A cold storage apple state multi-parameter real-time monitoring and early warning system

Abstract

This invention discloses a real-time monitoring and early warning system for multiple parameters of apple status in cold storage, relating to the field of agricultural product storage environment monitoring technology. It includes a reverse calculation module for collecting fin temperature sequences and air duct temperature sequences. Based on the fin temperature sequences, air duct temperature sequences, and the power input sequence of the defrost heater, reverse calculation is performed to obtain the absolute humidity and absolute temperature of the hot and humid air mass. This invention, through reverse calculation, utilizes existing sensors in the cold air blower and the power input sequence of the defrost heater to calculate the absolute humidity and absolute temperature of the hot and humid air mass in the air duct before the blower restarts. This eliminates the need for a dedicated humidity sensor to perceive the state of the hot and humid air mass after defrosting, providing initial parameters for migration trajectory prediction and proactive intervention. It predicts the migration trajectory and temperature and humidity decay process of the hot and humid air mass before it contacts the apple stack unit, making the destination of the hot and humid air mass after defrosting predictable rather than unknown, providing data support for condensation risk assessment.

Description

Technical Field

[0001] This invention relates to the field of agricultural product storage environment monitoring technology, and in particular to a real-time monitoring and early warning system for multiple parameters of apples in cold storage. Background Technology

[0002] In the long-term storage management of large-scale apple cold storage facilities, to ensure fruit quality and reduce losses, the industry commonly adopts automated control technology solutions based on environmental parameter monitoring. The core principle of this technology is to deploy sensors for temperature, humidity, and gas concentration within the cold storage to form a multi-parameter sensing network. By collecting real-time macroscopic air state data within the storage, and based on preset preservation process thresholds, closed-loop control is implemented for refrigeration, humidification, and ventilation equipment to maintain overall environmental stability. The design logic of this technology is based on the assumption that the air within the storage is uniformly mixed and that macroscopic steady-state parameters can effectively characterize the microenvironment of the fruit. The spatial density and temporal sampling frequency of its sensor deployment serve macroscopic trend recording at the minute to hour level. Specifically, the evaporator fins of the air cooler will gradually frost over during long-term low-temperature operation, affecting heat exchange efficiency. Therefore, the system needs to periodically initiate a defrosting program, commonly using electric defrosting or hot gas defrosting. After the fin temperature rises and the frost melts, the fan restarts to resume the refrigeration cycle. However, in practical applications, a transient microenvironmental change occurs after the defrosting process ends, which is often overlooked. Specifically, some of the melted frost evaporates during defrosting, forming a hot and humid air mass within the air duct. The moment the defrosting process ends and the evaporator fan restarts, this remaining hot and humid air mass is not immediately expelled from the cold storage. Instead, it is propelled out of the vent by the restarted fan blades in a piston-like manner, instantly sweeping into the cold storage. Because apples in the cold storage are kept in a near-freezing temperature environment, their current skin temperature is far lower than the dew point temperature of this hot and humid air mass. When this air mass flows over the apple surface, the moisture quickly condenses, forming a thin, all-encompassing liquid water film covering the fruit's pores. Unlike potatoes and other storable agricultural products with low pore density and strong tolerance to anaerobic environments, apple pores can be instantly and completely sealed under a liquid water cover. Furthermore, apples have an extremely low threshold for accumulating anaerobic metabolic toxins, making them highly physiologically vulnerable to short-term fluctuations in the local microenvironment. Although the liquid water film only lasts for a few minutes to tens of minutes, it is sufficient to completely block the normal gas exchange of oxygen and carbon dioxide between the apple and the outside world through the stomata on its skin. This forces the entire fruit tissue into an anaerobic respiration state within a few hours, accumulating fermentation metabolic toxins such as ethanol and acetaldehyde. After this process ends, the water film evaporates, and the apple surface dries, the attacked apple enters a pseudo-normal stage, and its routine sampling indicators such as firmness, color, and sugar content do not deviate from the standard range. Due to the gap in the sampling and polling cycle, the gas monitoring system in the storage area is highly likely to miss the signal of a sudden drop in oxygen concentration in this local microenvironment. Since the above technical solution relies entirely on the collection and control of macroscopic air steady-state parameters in the warehouse, its time response capability is much slower than the complete cycle from the occurrence of the transient event to the evaporation of the water film. Its spatial perception granularity is much larger than the local stack range of the apple stack unit downstream of the specific wind channel swept by the hot and humid air mass. Therefore, it does not have the ability to perceive and predict the two key intermediate events of the spatial migration path of the hot and humid air mass after defrosting and the transient condensation on the apple skin. It is impossible to take any targeted preventive and intervention measures in this process. This issue poses a serious threat to the storage safety and supply chain reliability of apples in cold storage. Regarding storage safety, apples exposed to this hot and humid air mass accumulate irreversible fermentation and metabolic toxins internally. However, during storage, they appear normal with no abnormal macroscopic monitoring parameters. After being removed from storage and placed on a shelf at room temperature, the rising temperature triggers delayed cell membrane oxidative breakdown, leading to rapid browning of the flesh, significant flavor deterioration, and other quality degradation, resulting in severe losses across the entire warehouse. In terms of supply chain management, this time lag between storage and deterioration means that quality problems are discovered at the end-consumer stage, while the root cause lies in the storage process months prior. The phased release plan established upon warehousing becomes ineffective due to significant deviations between the actual maturity of each batch and expectations, causing supply delays and customer trust crises in downstream distribution channels. This systemically impacts the modern fresh produce supply chain system based on brand promises and quality traceability, ultimately resulting in significant economic losses and damage to brand reputation. It fails to meet the core requirements of modern agricultural cold chain logistics for precise perception and intelligent early warning throughout the storage process. Summary of the Invention

[0003] The purpose of this invention is to solve the problems of existing technologies that cannot sense the state of hot and humid air masses in the air duct after defrosting, cannot predict the spatial migration path of the hot and humid air masses, and cannot take preventive interventions before condensation forms on the apple skin. In response, this invention proposes a real-time monitoring and early warning system for the state of apples in cold storage with multiple parameters.

[0004] To achieve the above objectives, the present invention employs the following technology: a real-time monitoring and early warning system for multiple parameters of apple status in cold storage, comprising: The reverse calculation module is used to collect fin temperature sequences and air duct temperature sequences. Based on the fin temperature sequences, air duct temperature sequences and the power input sequence of the defrost heater, reverse calculation is performed to obtain the absolute humidity and absolute temperature of the hot and humid air mass. The trajectory simulation module, based on the absolute humidity and absolute temperature of the hot and humid air mass, uses a digital twin flow field model of the cold storage to simulate and obtain the spatiotemporal distribution prediction data of the hot and humid air mass. The condensation risk index calculation module is used to obtain the current skin temperature of each apple stack unit. Based on the spatiotemporal distribution prediction data of the hot and humid air mass and the current skin temperature of each apple stack unit, the condensation risk index is calculated for each apple stack unit through which the peak of the hot and humid air mass passes. The high-risk screening module identifies apple stack units that exceed the safety threshold based on the condensation risk index of each apple stack unit and a preset safety threshold. The instruction generation module generates airflow guidance instructions based on apple stacking units that exceed safety thresholds; The correction module collects temperature response data of the fan outlet area during the execution of the airflow guidance command, based on the airflow guidance command; and adjusts the airflow guidance parameters based on the temperature response data.

[0005] Furthermore, based on the fin temperature sequence, duct temperature sequence, and defrost heater power activation timing, methods for reverse calculation to obtain the absolute humidity and absolute temperature of the hot and humid air mass include: The time point when the defrost heater stops heating and the fan has not yet restarted is taken as the estimated time node. Based on the estimated time node, the last value of the air duct temperature sequence before the estimated time node is taken as the temperature of the hot and humid air mass. Based on the power input sequence of the defrost heater, the total defrost input energy is obtained by integrating over time. Based on the total defrost input energy, the heat absorbed by the fins during heating, and the preset latent heat of phase change of ice and frost melting, the energy obtained by the air trapped in the air duct is obtained by subtracting the heat absorbed by the fins during heating and the latent heat of phase change of ice and frost melting from the total defrost input energy. Obtain the geometric volume of the air duct and the air density inside the air duct before frost. Based on the geometric volume of the air duct and the air density inside the air duct before frost, multiply the geometric volume of the air duct by the air density inside the air duct before defrosting to obtain the mass of the dry air trapped inside the air duct. Based on the energy gained by the trapped air, the temperature of the hot and humid air mass, and the mass of the dry air trapped inside the air duct, solve the moisture content of the hot and humid air mass in reverse using the enthalpy equation of moist air. Based on the moisture content, perform an absolute humidity conversion to obtain the absolute humidity of the hot and humid air mass.

[0006] Furthermore, based on the absolute humidity and absolute temperature of the hot and humid air mass, methods for simulating the spatiotemporal distribution prediction data of this air mass using a digital twin flow field model of a cold storage facility include: The trajectory simulation module constructs a thermodynamic state vector based on the absolute humidity and absolute temperature of the hot and humid air mass. Based on the thermodynamic state vector, the cold storage digital twin flow field model is injected in the form of a moving mass to generate spatiotemporal distribution prediction data of hot and humid air masses.

[0007] Furthermore, based on thermodynamic state vectors, methods for generating spatiotemporal distribution prediction data of hot and humid air masses by injecting them into the digital twin flow field model of a cold storage facility in the form of moving mass clusters include: The fan restart time is obtained. Based on the fan restart time and the preset time step, with the fan restart time as the time origin, the convective diffusion process of the moving material cluster in the airflow field of the storage is solved transiently. The spatial grid coordinates of the moving material cluster at each time step, as well as the corresponding temperature decay value and absolute humidity decay value are recorded. Based on the spatial grid coordinates, temperature decay values, and absolute humidity decay values ​​at each time step, the data are merged to generate the spatiotemporal distribution prediction data of the hot and humid air mass.

[0008] Furthermore, based on the spatiotemporal distribution prediction data of the hot and humid air mass and the current skin temperature of each apple stack unit, the method for calculating the condensation risk index for each apple stack unit through which the peak of the hot and humid air mass passes includes: Based on the spatiotemporal distribution prediction data of hot and humid air masses, the local absolute humidity and local air temperature values ​​corresponding to the peak surface of the hot and humid air mass passing through each apple stack unit are extracted. Based on the local absolute humidity value, the dew point temperature value of the surface of each apple stack unit is calculated according to the thermodynamic relationship of moist air. Based on the dew point temperature value and the current skin temperature value of the corresponding apple stack unit, the difference is calculated to obtain the condensation risk index of each apple stack unit.

[0009] Furthermore, based on the condensation risk index of each apple stack unit and a preset safety threshold, the method for identifying apple stack units that exceed the safety threshold includes: Based on the condensation risk index of each apple stack unit and the preset safety threshold, the condensation risk index of each apple stack unit is compared with the preset safety threshold one by one. If the condensation risk index is greater than the safety threshold, the apple stack unit is marked as a high-risk unit. Based on high-risk units, the spatial coordinates of each high-risk unit and the expected arrival time of the corresponding hot and humid air mass peak are extracted to generate a list of apple stack units that exceed the safety threshold.

[0010] Furthermore, the method for generating airflow guidance commands based on apple stacking units exceeding safety thresholds includes: Obtain the spatial coordinates of each high-risk unit in the list of apple pallet units that exceed the safety threshold, and the estimated arrival time of the corresponding hot and humid air mass peak. Based on the spatial coordinates of each high-risk unit, the deviation direction and magnitude of the migration path of the hot and humid air mass are calculated, and a slow acceleration start-up control sequence with an initial wind deflection angle is generated. The control sequence for slow acceleration is used as the airflow guidance command.

[0011] Furthermore, based on the spatial coordinates of each high-risk unit, the method for calculating the deviation direction and magnitude of the migration path of the hot and humid air mass, and generating a slow acceleration start-up control sequence with an initial wind deflection angle, includes: Based on the spatial coordinates of each high-risk unit, calculate the offset azimuth angle and offset distance of the center position of each high-risk unit relative to the axis of the fan outlet; Based on the offset azimuth angle, the initial deflection direction of the air guide plate is determined; based on the offset distance, the ratio of the offset distance to the air supply coverage of the air cooler is calculated to determine the initial deflection angle of the air guide plate. Based on the initial deflection direction and initial deflection angle, the initial wind deflection angle is obtained by combining them; Based on the initial air deflection angle, it is combined with the preset slow acceleration start-up curve to generate a slow acceleration start-up control sequence with the initial air deflection angle.

[0012] Furthermore, the method for collecting temperature response data of the fan outlet area during the execution of airflow guidance commands includes: During the execution of the airflow guidance command, the real-time temperature values ​​of each temperature measuring point in the area surrounding the air outlet of the air cooler are continuously collected at a preset sampling frequency. Based on the real-time temperature values ​​of each temperature measuring point, the real-time temperature values ​​of each temperature measuring point are arranged according to the time axis to obtain the measured temperature change curve of the air outlet area. Based on the measured temperature change curve, the measured temperature change curve is used as temperature response data.

[0013] Furthermore, methods for adjusting airflow guidance parameters based on temperature response data include: Based on the temperature response data, the temperature response data is input into the cold storage digital twin flow field model and compared with the temperature decay value in the spatiotemporal distribution prediction data of the hot and humid air mass to calculate the prediction deviation of the migration trajectory. Based on the prediction bias, the air deflection angle parameter in the slow acceleration start-up control sequence is corrected to generate an adjusted airflow guidance command. Repeatedly collect temperature response data of the fan outlet area during the execution of airflow guidance commands and adjust airflow guidance parameters until the condensation risk index of all apple stack units is no greater than the safety threshold.

[0014] In summary, due to the adoption of the above-mentioned technology, the beneficial effects of this invention, namely the real-time monitoring and early warning system for multiple parameters of apple status in cold storage, are as follows: This invention uses a reverse calculation method to calculate the absolute humidity and absolute temperature of the hot and humid air mass in the air duct before the fan restarts, utilizing the existing sensors and power input timing of the defrost heater in the evaporator. This eliminates the need for a dedicated humidity sensor to detect the state of the hot and humid air mass after defrosting, providing initial parameters for migration trajectory prediction and proactive intervention. By injecting the absolute humidity and absolute temperature of the hot and humid air mass into the digital twin flow field model of the cold storage for transient simulation, the invention predicts the migration trajectory and temperature and humidity decay process of the hot and humid air mass before it comes into contact with the apple stack unit, making the destination of the hot and humid air mass after defrosting predictable instead of unknown, and providing data support for condensation risk assessment. This invention uses a reduced-order model to replace the full-order model for online prediction. It extracts the dominant modes using the intrinsic orthogonal decomposition method and projects the high-dimensional flow field to a low-dimensional subspace, reducing the online prediction time from hours to seconds. This meets the real-time calculation requirements within the extremely short time window between fan restart and the arrival of the hot and humid air mass at the stack. By mapping the difference between the dew point temperature and the current surface temperature to a continuous condensation risk index, the system can distinguish between different levels of minor and severe risk, providing a quantitative basis for flexible setting of safety thresholds and graded early warning. This invention improves early warning efficiency by comparing the condensation risk index of each apple stack unit with a preset safety threshold and screening out high-risk units, thus concentrating intervention resources on a few units facing the threat of condensation. By calculating the offset azimuth angle and offset distance of the high-risk units relative to the fan outlet axis, a slow acceleration start-up control sequence with a guide deflection angle is applied at the moment the fan restarts, changing the direction of movement of the hot and humid air mass at its source, and transforming uncontrollable natural diffusion into guided and controlled migration. This invention uses the spatial coordinates of other apple stack units as constraints to iteratively verify the migration path of hot and humid air masses under candidate air deflection angles in the cold storage digital twin flow field model. This avoids the risk of secondary condensation that may be caused to other stacks by simply guiding hot and humid air masses away from high-risk units. By collecting temperature response data in the air outlet area and transmitting it back to the cold storage digital twin flow field model to correct prediction deviations, a complete closed loop from prediction to execution to feedback is formed. This invention employs a hierarchical strategy for closed-loop correction. Real-time correction is completed before the air mass peak reaches a high-risk unit. After the air mass passes the first high-risk unit, the correction results are applied to subsequent units. After the air mass passes all high-risk units, the deviation correction record is used for model pre-calibration in the next round of defrosting events. This approach balances immediate response within a limited intervention window with adaptive learning during long-term operation. The current skin temperature of the apple stack unit is obtained by combining an infrared temperature sensor array with an empirical model of internal and external temperature differences, avoiding fruit damage in a non-contact manner. A mapping table is used to correct the temperature difference between the inside and outside of the stack, providing reliable data for calculating the condensation risk index. Independent mapping tables are established for different varieties and packaging specifications and stored in the system. Attached Figure Description

[0015] Figure 1 A system block diagram of the present invention is shown; Figure 2 A flowchart of the present invention is shown. Detailed Implementation

[0016] The following will describe, with reference to the accompanying drawings of the embodiments of the present invention, a real-time monitoring and early warning system for multiple parameters of apple status in a cold storage facility. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0017] To more clearly and intuitively demonstrate the practical application effects and advantages of the real-time multi-parameter monitoring and early warning system for cold storage apples according to the present invention, and to verify its feasibility and effectiveness, the present invention will be further described below with reference to embodiments. Through specific scenario simulations and data calculations, the method is explained in detail how it plays a role in actual apple condition monitoring, helping readers to better understand the technical details and practical value of the invention. The present invention will be further described below with reference to embodiments; Example

[0018] See Figure 1 - Figure 2 A real-time monitoring and early warning system for multiple parameters of apple status in cold storage, comprising: The reverse calculation module is used to collect fin temperature sequences and air duct temperature sequences. Based on the fin temperature sequence, air duct temperature sequence and the power input sequence of the defrost heater, the thermodynamic state of the hot and humid air mass in the air duct after defrosting is calculated in reverse, and the absolute humidity and absolute temperature of the hot and humid air mass are obtained. It should be noted that the methods for obtaining the absolute humidity and absolute temperature of the hot and humid air mass by reverse calculation based on the fin temperature sequence, the duct temperature sequence, and the power input sequence of the defrost heater include: The time point when the defrost heater stops heating and the fan has not yet restarted is taken as the estimated time node. Based on the estimated time node, the last value of the air duct temperature sequence before the estimated time node is taken as the temperature of the hot and humid air mass. Based on the power input sequence of the defrost heater, the total defrost input energy is obtained by integrating over time. Based on the total defrost input energy, the heat absorbed by the fins during heating, and the preset latent heat of phase change of ice and frost melting, the energy obtained by the air trapped in the air duct is obtained by subtracting the heat absorbed by the fins during heating and the latent heat of phase change of ice and frost melting from the total defrost input energy. The heat absorption of the fins during heating is obtained in the following ways: the total mass of the evaporator fins and the heat exchange tubes in close contact with them is taken as the fin structure mass, which is obtained from the factory technical parameters of the air cooler or by on-site weighing; the specific heat capacity of the fin material is taken as the material's thermal property constant, which is looked up from the standard material thermal property table according to the metal material of the evaporator fins; the difference between the initial value of the fin temperature sequence at the start of defrosting heating and the last value of the fin temperature sequence at the end of defrosting heating is taken as the fin temperature difference before and after defrosting; the product of the fin structure mass, the specific heat capacity of the fin material, and the fin temperature difference before and after defrosting is the heat absorption of the fins during heating. The preset latent heat of phase change for frost melting is obtained as follows: First, the total mass of frost on the evaporator fin surface during this defrosting cycle is estimated. This total mass is determined by three factors: the cumulative operating time of the evaporator fan from the end of the previous defrost cycle to the start of this defrost cycle, the difference between the average absolute humidity of the air inside the evaporator and the saturated absolute humidity of the evaporator fin surface, and the frost rate coefficient calibrated experimentally. The total mass of frost is then multiplied by the latent heat of melting constant of water; the product is the latent heat of phase change for frost melting. The latent heat of melting constant of water is taken as the latent heat of phase change of ice melting into water at the same temperature under standard atmospheric pressure at zero degrees Celsius. The method involves obtaining the geometric volume of the air duct and the air density inside the air duct before frost. Based on these measurements, the geometric volume of the air duct is multiplied by the air density before defrosting to obtain the mass of the dry air trapped inside the air duct. Then, based on the energy gained by the trapped air, the temperature of the hot and humid air mass, and the mass of the dry air trapped inside the air duct, the moisture content of the hot and humid air mass is calculated using the enthalpy equation for moist air. Based on this moisture content, the absolute humidity is converted to obtain the absolute humidity of the hot and humid air mass. Finally, during the critical window period when defrosting heating stops but the fan has not yet restarted, the absolute humidity and absolute temperature of the hot and humid air mass inside the air duct are calculated using the power input sequence of the defrosting heater recorded by the existing fin temperature sensor, air duct temperature sensor, and controller. This method solves the problem that existing technologies rely entirely on sampling macroscopic steady-state air parameters within the storage facility and cannot perceive the state of the hot and humid air mass after defrosting. Existing technologies lack direct measurement methods for this hot and humid air mass, leaving them completely unaware of its existence, temperature, and humidity levels, and can only passively wait for condensation to occur before taking remedial action. This step, through reverse calculation, achieves for the first time quantitative perception of the thermodynamic state of this invisible hazard source—the defrosting hot and humid air mass—without adding any dedicated humidity sensors. This provides accurate initial state parameters of the hot and humid air mass for subsequent migration trajectory prediction and proactive intervention, filling the technical blind spot of existing technologies that are completely unaware of the state of the hot and humid air mass in the air duct after defrosting. In the enthalpy equation for moist air, the specific enthalpy of moist air is expressed as the sum of the specific enthalpy of dry air and the specific enthalpy of water vapor. The specific enthalpy of dry air is the product of the isobaric specific heat capacity of dry air and the air temperature, while the specific enthalpy of water vapor is the sum of the products of the latent heat of vaporization of water vapor, the isobaric specific heat capacity of water vapor, and the air temperature. The dew point temperature is obtained by referring to a table of thermal properties of moist air or by converting the absolute humidity value using an empirical correlation formula for dew point temperature.

[0019] Specifically, the empirical correlation for dew point temperature is an approximate conversion formula between local air absolute humidity and dew point temperature under normal pressure conditions. Its original form originates from the empirical relationship between saturated water vapor partial pressure and temperature in moist air thermodynamics. This solution pre-installs this empirical correlation in the system's calculation module. When online calculations need to avoid the time-consuming lookup of moist air thermal property tables, or when local air state parameters exceed the preset range of the moist air thermal property tables, the empirical correlation is invoked to directly calculate the dew point temperature value from the local air absolute humidity. This empirical correlation is applicable to normal pressure cold storage environments, and the calculation error is within an acceptable engineering range.

[0020] Specifically, the moist air thermal property table is a standard reference data table in the field of engineering thermodynamics, which records the correspondence between thermodynamic parameters such as density, specific enthalpy, moisture content, and dew point temperature of moist air under different temperatures and pressures. The original data of the moist air thermal property table comes from the internationally recognized standard for water vapor properties and has been widely used in HVAC, refrigeration engineering, and meteorology. In this solution, the moist air thermal property table is pre-stored in the system storage unit in the form of a two-dimensional lookup table. Using the local absolute humidity and local air temperature values ​​as index variables, the corresponding dew point temperature value is directly obtained by looking up the table. For intermediate values ​​not directly included in the table, a bilinear interpolation method is used for smoothing. This lookup method avoids the iterative solution process when calculating the dew point temperature online, further shortening the calculation time of the condensation risk assessment stage.

[0021] Specifically, the initial source of this enthalpy equation for moist air is the standard relational formula for calculating the thermodynamic properties of moist air in engineering thermodynamics. Its original form expresses the specific enthalpy of moist air as equal to the specific enthalpy of dry air plus the product of moisture content and the specific enthalpy of water vapor in the moist air. This application directly adopts this standard relational formula without formal modification, but in its application scenario, it is used to solve for the enthalpy of moist air trapped in the duct after defrosting. In conventional applications, this equation is used to calculate the enthalpy value when the moisture content and temperature are known. This application uses it in reverse, that is, to solve for the moisture content under the condition of known enthalpy change and temperature. The feasibility of this reverse application method is based on the critical window period when defrosting heating stops and the fan has not yet restarted. The trapped air in the duct is in a near-static closed state, and its enthalpy change can be accurately determined by subtracting the heat absorbed by the fins and the latent heat of phase change from the frost melting from the total defrosting input energy, thus locking the only unknown in the equation to the moisture content. The dimensions of both sides of the enthalpy equation for moist air are energy per unit mass, i.e., kilojoules per kilogram. The left side of the equation represents the specific enthalpy of moist air, with dimensions of kilojoules per kilogram. In the dry air specific enthalpy term on the right side of the equation, the dimension of the isobaric specific heat capacity of dry air is kilojoules per kilogram per degree Celsius, and its product with the air temperature is also kilojoules per kilogram. In the water vapor specific enthalpy term, the dimension of the latent heat of vaporization of water vapor is kilojoules per kilogram, and the dimension of the isobaric specific heat capacity of water vapor is kilojoules per kilogram per degree Celsius. Their product with the air temperature is also kilojoules per kilogram, and their dimensionless moisture content remains kilojoules per kilogram. The dimensions of both sides of the equation are strictly consistent, conforming to the fundamental dimensional principles of thermodynamics. In the reverse calculation, the dimension of the total energy input for defrosting is kilojoules, the dimension of the heat absorbed by the fins during heating is kilojoules, and the dimension of the latent heat of phase change from frost melting is kilojoules. Subtracting these three, the dimension of the energy gained by the air trapped in the duct is still kilojoules, consistent with the dimension of the enthalpy on the left side of the equation. The dimension of the mass of the dry air trapped in the duct is kilograms. Dividing by this dimension yields the enthalpy per unit mass of air, and the dimension conversion is correct.

[0022] Furthermore, regarding the acquisition of the mass of dry air retained in the duct, considering that the thermal expansion of air in the duct during defrosting heating will cause some air to escape from the duct, a duct sealing correction factor less than one can be applied by multiplying the duct's geometric volume by the air density in the duct before defrosting. This duct sealing correction factor is pre-calibrated according to the sealing level of the duct structure; the higher the sealing level, the closer the factor is to one. For the unmodeled heat loss after deducting the heat absorbed by the fins during heating and the latent heat of the frost melting phase change from the total defrosting input energy, a heat dissipation compensation term can be introduced for correction. This heat dissipation compensation term is estimated based on the defrosting heating time and the temperature difference between the inside and outside of the duct. The introduction of the duct sealing correction factor solves the problem that the thermal expansion of air in the duct during defrosting heating causes some air to escape, resulting in the actual mass of dry air retained in the duct being less than the theoretical value obtained by multiplying the duct's geometric volume by the air density in the duct before defrosting. In existing technologies, ignoring this thermal expansion overflow effect will systematically overestimate the mass of retained dry air, leading to an underestimation of the moisture content calculated in reverse. The system deviation is compensated by a pre-calibrated duct sealing correction coefficient, which significantly improves the accuracy of back-calculation. The introduction of a heat dissipation compensation term solves the problem of unmodeled heat dissipation loss after deducting the heat absorbed by the fins and the latent heat of the frost melting phase change from the total defrosting input energy. By estimating and compensating for heat dissipation loss based on the defrosting heating time and the temperature difference inside and outside the duct, the solution error of back-calculation is further reduced. The specific estimation method for the heat dissipation compensation item is as follows: the heat dissipation compensation item equals the heat dissipation coefficient of the outer wall of the duct multiplied by the defrosting heating duration, and then multiplied by the temperature difference between the inside and outside of the duct. The heat dissipation coefficient of the outer wall of the duct is pre-calculated using the ratio of the surface area of ​​the outer wall of the duct, the thermal conductivity of the insulation material of the duct wall, and the thickness of the insulation layer. This heat dissipation coefficient is a constant under the condition that the duct structure remains unchanged. The defrosting heating duration is the length of time from the start-up time to the stop time of the defrosting heater power input sequence. The temperature difference between the inside and outside of the duct is the difference between the average air temperature inside the duct and the air temperature inside the warehouse during the defrosting heating period. The average air temperature inside the duct is taken as the arithmetic mean of all sampled values ​​of the duct temperature sequence during the defrosting heating period.

[0023] Specifically, the duct sealing correction factor is pre-calibrated using the following method: After the cold storage is put into operation for the first time or after the duct structure is changed, select a typical defrosting cycle. At the point when defrosting heating has ended but the fan has not yet restarted, use a portable humidity meter to measure the absolute humidity of the hot and humid air mass at the duct access port as the baseline value. Compare this baseline value with the absolute humidity calculated in reverse without using the duct sealing correction factor. The ratio of the two values ​​is the calibration value of the duct sealing correction factor. For evaporative coolers that cannot provide an access port, the duct sealing level reference value provided by the manufacturer at the factory can be selected. Determine the initial coefficient based on the comparison table of sealing level and correction factor, and adjust it in subsequent operation based on statistical feedback of the actual condensation in the cold storage after each defrost. The duct sealing rating reference value is a rating indicator specified by the evaporative cooler manufacturer based on the sealing design of the duct structure at the time of equipment shipment. This rating indicator divides the duct sealing performance into several levels, with higher rating values ​​indicating a denser duct structure and lower air leakage rate. The manufacturer measures the air leakage rate of the duct under standard pressure differential conditions using a duct air tightness test bench, determines the corresponding duct sealing rating reference value based on the range of leakage rates, and notes this value on the equipment nameplate or technical manual. The table comparing the duct sealing rating reference value with the duct sealing performance correction factor is provided by the manufacturer with the equipment documentation, or compiled by the system provider based on field calibration data from multiple units of the same model.

[0024] The absolute humidity and absolute temperature of the hot and humid air mass are used as initial state parameters and injected into the digital twin flow field model of the cold storage. The spatial migration trajectory of the hot and humid air mass is simulated with the restart time of the fan as the time origin, so as to obtain the spatiotemporal distribution prediction data of the hot and humid air mass. It should be noted that the methods for obtaining the spatiotemporal distribution prediction data of the hot and humid air mass by simulating it using a digital twin flow field model of a cold storage facility, based on the absolute humidity and absolute temperature of the hot and humid air mass, include: Thermodynamic state vectors are constructed based on the absolute humidity and absolute temperature of the hot and humid air mass. Based on the thermodynamic state vector, the cold storage digital twin flow field model is injected in the form of a moving mass to generate spatiotemporal distribution prediction data of hot and humid air masses.

[0025] Specifically, methods for generating spatiotemporal distribution prediction data of hot and humid air masses by injecting them into the digital twin flow field model of a cold storage facility in the form of moving mass vectors, based on thermodynamic state vectors, include: The fan restart time is obtained. Based on the fan restart time and the preset time step, with the fan restart time as the time origin, the convective diffusion process of the moving material cluster in the airflow field of the storage is solved transiently. The spatial grid coordinates of the moving material cluster at each time step, as well as the corresponding temperature decay value and absolute humidity decay value are recorded. Based on the spatial grid coordinates, temperature decay value, and absolute humidity decay value of each time step, the data are merged to generate the spatiotemporal distribution prediction data of the hot and humid air mass. By constructing a thermodynamic state vector from the absolute humidity and absolute temperature of the hot and humid air mass obtained through reverse calculation, and injecting it into the cold storage digital twin flow field model in the form of a moving mass, and performing transient iterative solutions with the fan restart time as the time origin, this solves the problem of existing technologies being unable to predict the spatial migration path of hot and humid air masses after defrosting. Existing technologies only perform fixed fan delays or low-speed operation after defrosting, completely unaware of where the hot and humid air mass will be pushed or which apple stack units it will pass through, thus preventing any targeted preventative intervention. This step, through transient simulation of the cold storage digital twin flow field model, accurately predicts the complete migration trajectory and temperature and humidity decay process along the way of the hot and humid air mass before it even comes into contact with any apple stack units. This transforms the destination of the hot and humid air mass after defrosting from an unknown blind box state to a predictable transparent state, providing predictive data support for subsequent condensation risk assessment and proactive intervention decisions.

[0026] It should be noted that the method for constructing the digital twin flow field model of the cold storage is as follows: First, the three-dimensional geometric dimensions of the cold storage, the stacking layout of the apple stack units, the location and size of the air cooler outlets, and the location and size of the return air inlets are obtained to establish a geometric model of the internal space of the cold storage. Then, the geometric model is spatially meshed, and the mesh is refined near the air cooler outlets, return air inlets, and stack surfaces. Next, boundary conditions are set, including the air velocity direction of the air cooler outlets, the pressure boundary of the return air inlets, and the thermal boundary of the cold storage enclosure structure. Finally, a turbulence model based on the Reynolds-averaged equations is used to solve the steady-state flow field inside the cold storage to obtain a digital twin flow field model of the cold storage. The Reynolds-averaged equations are fundamental governing equations in computational fluid dynamics that decompose instantaneous turbulent motion into time-averaged and fluctuating motions. The core idea is to decompose the physical quantities in turbulent flow into the sum of time-averaged and fluctuating values, substitute them into the fundamental fluid dynamics equations, and then take the time average to obtain a closed set of equations concerning the time-averaged quantities. Since the airflow inside a cold storage facility is low-speed incompressible turbulence, using the Reynolds-averaged equations to describe its flow field can significantly reduce the computational load while ensuring engineering accuracy, making it one of the most commonly used methods in industrial flow field simulation. This scheme selects a two-equation turbulence model based on the eddy viscosity assumption. The Reynolds stress term is closed by solving the transport equations for turbulent kinetic energy and its dissipation rate, thereby completing the numerical solution of the steady-state flow field inside the cold storage facility. To meet the real-time calculation requirements within the extremely short time window between the fan restart and the arrival of the hot and humid air mass at the nearest apple stack unit, a reduced-order model is further constructed based on the aforementioned cold storage digital twin flow field model to replace the cold storage digital twin flow field model for online transient prediction. The reduced-order model is constructed by performing offline transient simulations of the cold storage digital twin flow field model under different combinations of air supply parameters and air guide angles, generating a large amount of sample data on the migration trajectory of the hot and humid air mass. The dominant modes of the hot and humid air mass migration process are extracted from the sample data using the intrinsic orthogonal decomposition method. The cold storage digital twin flow field model is then projected onto a low-dimensional subspace spanned by a small number of dominant modes to obtain the reduced-order model. During online prediction, the reduced-order model can complete the simulation of the hot and humid air mass migration trajectory within seconds, meeting the real-time requirements of intervention decisions. When the stacking state of goods in the warehouse changes significantly, the stack occupancy model is updated using a handheld scanning device, re-triggering the offline calculation of the cold storage digital twin flow field model and the update of the reduced-order model to maintain prediction accuracy. The construction of the reduced-order model addresses the issue that the transient simulation computation of the full-order computational fluid dynamics model is too cumbersome to meet the real-time prediction requirements within the extremely short time window between fan restart and the arrival of the hot and humid air mass at the stack. By extracting the dominant modes from the offline full-order simulation samples using the intrinsic orthogonal decomposition method, the high-dimensional flow field is projected onto a low-dimensional subspace, reducing the online prediction time from hours to seconds. This overcomes the core engineering obstacle of transitioning digital twin technology for cold storage flow field models from offline design tools to online real-time control. Furthermore, when the stacking state of goods in the cold storage changes significantly, the stack occupancy model is updated using a handheld scanning device, triggering offline calculations and the reduced-order model update again. This solves the problem of inaccurate flow field predictions caused by changes in stack layout, ensuring the model's continued effectiveness throughout the entire storage cycle.

[0027] Specifically, the reduced-order model must undergo accuracy verification after each update before it can be used online. The accuracy verification method is as follows: A subset of samples from the offline transient simulation data that did not participate in the extraction of the dominant mode was randomly selected as a validation set. The input parameters of each sample in the validation set were substituted into the reduced-order model and the full-order cold storage digital twin flow field model for solving. The average deviation of the spatial trajectory of the hot and humid air mass migration path and the maximum deviation of the hot and humid air mass peak surface at the time of arrival at each apple stack unit were compared between the two models. If both the average deviation and the maximum deviation were less than the preset accuracy deviation tolerance, the reduced-order model was deemed to have passed the accuracy verification. If either deviation exceeded the accuracy deviation tolerance, the number of dominant modes retained was increased or the sample size of the offline transient simulation was expanded, and the reduced-order model was reconstructed until it passed the accuracy verification.

[0028] Specifically, the tolerance value for accuracy deviation is determined jointly based on the geometric dimensions of the apple stack unit and the safe avoidance distance between the hot and humid air mass and the surface of the apple stack unit. If the average deviation of the migration path of the hot and humid air mass and the maximum deviation of the peak arrival time do not exceed half the length of a single apple stack unit in the airflow direction, then the prediction accuracy of the reduced-order model is considered sufficient to distinguish whether the hot and humid air mass contacts the surface of the apple stack unit. This deviation magnitude is the basis for setting the tolerance value for accuracy deviation. The tolerance value for accuracy deviation is preset in the system parameter configuration module and is calibrated and confirmed once during the initial deployment based on the actual apple stack unit dimensions of the target cold storage.

[0029] Specifically, the selection criteria for the dominant mode are as follows: The modes are arranged in descending order of their corresponding singular values. The proportion of the sum of squared singular values ​​of the top few modes to the sum of squared singular values ​​of all modes is calculated; this proportion represents the cumulative energy percentage of the top few modes. The top few modes corresponding to the first time their cumulative energy percentage exceeds a preset energy threshold are selected as the dominant modes. The preset energy threshold is set to a value close to one based on engineering accuracy requirements. The number of dominant modes that meet this preset energy threshold is the order of the reduced-order model. This selection criterion ensures that the reduced-order model achieves the maximum computational speed improvement while losing very little flow field detail information.

[0030] Specifically, the stacking space model is a data structure in the cold storage digital twin flow field model used to describe the geometric spacement information of each apple stack unit in the three-dimensional space within the cold storage. The stacking space model records the length, width, and height dimensions of each apple stack unit, the spatial coordinates of its bottom center point, the number of stacking layers, and the distance to adjacent apple stack units. This data structure was initially created during the initial modeling of the cold storage using a combination of manual measurement and 3D scanning. During the storage period, when the stacking layout of local apple stack units changes due to operations such as outbound sampling, stacking inspection, and replenishment, operators use handheld scanning devices to scan the changed areas, obtaining the actual spacement parameters of each apple stack unit after the change, and updating the stacking space model accordingly. The updated stacking space model is then input into the cold storage digital twin flow field model, replacing the original geometric boundary conditions to ensure that the spatial layout on which the flow field simulation is based remains consistent with the actual cold storage.

[0031] Specifically, intrinsic orthogonal decomposition (IVD) is a data-driven reduction technique that extracts the most energy-representative low-dimensional basis functions from high-dimensional datasets. Its basic principle is as follows: A large number of hot and humid air mass migration trajectory sample data generated by offline transient simulations are arranged into a snapshot matrix according to time series. Singular value decomposition (SVD) is then performed on this snapshot matrix to obtain a series of mutually orthogonal spatial modes and their corresponding singular values. The magnitude of the singular value represents the energy proportion of the corresponding mode in all sample data. By retaining the first few dominant modes with the largest singular values, the degrees of freedom of the original high-dimensional system can be significantly reduced while preserving most of the flow field energy. During online prediction, the full-order cold storage digital twin flow field model is projected onto the low-dimensional subspace spanned by these dominant modes, transforming the original partial differential equations into a low-dimensional ordinary differential equation system, reducing the solution time from hours to seconds. IVD has mature applications in the field of fluid dynamics reduction modeling, and its mathematical properties guarantee the prediction accuracy of the reduced model within the sample coverage range.

[0032] The condensation risk index calculation module is used to obtain the current skin temperature of each apple stack unit. Based on the spatiotemporal distribution prediction data of the hot and humid air mass and the current skin temperature of each apple stack unit, the condensation risk index is calculated for each apple stack unit through which the peak of the hot and humid air mass passes. The current skin temperature of each apple stack unit is obtained by deploying an array of infrared temperature sensors facing the surface of each stack unit within the cold storage. This non-contact measurement method of the infrared temperature sensor array solves the problem of contact temperature measurement damaging the apple skin and compromising the marketability of the fruit. Existing technologies using thermocouples or thermistors attached to the apple surface for contact temperature measurement can cause mechanical damage to the fruit skin, creating an entry point for microbial infection and increasing storage losses. This solution uses infrared non-contact temperature measurement to obtain the current skin temperature data of the outermost apple in the stack unit without damaging the fruit, ensuring the integrity and safety of the apples during ultra-long-term storage. For the inner apples hidden by the outer apples, their current skin temperature is estimated by subtracting the compensated temperature difference determined by an empirical model of the temperature difference between the inside and outside of the stack unit from the current skin temperature of the apple stack unit. The empirical model for internal and external temperature difference is pre-calibrated based on the size of the apple stack unit, the stacking density, and the air velocity inside the warehouse. It takes the difference between the current skin temperature of the outermost layer of the apple stack unit and the air temperature inside the warehouse as input and outputs the temperature offset of the inner layer of apples relative to the outermost layer of apples. Specifically, the compensated temperature difference is the output value obtained by the online application of the internal and external temperature difference empirical model, based on the difference between the current outermost skin temperature of the apple stack unit and the air temperature inside the warehouse, by looking up or interpolating from a pre-established mapping table. When looking up the table, if the real-time input difference exactly equals a record value of a certain set of input variables in the mapping table, the temperature offset corresponding to that record value is directly taken as the compensated temperature difference. If the real-time input difference lies between two adjacent input variable records in the mapping table, the compensated temperature difference is calculated by linearly interpolating these two records and their corresponding temperature offsets according to the proportional position of the input difference. When using the compensated temperature difference to estimate the current skin temperature of the inner apples, the current skin temperature of the outermost layer of the apple stack unit is subtracted from the compensated temperature difference; the difference is the estimated current skin temperature of the inner apples.

[0033] Specifically, the calibration method for this empirical model of internal and external temperature difference is as follows: A representative apple stack unit from the target cold storage was selected. Miniature temperature recorders were deployed on the outer surface of the apples and the center of the inner apples within each stack unit. During normal refrigeration cycles, data on the current outer and inner skin temperatures, as well as the airflow velocity within the storage area, were simultaneously collected, continuously for at least one complete refrigeration cycle. Based on the collected data, a mapping table was established, with the difference between the current outermost skin temperature and the internal air temperature of the apple stack unit as the input variable, and the temperature offset of the inner apples relative to the outermost apples as the output variable. Separate mapping tables were established for different varieties and packaging specifications of apples and stored categorized within the system. When the stored variety or packaging specifications change, the system automatically retrieves the corresponding mapping table without requiring recalibration. By combining an infrared temperature sensor array with an empirical model of internal and external temperature differences, this approach solves the problem of existing technologies' difficulty in obtaining the current skin temperature of apples inside an apple stack unit on a large scale, without damage, and accurately. Infrared temperature sensors can only measure the surface temperature of the outermost layer of apples in an apple stack unit. However, due to poor air circulation and heat accumulation during respiration, the skin temperature of apples inside the stack is usually higher than that of the outer layer. Directly using the current skin temperature of the outer layer to represent the current skin temperature of the entire stack of apples leads to an underestimation of the risk of condensation. The empirical model of internal and external temperature differences, through a pre-calibrated mapping relationship, calculates the temperature offset of the inner layer of apples based on the difference between the current skin temperature of the outer layer and the air temperature inside the storage room. This provides a more accurate estimate of the current skin temperature of the entire stack of apples, providing a reliable data foundation for the accurate calculation of the subsequent condensation risk index. Furthermore, separate mapping relationship tables are established for different varieties and packaging specifications and stored in the system, solving the problem of insufficient generalization ability of the empirical model when apple varieties or packaging methods change. This ensures that the system maintains high temperature estimation accuracy under various storage scenarios.

[0034] It should be noted that the method for calculating the condensation risk index for each apple stack unit through which the hot and humid air mass peak passes, based on the spatiotemporal distribution prediction data and the current skin temperature of each apple stack unit, includes: Based on the spatiotemporal distribution prediction data of hot and humid air masses, the local absolute humidity and local air temperature values ​​corresponding to the peak surface of the hot and humid air mass passing through each apple stack unit are extracted. Based on the local absolute humidity value, the dew point temperature value of the surface of each apple stack unit is calculated according to the thermodynamic relationship of moist air. Based on the dew point temperature value and the current skin temperature value of the corresponding apple stack unit, the difference is calculated to obtain the condensation risk index of each apple stack unit. The condensation risk index is defined as a continuous dimensionless value obtained through a preset risk mapping relationship based on the difference between the dew point temperature and the current skin temperature. When the dew point temperature is less than or equal to the current skin temperature, the condensation risk index is zero; as the difference increases, the condensation risk index monotonically increases; when the difference reaches or exceeds a preset critical temperature difference constant, the condensation risk index is one. By mapping the difference between the dew point temperature and the current skin temperature to a continuous condensation risk index, the problem of existing technologies being able to only determine whether condensation has occurred but not to quantify the degree of condensation risk is solved. Condensation is not a binary event that occurs instantaneously at a precise temperature difference. Influenced by factors such as the cleanliness of the apple skin, localized minor scratches, and surface micro-roughness, the actual probability of condensation varies among different individual apples at the same temperature difference. This scheme uses a monotonically increasing risk mapping relationship to generate continuous values, enabling the early warning system to distinguish between different levels of minor and severe risks. This provides a refined quantitative basis for the flexible setting of subsequent safety thresholds and graded early warnings, avoiding the dilemma of setting the threshold too low in traditional binary judgment, which leads to frequent false alarms, and setting the threshold too high, which leads to missed alarms.

[0035] The high-risk screening module compares the condensation risk index of each apple stack unit with the preset safety threshold to identify apple stack units that exceed the safety threshold. The preset safety thresholds are pre-set based on the apple variety's tolerance to dew formation on the skin. For late-maturing varieties with thicker skin and denser cuticles, a higher safety threshold is set, allowing for lower warning sensitivity to reduce unnecessary intervention. For early-maturing varieties with thinner skin and higher stomatal density, a lower safety threshold is set, increasing warning sensitivity to ensure no dew formation risk is overlooked. The specific values ​​of the safety thresholds are determined through postharvest physiological experiments. The experimental method involves using apple samples of different varieties in a temperature and humidity controlled chamber, under simulated frost-induced hot and humid air mass invasion, recording the critical temperature difference between the dew point temperature and the skin temperature at which a liquid water film appears on the apple skin, triggering the accumulation of anaerobic respiration metabolic toxins. The dew formation risk index corresponding to this critical temperature difference is then adjusted downwards by a certain safety margin as the preset safety threshold. The safety thresholds for each variety are pre-stored in the system storage unit and can be selected and retrieved by the operator after the storage variety is determined. It should be noted that, based on the condensation risk index of each apple pallet unit and the preset safety threshold, the methods for identifying apple pallet units that exceed the safety threshold include: Based on the condensation risk index of each apple stack unit and the preset safety threshold, the condensation risk index of each apple stack unit is compared with the preset safety threshold one by one. If the condensation risk index is greater than the safety threshold, the apple stack unit is marked as a high-risk unit. Based on high-risk units, the spatial coordinates and the expected arrival time of the corresponding hot and humid air mass peaks of each high-risk unit are extracted to generate a list of apple stack units exceeding the safety threshold. By comparing the condensation risk index of each apple stack unit with the preset safety threshold one by one, and marking units exceeding the threshold as high-risk units and extracting their spatial coordinates and the expected arrival time of the hot and humid air mass peaks, this solves the problem of existing technologies being unable to quickly locate a few dangerous areas requiring key attention when dealing with hundreds of apple stack units in a cold storage facility. In large-scale automated cold storage facilities, the number of apple stack units is enormous. Applying the same level of intervention measures to all units would not only overburden the implementing agency but also potentially cause unnecessary environmental disturbances to a large number of safe stacks that do not require intervention. This step, through a threshold screening mechanism, concentrates the system's attention and intervention resources on a few high-risk units that are truly facing the threat of condensation, achieving an upgrade from indiscriminate monitoring to precise targeted early warning, significantly improving early warning efficiency and the pertinence of intervention.

[0036] The instruction generation module generates airflow guidance instructions based on the apple stack unit exceeding the safety threshold, in order to change the migration path of the hot and humid air mass. It should be noted that the methods for generating airflow guidance commands based on apple pallet units exceeding safety thresholds include: Obtain the spatial coordinates of each high-risk unit in the list of apple pallet units that exceed the safety threshold, and the estimated arrival time of the corresponding hot and humid air mass peak. Based on the spatial coordinates of each high-risk unit, the deviation direction and magnitude of the migration path of the hot and humid air mass are calculated, and a slow acceleration start-up control sequence with an initial wind deflection angle is generated. The control sequence for slow acceleration is used as the airflow guidance command.

[0037] Specifically, the method for calculating the deviation direction and magnitude of the migration path of hot and humid air masses based on the spatial coordinates of each high-risk unit, and generating a slow acceleration start-up control sequence with an initial wind deflection angle, includes: Based on the spatial coordinates of each high-risk unit, calculate the offset azimuth angle and offset distance of the center position of each high-risk unit relative to the axis of the fan outlet; Based on the offset azimuth angle, the initial deflection direction of the air guide plate is determined; based on the offset distance, the ratio of the offset distance to the air supply coverage of the air cooler is calculated to determine the initial deflection angle of the air guide plate. Based on the initial deflection direction and initial deflection angle, the initial wind deflection angle is obtained by combining them; Based on the initial air guide angle, it is combined with the preset slow acceleration start curve to generate a slow acceleration start control sequence with the initial air guide angle. The preset slow-acceleration start-up curve is a standard control curve of fan speed versus time, pre-stored in the system control unit. This curve is characterized by: in the initial stage of fan start-up, the speed increase rate is slow and the acceleration is small, allowing the hot and humid air mass to be smoothly pushed out of the duct at a low speed, providing sufficient guidance time for the air guide plate in the initial section of the air mass migration path; in the middle and later stages of start-up, the speed increase rate gradually increases, and the acceleration increases accordingly, allowing the fan to quickly recover to the air delivery speed required for normal refrigeration cycle after the peak of the hot and humid air mass has entered the predetermined migration direction. The specific shape of this curve is determined through offline simulation results of the cold storage digital twin flow field model under different combinations of slow-acceleration parameters, with the selection criterion being the optimal compromise between the guiding effect of the hot and humid air mass and the refrigeration recovery speed. The preset slow-acceleration start-up curve is loaded during system initialization and recalibrated along with the update of the reduced-order model when the stacking state of goods in the cold storage changes significantly. By calculating the deviation direction and magnitude of the migration path of hot and humid air masses based on the spatial coordinates of high-risk units, a slow-acceleration start-up control sequence with an initial wind deflection angle is generated. This solves the problem that existing technologies can only passively wait for the natural migration of hot and humid air masses after defrosting, without being able to actively intervene in their path. The fixed-delay start-up or low-speed operation strategies of existing technologies cannot change the migration direction of hot and humid air masses, which may still be pushed onto the surface of the apple stack, causing condensation. This step, by applying a pre-calculated wind deflection angle and slow acceleration curve at the moment the fan restarts, changes the movement direction of the hot and humid air masses at their source, causing them to deviate from the high-risk apple stack units. This transforms the uncontrollable natural diffusion of defrosting hot and humid air masses into guided and controlled migration, realizing a paradigm shift from post-event remediation to pre-event proactive avoidance.

[0038] Furthermore, when generating airflow guidance commands, in addition to deviating the migration path of hot and humid air masses from high-risk units, it is also necessary to ensure that the guided migration path does not pass through the surfaces of other apple stack units. The specific implementation method is as follows: When calculating the direction and magnitude of deviation, the spatial coordinates of other apple stack units outside the list of apple stack units that exceed the safety threshold are used as constraints. The migration path of the hot and humid air mass under the candidate wind deflection angle is iteratively verified in the cold storage digital twin flow field model. The wind deflection angle that neither passes through the high-risk unit nor the surface of other apple stack units is selected as the final parameter. If a wind deflection angle that satisfies all constraints cannot be found, the wind deflection angle with the fewest apple stack units on the path will be selected first, and an early warning signal will be triggered to notify the operators to pay attention to the secondary risk area. Using the spatial coordinates of apple stack units outside the list of apple stack units exceeding the safety threshold as constraints, the migration path of the hot and humid air mass under candidate wind deflection angles is iteratively verified. A wind deflection angle that neither passes through high-risk units nor the surfaces of other apple stack units is selected, solving the problem that simply guiding the hot and humid air mass away from high-risk units may cause secondary condensation risks to other originally safe stacks. When a wind deflection angle satisfying all constraints cannot be found, the scheme that passes through the fewest apple stack units on the path is prioritized and an early warning signal is triggered, resolving the decision-priority issue between complete avoidance and partial avoidance.

[0039] The correction module, based on the airflow guidance command, collects temperature response data of the fan outlet area during the execution of the airflow guidance command; and adjusts the airflow guidance parameters based on the temperature response data. It should be noted that the methods for collecting temperature response data of the fan outlet area during the execution of airflow guidance commands include: During the execution of the airflow guidance command, the real-time temperature values ​​of each temperature measuring point in the area surrounding the air outlet of the air cooler are continuously collected at a preset sampling frequency. Each temperature measurement point refers to an array of temperature sensing elements deployed in a pre-defined spatial layout around the air outlet of the evaporative cooler. This array is deployed with one temperature sensing element at each of the four edges (top, bottom, left, right) of the air outlet, and several more at different distances from the outlet's centerline, creating multi-point synchronous sampling of the temperature field in the outlet area. Each temperature sensing element can reuse an existing temperature sensor in the air outlet area or can be deployed independently as a thermocouple or thermistor. The spatial coordinates of each temperature measurement point are pre-entered into the cold storage's digital twin flow field model. This allows for point-to-point comparison between the measured temperature values ​​at each measurement point and the corresponding temperature decay values ​​in the model after the temperature response data is transmitted back. The preset sampling frequency is continuous sampling at least once per second to meet the real-time capture requirement of temperature changes during the migration of hot and humid air masses. Based on the real-time temperature values ​​of each temperature measuring point, the real-time temperature values ​​of each temperature measuring point are arranged according to the time axis to obtain the measured temperature change curve of the air outlet area. Based on the measured temperature change curve, the measured temperature change curve is used as temperature response data.

[0040] It should be noted that the methods for adjusting airflow guidance parameters based on temperature response data include: Based on the temperature response data, the temperature response data is input into the cold storage digital twin flow field model and compared with the temperature decay value in the spatiotemporal distribution prediction data of the hot and humid air mass to calculate the prediction deviation of the migration trajectory. Based on the prediction bias, the air deflection angle parameter in the slow acceleration start-up control sequence is corrected to generate an adjusted airflow guidance command. Repeatedly collect temperature response data of the fan outlet area during the execution of airflow guidance commands and adjust airflow guidance parameters until the condensation risk index of all apple stack units is no greater than the safety threshold.

[0041] Furthermore, since the time window for the hot and humid air mass to migrate from the air outlet to the apple stack unit is extremely limited, the air deflection angle parameter in the above-mentioned modified slow-acceleration start-up control sequence is implemented according to the following stratified strategy in actual operation: For a currently migrating hot and humid air mass, if the calculation of the prediction deviation and the correction of the air guide angle parameters are completed before the peak of the hot and humid air mass reaches the high-risk unit, the adjusted airflow guidance command is immediately executed to change the subsequent migration direction of the hot and humid air mass; by collecting the temperature response data of the fan outlet area during the execution of the airflow guidance command and transmitting it back to the cold storage digital twin flow field model to correct the prediction deviation, the problem of error between the predicted trajectory and the actual trajectory is solved. If the peak of the hot and humid air mass has already crossed the first high-risk unit and there are still other high-risk units on the migration path of the hot and humid air mass, then the corrected wind deflection angle parameter will be applied to protect the subsequent high-risk units. If the peak of the hot and humid air mass has crossed all high-risk units, the prediction deviation obtained in this calculation is stored as a deviation correction record. In the reverse calculation stage of the next defrosting event, this deviation correction record is called to pre-correct the initial parameters of the cold storage digital twin flow field model, so that the prediction accuracy of the cold storage digital twin flow field model in subsequent events continues to improve. Through the above-mentioned hierarchical strategy, the contradiction between the extremely fast migration speed of the hot and humid air mass and the inherent delay in the feedback correction chain is resolved. Before the peak of the hot and humid air mass reaches the first high-risk unit, the limited time window is fully utilized to complete the real-time correction, directly changing the subsequent migration direction of the current hot and humid air mass. When the hot and humid air mass has crossed the first high-risk unit but there are still high-risk units to follow, the correction results are applied to the protection of subsequent units, avoiding the waste of the correction results of the units that have been crossed. This fully utilizes the limited intervention time window and realizes the adaptive learning of the system in long-term operation.

[0042] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the present invention's concept of a real-time monitoring and early warning system for multiple parameters of cold storage apples, should be covered within the scope of protection of the present invention.

Claims

1. A cold storage apple state multi-parameter real-time monitoring and early warning system, characterized in that, include: The reverse calculation module is used to collect fin temperature sequences and air duct temperature sequences. Based on the fin temperature sequences, air duct temperature sequences and the power input sequence of the defrost heater, reverse calculation is performed to obtain the absolute humidity and absolute temperature of the hot and humid air mass. The trajectory simulation module, based on the absolute humidity and absolute temperature of the hot and humid air mass, uses a digital twin flow field model of the cold storage to simulate and obtain the spatiotemporal distribution prediction data of the hot and humid air mass. The condensation risk index calculation module is used to obtain the current skin temperature of each apple stack unit. Based on the spatiotemporal distribution prediction data of the hot and humid air mass and the current skin temperature of each apple stack unit, the condensation risk index is calculated for each apple stack unit through which the peak of the hot and humid air mass passes. The high-risk screening module identifies apple stack units that exceed the safety threshold based on the condensation risk index of each apple stack unit and a preset safety threshold. The instruction generation module generates airflow guidance instructions based on apple stacking units that exceed safety thresholds; The correction module, based on the airflow guidance command, collects temperature response data of the fan outlet area during the execution of the airflow guidance command; Adjust airflow guidance parameters based on temperature response data.

2. The cold storage apple state multi-parameter real-time monitoring and early warning system according to claim 1, characterized in that, The methods for obtaining the absolute humidity and absolute temperature of the hot and humid air mass by reverse calculation based on the fin temperature sequence, the air duct temperature sequence, and the power input sequence of the defrost heater include: The time point when the defrost heater stops heating and the fan has not yet restarted is taken as the estimated time node. Based on the estimated time node, the last value of the air duct temperature sequence before the estimated time node is taken as the temperature of the hot and humid air mass. Based on the power input sequence of the defrost heater, the total defrost input energy is obtained by integrating over time. Based on the total defrost input energy, the heat absorbed by the fins during heating, and the preset latent heat of phase change of ice and frost melting, the energy obtained by the air trapped in the air duct is obtained by subtracting the heat absorbed by the fins during heating and the latent heat of phase change of ice and frost melting from the total defrost input energy. Obtain the geometric volume of the air duct and the air density inside the air duct before frost. Based on the geometric volume of the air duct and the air density inside the air duct before frost, multiply the geometric volume of the air duct by the air density inside the air duct before defrosting to obtain the mass of the dry air trapped inside the air duct. Based on the energy gained by the trapped air, the temperature of the hot and humid air mass, and the mass of the dry air trapped inside the air duct, solve the moisture content of the hot and humid air mass in reverse using the enthalpy equation of moist air. Based on the moisture content, perform an absolute humidity conversion to obtain the absolute humidity of the hot and humid air mass.

3. The cold storage apple state multi-parameter real-time monitoring and early warning system according to claim 1, characterized in that, The methods for predicting the spatiotemporal distribution of a hot and humid air mass by simulating its absolute humidity and absolute temperature using a digital twin flow field model of a cold storage facility include: Thermodynamic state vectors are constructed based on the absolute humidity and absolute temperature of the hot and humid air mass. Based on the thermodynamic state vector, the cold storage digital twin flow field model is injected in the form of a moving mass to generate spatiotemporal distribution prediction data of hot and humid air masses.

4. The cold storage apple state multi-parameter real-time monitoring and early warning system according to claim 3, characterized in that, Methods for generating spatiotemporal distribution prediction data of hot and humid air masses by injecting them into the digital twin flow field model of a cold storage facility in the form of moving mass vectors include: The fan restart time is obtained. Based on the fan restart time and the preset time step, with the fan restart time as the time origin, the convective diffusion process of the moving material cluster in the airflow field of the storage is solved transiently. The spatial grid coordinates of the moving material cluster at each time step, as well as the corresponding temperature decay value and absolute humidity decay value are recorded. Based on the spatial grid coordinates, temperature decay values, and absolute humidity decay values ​​at each time step, the data are merged to generate the spatiotemporal distribution prediction data of the hot and humid air mass.

5. A real-time monitoring and early warning system for multiple parameters of apple status in cold storage according to claim 1, characterized in that, Based on the spatiotemporal distribution prediction data of hot and humid air masses and the current skin temperature of each apple stack unit, the method for calculating the condensation risk index for each apple stack unit through which the peak of the hot and humid air mass passes includes: Based on the spatiotemporal distribution prediction data of hot and humid air masses, the local absolute humidity and local air temperature values ​​corresponding to the peak surface of the hot and humid air mass passing through each apple stack unit are extracted. Based on the local absolute humidity value, the dew point temperature value of the surface of each apple stack unit is calculated according to the thermodynamic relationship of moist air. Based on the dew point temperature value and the current skin temperature value of the corresponding apple stack unit, the difference is calculated to obtain the condensation risk index of each apple stack unit.

6. The real-time monitoring and early warning system for multiple parameters of apple status in cold storage according to claim 1, characterized in that, Based on the condensation risk index of each apple pallet unit and a preset safety threshold, methods for identifying apple pallet units that exceed the safety threshold include: Based on the condensation risk index of each apple stack unit and the preset safety threshold, the condensation risk index of each apple stack unit is compared with the preset safety threshold one by one. If the condensation risk index is greater than the safety threshold, the apple stack unit is marked as a high-risk unit. Based on high-risk units, the spatial coordinates of each high-risk unit and the expected arrival time of the corresponding hot and humid air mass peak are extracted to generate a list of apple stack units that exceed the safety threshold.

7. A real-time monitoring and early warning system for multiple parameters of apple status in cold storage according to claim 1, characterized in that, Methods for generating airflow guidance commands based on apple stack units that exceed safety thresholds include: Obtain the spatial coordinates of each high-risk unit in the list of apple pallet units that exceed the safety threshold, and the estimated arrival time of the corresponding hot and humid air mass peak. Based on the spatial coordinates of each high-risk unit, the deviation direction and magnitude of the migration path of the hot and humid air mass are calculated, and a slow acceleration start-up control sequence with an initial wind deflection angle is generated. The control sequence for slow acceleration is used as the airflow guidance command.

8. A real-time monitoring and early warning system for multiple parameters of apple status in cold storage according to claim 7, characterized in that, Based on the spatial coordinates of each high-risk unit, the method for calculating the deviation direction and magnitude of the migration path of the hot and humid air mass, and generating a slow acceleration start-up control sequence with an initial wind deflection angle, includes: Based on the spatial coordinates of each high-risk unit, calculate the offset azimuth angle and offset distance of the center position of each high-risk unit relative to the axis of the fan outlet; Based on the offset azimuth angle, the initial deflection direction of the air guide plate is determined; based on the offset distance, the ratio of the offset distance to the air supply coverage of the air cooler is calculated to determine the initial deflection angle of the air guide plate. Based on the initial deflection direction and initial deflection angle, the initial wind deflection angle is obtained by combining them; Based on the initial air deflection angle, it is combined with the preset slow acceleration start-up curve to generate a slow acceleration start-up control sequence with the initial air deflection angle.

9. A real-time monitoring and early warning system for multiple parameters of apple status in cold storage according to claim 1, characterized in that, Methods for collecting temperature response data in the fan outlet area during the execution of airflow guidance commands, based on airflow guidance commands, include: During the execution of the airflow guidance command, the real-time temperature values ​​of each temperature measuring point in the area surrounding the air outlet of the air cooler are continuously collected at a preset sampling frequency. Based on the real-time temperature values ​​of each temperature measuring point, the real-time temperature values ​​of each temperature measuring point are arranged according to the time axis to obtain the measured temperature change curve of the air outlet area. Based on the measured temperature change curve, the measured temperature change curve is used as temperature response data.

10. A real-time monitoring and early warning system for multiple parameters of apple status in cold storage according to claim 1, characterized in that, Methods for adjusting airflow guidance parameters based on temperature response data include: Based on the temperature response data, the temperature response data is input into the cold storage digital twin flow field model and compared with the temperature decay value in the spatiotemporal distribution prediction data of the hot and humid air mass to calculate the prediction deviation of the migration trajectory. Based on the prediction bias, the air deflection angle parameter in the slow acceleration start-up control sequence is corrected to generate an adjusted airflow guidance command. Repeatedly collect temperature response data of the fan outlet area during the execution of airflow guidance commands and adjust airflow guidance parameters until the condensation risk index of all apple stack units is no greater than the safety threshold.

Images