost trend appears, switch to 70%~85%RH. This actively reduces frost thermal resistance while suppressing evaporation and glaze loss, avoids humidity fluctuations and false triggering, shortens the 0~-5℃ dwell time, improves surface-core synchronization and batch temperature uniformity, reduces energy consumption and stabilizes the subsequent ice crystal nucleation quality.

[0023] It should be noted that, as Figure 3As shown, the micro-snow nucleation unit includes, in sequence, a stabilizing tank, a vacuum-insulated liquid supply pipeline, a main solenoid valve, a zone shut-off valve, a Coriolis mass flow meter, an atomizing chamber, a back pressure regulating valve, and a nozzle. An online laser particle size monitor is also installed to measure the D50 of the spray cone. After the predictive control unit is triggered, the main solenoid valve opens for 30–120 seconds according to the three-segment flux curve of ignition-platform-fallback. The Coriolis mass flow meter, in a closed loop, stabilizes the mass flow rate of a single nozzle at 5–20 g·s⁻¹. -1 The laser particle size online monitor outputs D50 in real time and feeds it back to the predictive control unit to control D50 within 50–150 μm via a back pressure regulating valve. The zone cut-off valves operate in staggered phases according to the stacking obstruction to ensure uniform coverage. This is in conjunction with the ultrasonic unit's 20%–40% duty cycle pulses and the electric field unit's 0.5–3.0 kV·cm² pulses. -1 It works in conjunction with the magnetron control unit (0.5–2.0 mT) to induce multi-point nucleation; when CO2 ≥ 5000 ppm is maintained for ≥ 5 s or particle size deviation > |10|% is maintained for ≥ 2 s, it automatically reduces the flux to 50% of the initial value or terminates the injection and resets according to the hysteresis logic.

[0024] Unlike traditional spraying methods that rely on fixed spray volume, open-loop atomization, and single triggering, this invention utilizes a liquid stabilizing tank, a vacuum-insulated liquid supply pipeline, a Coriolis mass flow meter, a laser particle size online monitor, and a back pressure regulating valve to form a "flux-particle size" closed loop. This is combined with a three-stage flux curve (ignition-plateau-fall) from the main solenoid valve and staggered phase switching by a zone-based shut-off valve, along with 20%–40% duty cycle ultrasonic spraying at 0.5–3.0 kV·cm⁻¹. -1 The electric field and the 0.5–2.0 mT magnetic field work together to self-tune according to D50 feedback within a 30–120 s window. When CO2 ≥ 5000 ppm or D50 deviation > |10|% for ≥ 2 s, the flux is reduced to 50% or the spraying is stopped with hysteresis recovery, thereby stabilizing D50 at 50–150 μm and 5–20 g·s per nozzle. -1 Controlled, with sufficient shading compensation, improved nucleation density and spatial uniformity, accelerated 0 to -5℃ crossing, reduced frost thermal resistance and dripping, temperature uniformity ≤2.0℃, suppressed recrystallization and batch-to-batch fluctuations while taking into account energy consumption and safety.

[0025] It should be noted that the ultrasonic unit, electric field unit, and magnetocontrol unit employ unified time-base phase coupling control issued by the predictive control unit: within the injection window of the micro-snow nucleation unit, a phase of 0.5–3.0 kV·cm is applied between the two parallel electrode plates in the sample chamber by the electric field unit for 0.3–1.0 s prior to the phase coupling. -1 An electrostatic field is used for orientation pretreatment, followed by an ultrasonic unit operating with a 20–40 kHz pulse train with 20%–40% duty cycle, overlapping the electric field for 60%–80% of the time. A magnetron unit generates a uniform static magnetic field of 0.5–2.0 mT on the outer sidewall of the sample chamber via a pair of Helmholtz coils, with a pulse duration ≤0.2 mT·5 s. -1The change is gradual, and the direction of the magnetic field is orthogonal to the direction of ultrasonic propagation; The acoustic cavitation intensity and temperature uniformity indices of the online detection unit, along with the infrared thermal gradient and near-infrared dripping prediction, form a feedback loop and are adjusted according to the following coupling law in a closed loop: when the acoustic cavitation intensity exceeds the set value and lasts for ≥2s, the ultrasonic duty cycle is reduced by 5%–10%, and the electric field strength is increased by 0.2–0.5 kV·cm. -1 When the temperature uniformity is >2.0℃ and lasts for ≥5s, the magnetic field strength is increased by 0.2–0.5mT and the ultrasonic frequency is decreased by 2–5kHz; when the predicted droplet loss is >3%, the electric field preprocessing window is expanded to 0.6–1.2s and the ultrasonic duty cycle is reduced to 20%–30%; the variation of each parameter is limited to an electric field ≤0.5kV·cm every 5s. -1 The system features a magnetic field ≤ 0.2 mT, an ultrasonic duty cycle ≤ 10%, and automatically pauses ultrasound when the CO2 concentration is ≥ 5000 ppm and lasts for ≥ 5 seconds, while reducing the electric field to 0.5 kV·cm. -1 .

[0026] Unlike traditional schemes that drive ultrasonic, electric field, or magnetocontrol units separately, in an open-loop, fixed-parameter manner (without synchronization, phase relationship, or multi-source feedback and orientation preprocessing), this invention employs unified time-base phase coupling control issued by a predictive control unit: electric field orientation preprocessing is first performed within the injection window of the micro-snow nucleation unit, followed by 60%–80% time overlap with 20–40kHz ultrasonic pulses with 20%–40% duty cycle, and cavitation and heat transfer are stabilized by an orthogonal, slowly varying 0.5–2.0mT static magnetic field from the magnetocontrol unit. The online detection unit performs closed-loop parameter adjustment and sets the rate of change and CO2 over-limit safety limits, thereby improving the consistency and density of multi-point nucleation, shortening the 0–-5℃ traversal time, suppressing coarse ice crystals and recrystallization, reducing dripping and cracking, controlling temperature uniformity to ≤2.0℃, and reducing batch-to-batch fluctuations and energy consumption.

[0027] It should be noted that, as Figure 4As shown, the jet flow humidity control unit includes a second fan, an air supply box, a guide duct, two rectifier louvers, zoned air supply outlets, an air volume regulating valve, a zoned return air damper, a condenser dehumidification coil, a hot gas reheat valve, a bypass ventilation damper, an ultrapure water mist generator, a dew point mirror sensor, a temperature and humidity sensor, a wind speed sensor, an infrared thermal imager observation window, a feed air curtain, and an air knife. The second fan is connected to the guide duct via the air supply box. The guide duct is equipped with two rectifier louvers and zoned air supply outlets at its front end. A pre-installed airflow regulating valve is placed at the air supply outlet of the zone, and a zoned return air damper is arranged on the return air side. The condenser dehumidification coil and the bypass ventilation damper are connected in parallel to the air supply path. The hot gas reheat valve and the ultrapure water micro-mist generator are connected in series after the condenser dehumidification coil. Dew point mirror sensor, temperature and humidity sensor, and wind speed sensor are respectively installed at preset positions in the supply and return air. The feed air curtain and air knife are located at the air supply inlet to form an isolation air curtain. Under the control of the predictive control unit, the fan frequency is 10-60Hz and the outlet air velocity is 8-15m·s. -1 Based on the dew point margin and CO2 injection events, the condenser dehumidification coil, hot gas reheat valve, bypass ventilation damper and ultrapure water micro-mist device are coordinated to form a relative humidity range of 80% to 95% or 70% to 85%, and the temperature uniformity is maintained at ≤2.0℃ by cooperating with the air volume regulating valve and the zoned return air damper.

[0028] Unlike traditional solutions with fixed wind speed, single dehumidification, or single PID humidity control that lack direct dew point measurement and nucleation linkage, the jet flow humidity control unit of this invention uses a reconfigurable air path consisting of a second fan, an air supply box, a guide duct, a second rectifier louver, zoned air supply outlets, an air volume regulating valve, a zoned return air damper, a condenser dehumidification coil, a bypass ventilation damper, a hot gas reheat valve, and an ultrapure water micro-mist generator. Under the command of the predictive control unit, it is driven in a closed loop using an infrared thermal imager observation window, a dew point mirror sensor, a temperature and humidity sensor, and a wind speed sensor, keeping the second fan at 10–60 Hz and the outlet wind speed at 8–15 m / s. -1 The system adaptively switches between 80%–95%RH and 70%–85%RH based on the dew point margin and CO2 injection events, and balances the recirculation in each zone with the air volume regulating valve and the zoned recirculation damper. The feed air curtain and air knife suppress the intrusion of external heat and moisture, thereby maintaining temperature uniformity ≤2.0℃ and dew point margin ≥3℃. In the critical zone of 0 to -5℃, it simultaneously suppresses evaporation dry loss and frosting thermal resistance, protects the glaze layer, avoids surface overcooling and cracking, improves CO2 jet coverage and convective heat transfer, shortens the time to cross the ice crystal zone, and reduces energy consumption and batch-to-batch fluctuations.

[0029] It should be noted that when the online detection unit determines that the CO2 concentration reaches or exceeds 5000ppm and lasts for at least 5 seconds, or the acoustic cavitation intensity exceeds the limit, or the output of the electric field unit and the magnetic control unit exceeds the limit, the cleaning interlock unit will sequentially issue cleaning interlocks within 0.2 seconds: instructing the micro-snow nucleation unit to close the main solenoid valve and the zone shut-off valve, and to place the back pressure regulating valve in the pressure relief position, and to use the air knife of the jet wet control unit to dry blow the nozzle for 3-10 seconds to remove residual micro-snow and frost; instructing the jet wet control unit to raise the fan speed to no less than 50Hz and fully open the feed air curtain door and the zone return air door to establish a fresh air channel, and at the same time start the ultrapure water micro-mist generator to flush the atomization chamber outlet with a micro-atomization of 1.0-1.5 liters per minute per nozzle for 2-4 minutes. The air supply dew point is kept below 5°C for 5-10 minutes by the condensation dehumidification coil and the bypass ventilation door to condense and collect the dissolved and entrained contaminants. Then, the hot air reheat valve is opened to raise the air supply above the dew point and maintain it for 10-15 minutes to complete the drying. During this process, the infrared thermal imager, fiber optic multi-point temperature probe, temperature and humidity sensor and dew point mirror sensor are used as release criteria. After the CO2 concentration is not higher than 1000ppm, the relative humidity is not higher than 85% and the equipment temperature is not higher than 30 degrees Celsius and stabilizes for 60 seconds, the cleaning interlock unit is released, the jet humidity control unit returns to the normal humidity range, and the stacking unit moves the in-place product out of the cleaning area at the guaranteed bottom speed. The entire process timestamp and threshold status are sent back to the predictive control unit for archiving.

[0030] After the above-mentioned cleaning interlock is detected by the online detection unit, the predictive control unit issues a sequence of isolation, ventilation, flushing, dehumidification, reheating, and reset within 0.2 seconds. This causes the main solenoid valve and the zone shut-off valve of the micro-snow nucleation unit to close instantaneously, and the back pressure regulating valve to release pressure. The air knife of the jet humidity control unit briefly dries and removes frost from the nozzles. At the same time, the second fan is raised to ≥50Hz, and the feed air curtain and the zone return air door are fully opened to create a high ventilation channel. In conjunction with the ultrapure water micro-mist device, a small amount of atomized rinsing is performed on the outlet area of ​​the atomization chamber and the wall of the sample chamber. Then, the supply air dew point is lowered to <5℃ by the condensation dehumidification coil and the bypass ventilation door and maintained for 5-10 minutes to promote the condensation and removal of pollutants. Subsequently, the hot gas reheat valve is heated. Dry for 10-15 minutes; under the combined monitoring and release criteria of infrared thermal imager, fiber optic multi-point temperature probe, temperature and humidity sensor and dew point mirror sensor, the system automatically unlocks when CO2 ≤ 1000ppm, relative humidity ≤ 85%, equipment temperature ≤ 30℃ and stabilizes for 60 seconds. The stacking unit removes the product at the guaranteed minimum speed. The entire process timestamp and threshold status are backed up and archived, thereby achieving integrated personnel and equipment safety, rapid risk reduction, on-site decontamination and drying, avoiding secondary pollution and batch cross-contamination, reducing downtime and restoring steady-state humidity control and heat exchange conditions, and ensuring unobstructed nozzles and stable sample chamber cleanliness level, traceable parameters, and compliance with regulatory limits.

[0031] It should be noted that the stable cooling unit adopts a three-stage temperature step plus perturbation stabilization method: the first stage is -38℃ for 4-8 minutes, the second stage is -40℃ for 6-12 minutes, and the third stage is -42℃ for 3-6 minutes; throughout the process, temperature perturbation amplitude is 0.3-0.6℃, oscillation period is 120-240s, and airflow micro-pulse amplitude is 10%-15%, single pulse duration is 10-20s, interval is 2-4 minutes, and relative humidity is 65%-80%; the stable cooling unit calculates the recrystallization risk index based on the surface and core temperatures, temperature uniformity, and frost thickness measured by the near-infrared spectrum of the online detection unit and the infrared thermal imager. The recrystallization risk index is equal to the unfrozen index. The water ratio is multiplied by the relative temperature margin, the temperature fluctuation intensity, and the proportion of time the sample spends in the active region of -15 to -5℃. The relative temperature margin is the proportion of the current temperature above the glass transition temperature to the difference between the melting point and the glass transition temperature. If the temperature is not higher than the glass transition temperature, the relative temperature margin is zero. The temperature fluctuation intensity is the product of the temperature fluctuation amplitude and the zero-crossing frequency within the near time window. When the recrystallization risk index is >1.0 and lasts for ≥5s, the stabilization cooling unit automatically extends the second stage by 2-4 minutes and increases the air volume by 10%, while reducing the temperature perturbation amplitude by 0.1℃. Stabilization cooling is released and the sample is switched to the out-of-pack mode when the recrystallization risk index is <0.8 and the temperature difference between the core and the near surface is ≤0.8℃.

[0032] A stable cooling strategy of "three-stage temperature dwell + temperature perturbation + airflow micro-pulse + relative humidity constraint" is adopted, and closed-loop parameter tuning is performed using a "recrystallization risk index" composed of near-infrared and infrared radiation. This is because unfrozen water and minor thermal disturbances are most likely to trigger secondary ice crystal growth during the initial stable glass phase. Segmented dwell provides a heat release channel from the surface to the interior, while 0.3–0.6℃ perturbation and 10%–15% airflow pulses at 65%–80% RH disperse localized overcooling, reduce dry loss, and decrease the tendency for surface cracking. The system comprehensively judges the performance based on the proportion of unfrozen water, relative temperature margin, temperature fluctuation intensity, and time weight in the active range of -15 to -5℃. When the index is >1.0 and lasts for ≥5s, the -40℃ range is adaptively extended by 2 to 4 minutes, the air volume is increased by 10%, and the micro-disturbance is reduced by 0.1℃ until the index is <0.8 and the core-near-surface ΔT is ≤0.8℃. This suppresses recrystallization, stabilizes the microcrystalline structure, reduces dripping, improves color and texture fidelity, balances energy consumption and batch consistency, and strengthens traceability safety boundaries.

[0033] It should be noted that before the stabilization cooling unit begins, the online detection unit generates a batch profile vector of large seafood products. This batch profile vector includes thickness, fat percentage, glaze ratio, near-infrared feature values, conveyor line speed, and jet parameters. The predictive control unit searches for similar batch templates in the historical parameter database with a nearest neighbor count of 5 and outputs the temperature level setting, residence time, temperature perturbation amplitude, and airflow pulse amplitude as the initial strategy. During execution, exploratory fine-tuning is performed within the safety boundary with a probability of 5%, with temperature level fine-tuning not exceeding ±0.5℃ and single-segment residence time fine-tuning not exceeding [a certain value]. The adjustment interval is no less than 60 seconds. The objective function is composed of a weighted average of dripping prediction, temperature uniformity, and unit energy consumption. If the objective function decreases, the current parameters are fixed and written into the historical parameter library to increase the weight. If it does not decrease, it is rolled back to the previous parameter set. The CO2 concentration is constrained to be no higher than 5000ppm and the temperature uniformity is no higher than 2.0℃ throughout the process. When the temperature difference between the core and the near surface is no higher than 0.8℃ and the dripping prediction is no higher than 3%, the stable cooling is considered to be completed.

[0034] Before the start of stable cooling, the online detection unit generates a batch profile including thickness, fat content, glazing ratio, near-infrared characteristic values, conveyor line speed, and jet parameters. The predictive control unit searches the historical parameter database for similar templates with a nearest neighbor count of 5 to provide an initial strategy for temperature level, dwell time, perturbation, and airflow pulse. It explores in small steps within the safety boundary with a 5% probability (temperature level ±0.5℃, dwell time ±10%, airflow pulse ±5%, relative humidity ±3%, interval ≥60s). The objective function of "drip prediction + temperature uniformity + unit energy consumption" is used for optimization. If the temperature decreases, it is solidified and stored in the database; if it does not decrease, it is rolled back. At the same time, CO2 ≤5000ppm and temperature uniformity ≤2.0℃ are constrained throughout the process. The completion is determined by the core-near surface temperature difference ≤0.8℃ and drip prediction ≤3%. This is an adaptive stable cooling and empirical deposition that takes into account the differences in raw materials and thermal inertia between different batches. The technical effect is to suppress recrystallization and dry loss, reduce energy consumption and inter-batch fluctuations, ensure color and texture, and enhance process traceability and safety boundaries.

[0035] It should be noted that in the predictive control unit, the BP-ANN model uses process quantity branches and near-infrared branches for input encoding. Each of the process quantity and near-infrared branches undergoes two fully connected layers for extraction. After branch fusion, a time-series modeling layer is introduced to capture the dynamics of the critical region. A multi-task output head is used at the top, simultaneously regressing droplet loss, hardness, recrystallization risk index, and the target time of the critical region. A physical constraint gating layer is connected between the fusion and output layers to perform key monotonic relationships and boundary projections. A physical constraint loss is added to the total loss, resulting in the following total loss function: ; In the formula: For the total loss function, For the number of tasks, , For task indexing, For the first The loss weight of each task, For the first The target variable for each task For the first Mean squared error of each task objective variable The weight of the physical constraint penalty term. As a physical consistency penalty item, for violations , , and The degree of monotonicity is measured, and the positive deviations of each constraint are aggregated using a squared penalty. , , , These are the model's prediction functions for dripping loss, hardness, recrystallization risk index, and critical zone target time, respectively. , , These are relative humidity, dew point margin, and temperature uniformity index, respectively. Dew point margin is equal to the measured surface temperature minus the dew point temperature, and temperature uniformity index is equal to the standard deviation of the temperature at each measuring point. The bypass branch outputs heteroscedasticity and supports Monte Carlo estimation confidence intervals; a missing measurement mask and time decay interpolation layer are added to the input side; a sensitivity linearization layer is connected in parallel at the output to generate gradients for linear velocity, air volume, relative humidity, CO2 flux, ultrasonic duty cycle, and magnetic field strength for predictive control; at the same time, a drift correction layer and a soft limiting layer are arranged, the former performing statistical alignment across batches and product types, and the latter mapping the suggested settings to the device safety domain.

[0036] Addressing the technical challenges of sparse nucleation, discrete temperature field, and narrow parameter windows in the critical region of 0–-5℃ for large seafood and shellfish such as shrimp and crabs with large volume, thick cuts, and uneven distribution of fat and fascia, the proposed BP-ANN, based on the fusion of process quantities and near-infrared dual branches, time-series modeling, and physical constraint gating, combines uncertainty estimation, missing measurement masking and time decay interpolation, sensitivity linearization, drift correction, and soft limiting. It outputs controllable gradients for linear velocity, air volume, relative humidity, CO2 flux, ultrasonic duty cycle, electric field strength, and magnetic field strength, and maps them to the safety region. This enables predictive control to robustly shorten the critical region crossing time, reduce temperature uniformity to ≤2.0℃, stabilize D50 and flux to increase multi-point nucleation density, thereby reducing dripping and desiccation, inhibiting recrystallization and reducing cracking and glaze loss, and reducing energy consumption and the risk of exceeding limits.

[0037] In summary, this invention, through a pre-cooling unit operating under zoned convection and dew point adaptive humidity control, stably lowers the surface temperature of large seafood products or shellfish products to -0.5 to -2.5°C. A temperature recovery tip precisely triggers a micro-snow nucleation unit to inject CO2 with a D50 of 50 to 150 μm within 30 to 120 seconds. This is combined with an ultrasonic unit providing 20 to 40 kHz pulses with 20% to 40% duty cycle and an electric field unit providing 0.5 to 3.0 kV·cm. -1 Coupled with a unified time base of 0.5–2.0 mT by the magnetron control unit, the nucleation density is increased, the critical region of 0–-5℃ is rapidly traversed, and the temperature uniformity is ≤2.0℃; the jet humidity control unit operates according to a two-stage strategy of 80%–95%RH and 70%–85%RH at 8–15 m·s. -1 To suppress evaporation and drying loss and frost thermal resistance, the online detection unit and BP-ANN predictive control closed-loop rolling optimization optimize linear speed, air volume, humidity, CO2 flux, and ultrasonic, electric field, and magnetic field strength. It takes into account particle size and flux closed-loop and spray zone wheel shear compensation stacking obstruction to reduce dripping, softening, and cracking, reduce recrystallization, and control the core-near-surface temperature difference to ≤0.8℃. At the same time, the cleaning interlock unit interlocks and shuts down and performs CIP cleaning within 0.2s for CO2, noise, and electromagnetic over-limit. This achieves simultaneous optimization of hygiene and safety, batch consistency, energy consumption, and yield.

[0038] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0039] Finally: The above description is only a specific embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A microcrystalline cryogenic resonance chamber, characterized in that, It includes a pre-cooling unit, a jet humidity control unit, a micro-snow nucleation unit, an ultrasonic unit, an electric field unit, a magnetocontrol unit, a transport unit, an online detection unit, a predictive control unit, a cleaning interlock unit, and a stable cooling unit. The transport unit places the product in the center of two parallel electrode plates within the sample chamber. The electric field unit applies 0.5–3.0 kV·cm⁻¹ to the high-voltage electrode plate above and the grounding plate below. -1 The electrostatic field and magnetron unit include a pair of Helmholtz coils located outside the sample chamber sidewall, forming a uniform static magnetic field of 0.5–2.0 mT in the central region; the ultrasonic unit's transducer surface is attached to the sample chamber sidewall and the bottom acoustic coupling window faces the tray, and it is excited with 20%–40% duty pulses to nucleate in coordination with the electromagnetic field; the pre-cooling unit, under zoned temperature and humidity constraints, lowers the surface temperature to -0.5–-2.5℃ and triggers the micro-snow nucleation unit with a temperature recovery tip to spray CO2 micro-snow with a D50 of 50–150 μm in 30–120 s; the jet humidity control unit sprays CO2 micro-snow with a flow rate of 8–15 m·s. -1 And it rapidly traverses the 0 to -5℃ critical zone using a relative humidity strategy of 80% to 95% or 70% to 85%.

2. The microcrystalline cryogenic resonance chamber according to claim 1, characterized in that, The online detection unit continuously collects fiber optic temperature, infrared thermal images, wind speed, humidity, carbon dioxide concentration, acoustic cavitation, and near-infrared spectra; the predictive control unit predictively controls and optimizes linear velocity, airflow, humidity, carbon dioxide flux, ultrasonic duty cycle, electric field strength, and magnetic field strength based on the BP-ANN model to ensure temperature uniformity ≤2.0℃ and suppress dripping and recrystallization; the cleaning interlock unit shuts down and cleans on-site in response to CO2, noise, and electromagnetic intrusion; and the stabilizing cooling unit uses low-temperature short-term holding and micro-perturbation stabilizing cooling to solidify the microcrystalline structure and suppress recrystallization.

3. The microcrystalline cryogenic resonance chamber according to claim 1, characterized in that, The precooling unit includes a low-temperature evaporator, a fan, a guide duct, rectifier louvers, an infrared thermal imager, and fiber optic multi-point temperature probes. The variable-frequency convection fan supplies air to the low-temperature evaporator through the guide duct. The rectifier louvers are sequentially connected to the air outlet side of the low-temperature evaporator, forming a parallel airflow above the stacking unit. The infrared thermal imager provides a top-down view of the product surface, and the fiber optic multi-point temperature probes are positioned near the product surface and core. The low-temperature evaporator achieves an evaporation temperature of -15 to -25°C at 8–12 m / s thanks to the frequency modulation of the fan and the coordination of the rectifier louvers. -1 Zoned convection is implemented to maintain a relative humidity of 80%–95% or 70%–85%. Gradual cooling and convection homogenization are carried out on the products in the transport unit to reduce the surface temperature to -0.5 to -2.5℃ under conditions of temperature uniformity ≤2.0℃. A monitoring array consisting of an infrared thermal imager and fiber optic multi-point temperature probes calculates the temperature change rate within a 3–5s sliding window. When ≥60% of the measuring points show a change from negative to positive and ≥0.05℃·s, the temperature change rate is recorded. -1 And when the temperature rises for ≥5 seconds, a trigger command is output to the micro-snow nucleation unit; In the pre-cooling unit, the relative humidity is adaptively selected by the predictive control unit based on dew point margin and quality risk: when the dew point margin calculated by the infrared thermal imager and temperature and humidity sensor is equal to the surface temperature minus the dew point temperature, the dew point margin is ≥3~5℃, and the dripping risk predicted by the online detection unit is >3% or the glaze layer needs to be protected, the RH is maintained at 80%~95%; when the dew point margin is ≤2℃ or the temperature uniformity is >2.0℃ or a frosting trend appears, the frosting trend criterion is a relative humidity rise rate ≥1%·10s. -1 Furthermore, if the infrared cold spot area accounts for ≥10%, switch to 70%~85%RH; and set a ≥60s hysteresis, switch back to 80%~95%RH after the dew point margin recovers to ≥3℃ and the frost criterion is removed.

4. A microcrystalline cryogenic resonance chamber according to claim 3, characterized in that, The micro-snow nucleation unit includes, in sequence, a stabilizing tank, a vacuum-insulated liquid supply pipeline, a main solenoid valve, a zone shut-off valve, a Coriolis mass flow meter, an atomizing chamber, a back pressure regulating valve, and a nozzle. An online laser particle size monitor is also installed to measure the D50 of the spray cone. After the predictive control unit is triggered, the main solenoid valve opens for 30–120 seconds according to a three-stage flux curve of ignition-plateau-decline. The Coriolis mass flow meter, in a closed loop, stabilizes the mass flow rate of a single nozzle at 5–20 g·s⁻¹. -1 The laser particle size online monitor outputs D50 in real time and feeds it back to the predictive control unit to control D50 within 50–150 μm via a back pressure regulating valve. The zone cut-off valves operate in staggered phases according to the stacking obstruction to ensure uniform coverage. This is in conjunction with the ultrasonic unit's 20%–40% duty cycle pulses and the electric field unit's 0.5–3.0 kV·cm² pulses. -1 It works in conjunction with the magnetron control unit (0.5–2.0 mT) to induce multi-point nucleation; when CO2 ≥ 5000 ppm is maintained for ≥ 5 s or particle size deviation > |10|% is maintained for ≥ 2 s, it automatically reduces the flux to 50% of the initial value or terminates the injection and resets according to the hysteresis logic.

5. A microcrystalline cryogenic resonance chamber according to claim 1, characterized in that, The ultrasonic unit, electric field unit, and magnetocontrol unit employ unified time-base phase coupling control issued by the predictive control unit: within the injection window of the micro-snow nucleation unit, a period of 0.3–1.0 s is first established by the electric field unit applying 0.5–3.0 kV·cm⁻¹ between the two parallel electrode plates in the sample chamber. -1 An electrostatic field is used for orientation pretreatment, followed by an ultrasonic unit operating with a 20–40 kHz pulse train with 20%–40% duty cycle, overlapping the electric field for 60%–80% of the time. A magnetron unit generates a uniform static magnetic field of 0.5–2.0 mT on the outer sidewall of the sample chamber via a pair of Helmholtz coils, with a pulse duration ≤0.2 mT·5 s. -1 The change is gradual, and the direction of the magnetic field is orthogonal to the direction of ultrasonic propagation; The acoustic cavitation intensity and temperature uniformity indices of the online detection unit, along with the infrared thermal gradient and near-infrared dripping prediction, form a feedback loop and are adjusted according to the following coupling law in a closed loop: when the acoustic cavitation intensity exceeds the set value and lasts for ≥2s, the ultrasonic duty cycle is reduced by 5%–10%, and the electric field strength is increased by 0.2–0.5 kV·cm. -1 When the temperature uniformity is >2.0℃ and lasts for ≥5s, the magnetic field strength is increased by 0.2–0.5mT and the ultrasonic frequency is decreased by 2–5kHz; when the predicted droplet loss is >3%, the electric field preprocessing window is expanded to 0.6–1.2s and the ultrasonic duty cycle is reduced to 20%–30%; the variation of each parameter is limited to an electric field ≤0.5kV·cm every 5s. -1 The system automatically pauses ultrasound and reduces the electric field to 0.5 kV·cm when the magnetic field is ≤0.2 mT, the ultrasound duty cycle is ≤10%, and the CO2 concentration is ≥5000 ppm for ≥5 s. -1 .

6. A microcrystalline cryogenic resonance chamber according to claim 1, characterized in that, The jet flow humidity control unit includes a second fan, an air supply box, a guide duct, two rectifier louvers, zoned air supply outlets, an air volume regulating valve, zoned return air dampers, a condensate dehumidification coil, a hot gas reheat valve, a bypass ventilation damper, an ultrapure water mist generator, a dew point mirror sensor, a temperature and humidity sensor, a wind speed sensor, an infrared thermal imager observation window, a feed air curtain, and an air knife. The second fan is connected to the guide duct via the air supply box. The guide duct is equipped with two rectifier louvers and zoned air supply outlets at its front end. Each zone's air supply... A pre-installed airflow regulating valve is placed at the air outlet, and a zoned return air damper is arranged on the return air side. The condenser dehumidification coil and the bypass ventilation damper are connected in parallel to the supply air path. The hot gas reheat valve and the ultrapure water micro-mist generator are connected in series after the condenser dehumidification coil. Dew point mirror sensors, temperature and humidity sensors, and wind speed sensors are respectively installed at preset positions in the supply and return air. The feed air curtain and air knife are located at the supply air inlet to form an isolation air curtain. Under the control of the predictive control unit, the fan frequency is 10-60Hz and the outlet air velocity is 8-15m·s. -1 Based on the dew point margin and CO2 injection events, the condenser dehumidification coil, hot gas reheat valve, bypass ventilation damper and ultrapure water micro-mist device are coordinated to form a relative humidity range of 80% to 95% or 70% to 85%, and the temperature uniformity is maintained at ≤2.0℃ by cooperating with the air volume regulating valve and the zoned return air damper.

7. A microcrystalline cryogenic resonance chamber according to claim 4, characterized in that, When the online detection unit determines that the CO2 concentration reaches or exceeds 5000 ppm and lasts for at least 5 seconds, or the acoustic cavitation intensity exceeds the limit, or the output of the electric field unit and the magnetic control unit exceeds the limit, the cleaning interlock unit will sequentially issue cleaning interlocks within 0.2 seconds: instructing the micro-snow nucleation unit to close the main solenoid valve and the zone shut-off valve, and to place the back pressure regulating valve in the pressure relief position, and to use the air knife of the jet humidity control unit to dry blow the nozzle for 3-10 seconds to remove residual micro-snow and frost; instructing the jet humidity control unit to raise the fan speed to no less than 50 Hz and fully open the feed air curtain door and the zone return air door to establish a fresh air channel, and at the same time start the ultrapure water micro-mist generator to rinse the atomization chamber outlet area and the sample with micro-atomization at 1.0-1.5 liters per minute per nozzle for 2-4 minutes. The indoor walls are then condensed and dehumidified by a condensing dehumidifying coil in conjunction with a bypass ventilation damper to maintain the supply air dew point below 5°C for 5-10 minutes to condense and collect the dissolved and entrained contaminants. Subsequently, the hot air reheat valve is opened to raise the supply air above the dew point and maintain this for 10-15 minutes to complete the drying process. During this period, an infrared thermal imager, fiber optic multi-point temperature probe, temperature and humidity sensor, and dew point mirror sensor are used together as release criteria. After the CO2 concentration is not higher than 1000ppm, the relative humidity is not higher than 85%, and the equipment temperature is not higher than 30°C and stabilizes for 60 seconds, the cleaning interlock unit is released, the jet humidity control unit returns to the normal humidity range, and the stacking unit moves the in-place products out of the cleaning area at the guaranteed baseline speed. The entire process timestamp and threshold status are then transmitted back to the predictive control unit for archiving.

8. A microcrystalline cryogenic resonance chamber according to claim 1, characterized in that, The stabilization cooling unit employs a three-stage temperature gradient plus perturbation stabilization method: the first stage is a residence at -38℃ for 4–8 minutes, the second stage is a residence at -40℃ for 6–12 minutes, and the third stage is a residence at -42℃ for 3–6 minutes. Throughout the process, temperature perturbation amplitudes of 0.3–0.6℃ and oscillation periods of 120–240 seconds are superimposed, along with airflow micro-pulses with amplitudes of 10%–15%, durations of 10–20 seconds per pulse, and intervals of 2–4 minutes. Relative humidity is maintained at 65%–80%. The stabilization cooling unit calculates a recrystallization risk index based on surface and core temperatures, temperature uniformity, and frost thickness measured by the near-infrared spectroscopy of the online detection unit and an infrared thermal imager. The recrystallization risk index is equal to the proportion of unfrozen water multiplied by... The relative temperature margin is multiplied by the temperature fluctuation intensity and the proportion of time the sample spends in the active region of -15 to -5℃. The relative temperature margin is the proportion of the current temperature above the glass transition temperature to the difference between the melting point and the glass transition temperature. If it is not higher than the glass transition temperature, the relative temperature margin is zero. The temperature fluctuation intensity is the product of the temperature fluctuation amplitude and the zero crossing frequency within the near time window. When the recrystallization risk index is >1.0 and lasts for ≥5s, the stabilization cooling unit automatically extends the second stage by 2 to 4 minutes and increases the air volume by 10%, while reducing the temperature perturbation amplitude by 0.1℃, until the recrystallization risk index is <0.8 and the temperature difference between the core and the near surface is ≤0.8℃, at which point the stabilization cooling is released and the sample is switched to the out-of-pack mode.

9. A microcrystalline cryogenic resonance chamber according to claim 1, characterized in that, Before the stabilization cooling unit begins, the online detection unit generates a batch profile vector of large seafood products. This batch profile vector includes thickness, fat percentage, glaze ratio, near-infrared feature values, conveyor line speed, and jet parameters. The predictive control unit searches the historical parameter database for similar batch templates with a nearest neighbor count of 5 and outputs the temperature level setting, residence time, temperature perturbation amplitude, and airflow pulse amplitude as the initial strategy. During execution, exploratory fine-tuning is performed within the safety boundary with a probability of 5%, with temperature level fine-tuning not exceeding ±0.5℃ and single-segment residence time fine-tuning not exceeding ±1℃. 0%, airflow pulse amplitude fine-tuning not exceeding ±5%, relative humidity fine-tuning not exceeding ±3%, single adjustment interval not less than 60s; real-time evaluation is performed using an objective function composed of dripping prediction, temperature uniformity, and unit energy consumption weighted together. If the objective function decreases, the current parameters are fixed and written into the historical parameter library to increase the weight; if it does not decrease, it is rolled back to the previous parameter set; throughout the process, CO2 concentration is constrained not to exceed 5000ppm, temperature uniformity is not greater than 2.0℃, and stable cooling is determined to be completed when the core and near-surface temperature difference is not greater than 0.8℃ and the dripping prediction is not greater than 3%.

10. A microcrystalline cryogenic resonance chamber according to claim 9, characterized in that, In the predictive control unit, the BP-ANN model employs process quantity branch and near-infrared branch input encoding. Each of the process quantity branch and near-infrared branch undergoes two layers of fully connected extraction. After branch fusion, a time-series modeling layer is introduced to capture the dynamics of the critical region. A multi-task output head is used at the top, simultaneously regressing droplet loss, hardness, recrystallization risk index, and the critical region target time. A physical constraint gating layer is connected between the fusion and output to perform key monotonic relationships and boundary projections, adding physical constraint losses to the total loss. Bypass settings include uncertainty branch output heteroscedasticity and support Monte Carlo estimation confidence intervals. Missing measurement masks and time decay interpolation layers are added on the input side. A sensitivity linearization layer is connected in parallel at the output to generate gradients for linear velocity, air volume, relative humidity, CO2 flux, ultrasonic duty cycle, and magnetic field strength for predictive control. Simultaneously, a drift correction layer and a soft limiting layer are deployed. The former performs cross-batch and cross-product type statistical alignment, while the latter maps suggested settings to the equipment safety domain.

Images