Intelligent fresh-keeping parameter control method and device for blueberry fresh-keeping refrigerator

Abstract

This application relates to the field of computer technology, and in particular to an intelligent preservation parameter control method and device for a blueberry cold storage facility. The method includes: upon blueberries entering the cold storage, detecting and acquiring initial quality data and varietal characteristic data of the blueberries to generate an initial preservation parameter plan for the corresponding batch; using a multi-source sensor array arranged within the cold storage facility, real-time data on blueberry respiration intensity, ethylene release, surface image data, packaging microenvironment data, and temperature, humidity, and gas concentration data within the storage facility are collected; the dynamic adjustment ratio of oxygen and carbon dioxide, the optimal operating parameters of the variable frequency refrigeration unit and the guide fan, and humidity correction parameters are determined, and the remaining shelf life of the blueberries and the trend of surface microbial quantity changes are predicted; optimal control commands are generated through a preset decision model; and the optimal control commands are sent to a fuzzy PID controller, which drives each actuator. This application helps improve the preservation effect of blueberries.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to an intelligent preservation parameter control method and device for a blueberry cold storage. Background Technology

[0002] Blueberries, as a type of berry with rich nutritional value, have thin skin, juicy flesh, and the bloom covering their surface is easy to fall off. In addition, they have a high rate of respiration and metabolism after harvesting, making them prone to mold, softening, bloom shedding, and other quality deterioration during storage, which seriously affects their commercial value and shelf life.

[0003] Existing blueberry cold storage facilities primarily employ static parameter control, resulting in low precision and data disconnect between different stages. This leads to uneven temperature distribution within the storage, severe bloom shedding, and accelerated fruit aging due to ethylene accumulation, resulting in short shelf life and rapid quality deterioration of blueberries. Furthermore, traditional control methods rely heavily on manual intervention, failing to achieve dynamic closed-loop control of preservation parameters and thus unable to meet the intelligent and precise requirements of large-scale blueberry preservation.

[0004] Therefore, there is an urgent need for an intelligent method for controlling the preservation parameters of blueberry cold storage, which can achieve precise and dynamic control of blueberry preservation parameters through a fully intelligent control system to improve the preservation effect of blueberries. Summary of the Invention

[0005] Therefore, it is necessary to provide an intelligent preservation parameter control method and device for blueberry cold storage that can improve the preservation effect of blueberries, in order to address the above-mentioned technical problems.

[0006] In a first aspect, this application provides a method for intelligent preservation parameter control in a blueberry cold storage facility, the method comprising: When blueberries are put into storage, the initial quality data and varietal characteristic data of blueberries are obtained by testing, and the initial preservation parameter plan for the corresponding batch is generated to determine the initial control benchmark. The system uses a multi-source sensor array deployed inside the cold storage to collect real-time data on blueberry respiration intensity, ethylene release, surface image data, packaging microenvironment data, and temperature, humidity, and gas concentration data within the storage. The collected data were comprehensively analyzed to determine the dynamic adjustment ratio of oxygen and carbon dioxide, the optimal operating parameters of the variable frequency refrigeration unit and the guide fan, the humidity correction parameters, and to predict the remaining shelf life of blueberries and the trend of surface microbial growth. Based on the dynamic adjustment ratio, the optimal operating parameters, the humidity correction parameters, the remaining shelf life, and the trend of surface microbial quantity changes, the optimal control command is generated through a preset decision model. The optimal control command is sent to the fuzzy PID controller, which drives each actuator to dynamically control the preservation parameters of the cold storage in a closed loop.

[0007] In one embodiment, the comprehensive analysis of the collected data to determine the dynamic adjustment ratio of oxygen and carbon dioxide, the optimal operating parameters of the variable frequency refrigeration unit and the guide fan, the humidity correction parameters, and to predict the remaining shelf life of blueberries and the trend of surface microbial growth includes: The respiratory intensity data and ethylene release data are analyzed based on a machine learning model to determine the physiological stage of the blueberry and to determine the dynamic regulation ratio of oxygen and carbon dioxide based on the physiological stage. By combining computational fluid dynamics simulation models to analyze the temperature, humidity and gas concentration data in the warehouse, the optimal operating parameters of the variable frequency refrigeration unit and the guide fan are determined. The surface image data is processed by an image analysis algorithm to quantify the integrity of the blueberry bloom, and humidity correction parameters are determined based on the bloom integrity. The temperature, humidity and gas concentration data in the warehouse, as well as the packaging microenvironment data, are input into a preset shelf-life prediction model to predict the remaining shelf life of the blueberries. A microbial prediction model is constructed based on the temperature, humidity, and gas concentration data in the storage room and the packaging microenvironment data. The microbial prediction model is used to predict the trend of microbial quantity changes on the surface of blueberries.

[0008] In one embodiment, the analysis of the temperature, humidity, and gas concentration data within the storage chamber using a computational fluid dynamics simulation model to determine the optimal operating parameters for the variable frequency refrigeration unit and the guide fan includes: The temperature, humidity and gas concentration data inside the cold storage, the spatial structure parameters of the cold storage, and the blueberry loading data are input into the computational fluid dynamics simulation model to construct a three-dimensional simulation model of airflow organization and temperature field distribution inside the cold storage. Acquire the precooling curve data and precooling final temperature data synchronized during the blueberry precooling stage, and combine the precooling curve data and precooling final temperature data to iteratively correct the three-dimensional simulation model, and predict the dynamic change trend of the cold storage temperature field under different refrigeration power and different fan speed. Based on the predicted dynamic change trend, with the uniformity of the temperature field inside the cold storage as the core objective, and taking into account both refrigeration energy consumption and equipment operating losses, the optimal output power of the variable frequency refrigeration unit and the differentiated speed parameters of each guide fan are obtained.

[0009] In one embodiment, the step of acquiring precooling curve data and precooling final temperature data synchronized during the blueberry precooling stage, and combining the precooling curve data and precooling final temperature data to iteratively correct the three-dimensional simulation model and predict the dynamic change trend of the cold storage temperature field under different refrigeration powers and different fan speeds includes: Acquire the precooling curve data and precooling final temperature data synchronized during the blueberry precooling stage, and extract the temperature drop characteristic data in the precooling curve data and the temperature difference characteristic data in the precooling final temperature data as correction factors. The correction factor is incorporated into the three-dimensional simulation model to dynamically calibrate the temperature conduction coefficient and fruit heat exchange coefficient of the three-dimensional simulation model, thus completing the first iterative correction of the three-dimensional simulation model. Based on the corrected three-dimensional simulation model, historical refrigeration control data of the cold storage were substituted to verify the deviation between the temperature field simulation results of the three-dimensional simulation model under different refrigeration power and different fan speed and the actual monitoring results. If the deviation value exceeds the preset range, the airflow organization calculation parameters of the three-dimensional simulation model are corrected a second time according to the deviation value to achieve iterative convergence of the model. Based on the three-dimensional simulation model that has completed the iterative correction, the dynamic change trend of the cold storage temperature field under different refrigeration power and different fan speed is predicted.

[0010] In one embodiment, the step of processing the surface image data using an image analysis algorithm to quantify the integrity of the blueberry bloom and determining the humidity correction parameter based on the bloom integrity includes: Based on the surface image data, the peel area of ​​a single fruit is extracted using an image analysis algorithm, and interference from the fruit stem, background, and packaging texture is removed. Based on the diffuse reflection characteristics of fruit powder to light of a specific wavelength, a fruit powder feature vector is constructed using gray-level co-occurrence matrix and texture entropy algorithm, and the fruit powder shedding rate of the fruit peel area is calculated to quantify the integrity of blueberry fruit powder. If the fruit powder shedding rate exceeds the preset humidity threshold, the humidity baseline value is lowered to obtain the humidity correction parameter; After humidity correction, the image analysis algorithm continuously tracks the trend of the fruit powder shedding rate. If the fruit powder shedding rate falls back to within the preset humidity threshold, the humidification device is driven to restore the baseline humidity.

[0011] In one embodiment, the step of constructing a fruit powder feature vector using a gray-level co-occurrence matrix and texture entropy algorithm based on the diffuse reflection characteristics of fruit powder to specific wavelengths of light, and calculating the fruit powder shedding rate of the peel region to quantify the integrity of blueberry fruit powder includes: A specific near-infrared band with the most significant diffuse reflection characteristics of fruit powder was selected as the image acquisition light source. The extracted fruit peel area was subjected to directional imaging to obtain the characteristic band grayscale image of the fruit peel area. Based on the grayscale image of the feature band, multiple pixel sub-regions are divided and grayscale co-occurrence matrices of each sub-region are constructed to extract texture feature parameters. Based on the texture feature parameters, the texture entropy of each sub-region is calculated, and the texture entropy difference threshold between the powder-covered area and the powder-free area is combined to construct the powder feature vector. Based on the fruit bloom feature vector, the fruit bloom shedding rate of the fruit peel region is calculated to quantify the integrity of the blueberry fruit bloom.

[0012] In one embodiment, the step of inputting the temperature, humidity, and gas concentration data of the storage area, as well as the packaging microenvironment data, into a preset shelf-life prediction model to predict the remaining shelf life of the blueberries includes: A shelf-life prediction model integrating the XGBoost algorithm and time series analysis was constructed. The initial quality data and varietal characteristic data of the blueberries were used as the basic input dimensions, and the temperature, humidity and gas concentration data in the warehouse and the packaging microenvironment data were used as dynamic input dimensions. At the same time, the results of the blueberry physiological stage judgment were introduced as the correlation correction dimension. The shelf life prediction model was trained and validated using historical data from multiple batches of blueberries throughout their entire shelf life cycle, and the model hyperparameters were optimized using a grid search method. The real-time collected data on temperature, humidity, and gas concentration in the warehouse, as well as the packaging microenvironment data, are input into the trained shelf-life prediction model. The shelf-life prediction model outputs the remaining shelf-life prediction value for the corresponding batch of blueberries based on time series trend extrapolation.

[0013] In one embodiment, the step of detecting and acquiring initial quality data and varietal characteristic data of blueberries upon their arrival at the warehouse, and generating an initial preservation parameter plan for the corresponding batch to determine the initial control benchmark includes: When blueberries are received into the warehouse, a full-area non-destructive scanning inspection is performed to extract the initial quality data of the blueberries, and the initial quality data is then standardized and graded. Blueberry varietal characteristic data are collected using RFID radio frequency identification technology. Based on the varietal characteristic data, climacteric and non-climacteric varieties are distinguished. Combined with the grading results of the initial quality data, a preset preservation parameter database is retrieved. The parameters in the preservation parameter database are individually adapted and adjusted to generate an initial preservation parameter plan for the corresponding batch of blueberries, so as to determine the initial control benchmark.

[0014] Secondly, this application also provides an intelligent preservation parameter control device for a blueberry cold storage facility. The device includes: The parameter baseline determination module is used to detect and obtain the initial quality data and variety characteristic data of blueberries when they are put into storage, and generate the initial preservation parameter plan for the corresponding batch to determine the initial control baseline. The data acquisition module is used to collect real-time data on blueberry respiration intensity, ethylene release, surface image data, packaging microenvironment data, and temperature, humidity and gas concentration data inside the cold storage through a multi-source sensor array arranged inside the cold storage. The data analysis module is used to comprehensively analyze various types of data collected, determine the dynamic adjustment ratio of oxygen and carbon dioxide, the optimal operating parameters of the variable frequency refrigeration unit and the guide fan, the humidity correction parameters, and predict the remaining shelf life of blueberries and the trend of surface microbial quantity changes. The control command generation module is used to generate optimal control commands based on the dynamic adjustment ratio, the optimal operating parameters, the humidity correction parameters, the remaining shelf life, and the trend of surface microbial quantity changes through a preset decision model. The control command execution module is used to send the optimal control command to the fuzzy PID controller, and drive each actuator through the fuzzy PID controller to dynamically control the preservation parameters of the cold storage in a closed loop.

[0015] In summary, this application includes the following beneficial technical effects: When blueberries are received into storage, an initial preservation parameter plan for each batch is generated, avoiding the problem of mismatch between the traditional single parameter setting and the characteristics of blueberries themselves. This lays a precise benchmark foundation for subsequent dynamic control. Real-time collection and comprehensive analysis of various blueberry data enables the accurate determination of various core preservation control parameters. Simultaneously, it allows for advance understanding of blueberry quality degradation and microbial growth trends, shifting preservation control from passive response to proactive prediction, thus improving the foresight and accuracy of control. Combining dynamic adjustment ratios, optimal operating parameters, humidity correction parameters, remaining shelf life, and surface microbial quantity trends, optimal control commands are generated through a pre-set decision model. This avoids the imbalance between preservation effect and operating efficiency caused by traditional single-target control, ensuring the scientific and optimal nature of the control commands. A fuzzy PID controller enables precise and rapid execution of optimal control commands, allowing preservation parameters to be adjusted in real time according to blueberry conditions and environmental changes, always maintaining them within the optimal preservation range. This solves the technical problem of parameters not being able to dynamically adapt in traditional static control, effectively guaranteeing the preservation quality of blueberries. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating an intelligent preservation parameter control method for a blueberry cold storage facility in one embodiment. Figure 2This is a flowchart illustrating the intelligent preservation parameter control method for a blueberry cold storage facility in another embodiment. Figure 3 This is a structural block diagram of an intelligent preservation parameter control device for a blueberry cold storage in one embodiment. Detailed Implementation

[0017] This invention provides a method and device for intelligent preservation parameter control in a blueberry cold storage facility.

[0018] The embodiments of the present invention will now be described in more detail with reference to the accompanying drawings. While some embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the present disclosure. It should be understood that the accompanying drawings and embodiments are for illustrative purposes only and are not intended to limit the scope of protection of the present invention.

[0019] In the description of the embodiments disclosed in this invention, the term "comprising" and similar terms should be understood as open-ended inclusion, i.e., "including but not limited to". The term "based on" should be understood as "at least partially based on". The term "one embodiment" or "the embodiment" should be understood as "at least one embodiment". The terms "first", "second", etc., may refer to different or the same objects. Other explicit and implicit definitions may also be included below.

[0020] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 One embodiment of the intelligent preservation parameter control method for blueberry cold storage in this invention includes: S100 detects and obtains initial quality data and varietal characteristic data of blueberries when they are put into storage, and generates an initial preservation parameter plan for the corresponding batch to determine the initial control benchmark.

[0021] Specifically, after blueberries are harvested, they are immediately transported to cold storage for pre-cooling using vacuum or differential pressure pre-cooling technology. This rapidly lowers the fruit temperature and inhibits initial respiration and metabolism. This pre-cooling stage breaks the traditional disconnect between pre-cooling and storage, providing fundamental data support for precise temperature control within the storage facility. This ensures seamless temperature management and prevents quality deterioration due to insufficient pre-cooling or temperature fluctuations. IoT sensors are integrated into the pre-cooling equipment to collect and record the pre-cooling curve of each batch of blueberries in real time—that is, the rate at which temperature decreases over time. Simultaneously, this pre-cooling data (including final pre-cooling temperature and cooling duration) is synchronized to the central cold storage intelligent control system. After pre-cooling, the blueberries are transported to cold storage. Upon arrival, they undergo batch testing using professional equipment to obtain initial quality data for each batch, including indicators such as fruit sugar content, acidity, firmness, and surface defects. Simultaneously, varietal characteristic data is collected, including variety, origin, harvest time, and harvest maturity. Based on the initial quality and varietal characteristic data for each batch, and combined with the fundamental principles of blueberry preservation, differentiated initial preservation parameter plans are generated for different batches, clearly defining the control benchmarks for core parameters such as initial temperature and humidity, gas concentration, and fan operation during storage for each batch.

[0022] The S200 uses a multi-source sensor array deployed inside the cold storage to collect real-time data on blueberry respiration intensity, ethylene release, surface image data, packaging microenvironment data, and temperature, humidity, and gas concentration data within the storage.

[0023] Specifically, a multi-source sensor array is deployed within the central cold storage to construct a comprehensive data acquisition system. This system continuously and in real-time collects various data related to blueberry preservation, including physiological data such as blueberry respiration rate and ethylene release, reflecting the metabolic activity of blueberries; blueberry surface image data, reflecting the appearance and bloom condition of blueberries; packaging microenvironment data such as temperature, humidity, and gas concentration within RFID smart packaging boxes, obtaining environmental parameters close to the fruit; and macro-environmental data such as temperature, humidity, oxygen, carbon dioxide, and ethylene concentration in various areas of the cold storage, to understand the overall environmental status of the cold storage.

[0024] The S300 performs comprehensive analysis on the collected data to determine the dynamic adjustment ratio of oxygen and carbon dioxide, the optimal operating parameters of the variable frequency refrigeration unit and the guide fan, the humidity correction parameters, and predicts the remaining shelf life of blueberries and the changing trend of surface microbial counts.

[0025] Specifically, a comprehensive analysis was conducted on the collected data from multiple dimensions, including fruit physiology, packaging microenvironment, cold storage environment, and surface images. Through analytical logic, various data were transformed into practical preservation control parameters. This determined the dynamic adjustment ratios of oxygen and carbon dioxide in the cold storage, the operating parameters of the variable frequency refrigeration unit, the operating parameters of the guide fan, and humidity correction parameters, clarifying the core control direction under the current storage conditions. Simultaneously, based on the data change patterns and the quality degradation patterns of blueberries, the remaining shelf life of the current batch of blueberries was predicted, along with the trend of changes in the number of microorganisms on the blueberry surface, providing a trend reference for subsequent control decisions.

[0026] The S400 generates optimal control commands based on a preset decision model, taking into account dynamic adjustment ratios, optimal operating parameters, humidity correction parameters, remaining shelf life, and the changing trend of surface microbial counts.

[0027] Specifically, the dynamic adjustment ratios of oxygen and carbon dioxide obtained from comprehensive analysis, the optimal operating parameters of the variable frequency refrigeration unit and the guide fan, humidity correction parameters, and the prediction results of the remaining shelf life of blueberries and the changing trend of surface microorganisms based on multi-source data are all imported into a pre-set multi-objective optimization decision model as the overall input set. This pre-set decision model takes ensuring the freshness and quality of blueberries, minimizing the energy consumption of cold storage equipment, and reducing equipment operating losses as core optimization objectives, while incorporating constraints such as shelf life warning thresholds and microbial safety control thresholds. It constructs a multi-objective collaborative optimization computing system that can adjust the input gas adjustment ratios, temperature and humidity, equipment operation and other control parameters in relation to shelf life and microbial growth. By deeply correlating and dynamically weighting risk trend data such as plant growth, and combining the real-time physiological state of blueberries and the changing characteristics of the cold storage environment, the optimal solution for the coordinated operation of each actuator is obtained through model iterative calculation, under the premise of meeting the core parameter control requirements for blueberry preservation and avoiding preservation risks such as excessive microbial growth and significant shortening of shelf life. Finally, the optimal control command with accuracy, adaptability and economy is generated. This command not only clarifies the specific operating parameters and action instructions of each actuator, but also realizes the control coordination between different actuators. It breaks through the shortcomings of traditional control where each device operates alone and the parameter matching degree is low, and provides a scientific, unified and forward-looking decision-making basis for the closed-loop precise execution of subsequent preservation parameters.

[0028] The S500 sends the optimal control command to the fuzzy PID controller, which drives each actuator to dynamically control the preservation parameters of the cold storage in a closed loop.

[0029] Specifically, the optimal control commands generated by the preset decision model are directed to the fuzzy PID controller in the cold storage. This controller integrates the adaptive advantages of fuzzy control with the precise adjustment characteristics of PID control, breaking through the limitations of traditional single controllers in adapting to complex preservation environments. It can dynamically adapt and accurately parse control commands based on the real-time changes in multi-source data within the cold storage, breaking down the overall control commands into detailed action commands that match the operating logic of each actuator. Simultaneously, this fuzzy PID controller constructs the collaborative drive logic for each actuator, synchronously driving the variable frequency refrigeration unit, nitrogen / carbon dioxide generator, zoned humidifier / dehumidifier, directional flow fan, ethylene removal device, and other actuators to operate collaboratively according to commands, achieving… The controller enables synchronized and precise adjustment of preservation parameters such as temperature, humidity, controlled atmosphere parameters, and airflow organization. During the operation of the actuators, the controller continuously receives real-time feedback data from a multi-source sensor array, including the cold storage environment and the physiological state of the blueberries. It compares the feedback data with the target values ​​of the parameters corresponding to the optimal control commands in real time and performs deviation analysis. If the parameter deviation is detected to exceed the preset range, the controller immediately and automatically adjusts the operating parameters of each actuator. This overcomes the technical drawbacks of traditional control methods, which lack feedback and correction after parameter adjustment. It allows various preservation parameters in the cold storage to adapt and adjust precisely according to the dynamic changes in the physiological state of the blueberries and the cold storage environment, always remaining stable within the optimal range that meets the preservation needs of blueberries. This ensures the real-time, accurate, and coordinated control of preservation parameters.

[0030] In one embodiment, such as Figure 2 As shown, S300 includes: S310 uses a machine learning model to analyze respiratory intensity data and ethylene release data to determine the physiological stage of blueberries and to determine the dynamic regulation ratio of oxygen and carbon dioxide based on the physiological stage. S320, combined with computational fluid dynamics simulation model, analyzes the temperature, humidity and gas concentration data in the warehouse to determine the optimal operating parameters of the variable frequency refrigeration unit and the guide fan; The S330 processes surface image data through image analysis algorithms, quantifies the integrity of blueberry bloom, and determines humidity correction parameters based on the bloom integrity. S340 inputs the temperature, humidity and gas concentration data in the warehouse and the packaging microenvironment data into the preset shelf life prediction model to predict the remaining shelf life of blueberries; S350 uses data on temperature, humidity, and gas concentration in the storage room, as well as data on the packaging microenvironment, to build a microbial prediction model. This model is used to predict the changing trend of the number of microorganisms on the surface of blueberries.

[0031] Specifically, firstly, a machine learning model is used to deeply analyze blueberry respiration intensity and ethylene release data. Leveraging the model's ability to fit and analyze fruit physiological metabolic data, the physiological stage of the blueberry is accurately determined. Based on the respiratory metabolic characteristics of different physiological stages, the appropriate oxygen and carbon dioxide regulation ratio is dynamically determined, ensuring a high degree of match between controlled atmosphere parameters and the real-time metabolic needs of the blueberry. Specifically, time-series data on respiration intensity and ethylene release from multiple batches of blueberries throughout their storage period are used, combined with manually calibrated physiological stage labels (such as the stable respiratory period, the pre-peak respiratory period, and the metabolic decline period), to train and iteratively optimize the machine learning model. This allows the model to deeply fit the nonlinear mapping relationship between the changes in blueberry metabolic indicators and the physiological stages. During actual cold storage, real-time data on blueberry respiration intensity and ethylene release collected by a multi-source sensor array are input into the trained model. The model determines the specific physiological stage of blueberries by extracting and matching real-time metabolic features. Furthermore, this application does not assign a single, fixed oxygen and carbon concentration to each physiological stage. Instead, based on the physiological stage determined by the model, and combined with the respiratory metabolic characteristics of blueberries at that stage and the core requirements of controlled atmosphere preservation, it pre-sets differentiated dynamic oxygen and carbon regulation strategies. For example, during the stable respiratory period, it maintains an oxygen and carbon concentration suitable for basal metabolism to ensure fruit activity. In the early stages of the respiratory peak, it triggers a gradient parameter adjustment mechanism to precisely lower the oxygen concentration and moderately increase the carbon dioxide concentration, fundamentally inhibiting the respiratory metabolic rate and ethylene synthesis and release of blueberries. The adjustment ratio can be finely adjusted according to the real-time changes in respiratory intensity and ethylene release, achieving dynamic matching between controlled atmosphere parameters and the real-time metabolic state of blueberries. This ensures that the cold storage controlled atmosphere environment is always highly adapted to the physiological stage of blueberries, significantly improving the accuracy and effectiveness of controlled atmosphere preservation.Secondly, computational fluid dynamics (CFD) simulation models were used to analyze the temperature, humidity, and gas concentration data within the cold storage. Leveraging the model's fluid dynamics and temperature field simulation characteristics, the optimal operating parameters for the variable frequency refrigeration unit and the guide fan were accurately calculated and determined to ensure the uniformity of the temperature and airflow fields in the cold storage. Simultaneously, image analysis algorithms were used to professionally process blueberry surface images, accurately quantifying the integrity of the blueberry bloom and dynamically determining humidity correction parameters based on changes in bloom integrity, thus addressing both the humidity preservation and bloom protection requirements of blueberries. Furthermore, the temperature, humidity, and gas concentration data within the cold storage, along with packaging microenvironment data, were input into a pre-set shelf-life prediction model. Through the model's data analysis and trend extrapolation capabilities, the remaining shelf life of the blueberries was accurately predicted. Finally, a dedicated microbial prediction model was constructed based on the temperature, humidity, and gas concentration data within the cold storage, and the packaging microenvironment data. By fitting the model to the microbial growth and reproduction patterns, the trend of microbial quantity changes on the blueberry surface was predicted, providing a basis for subsequent preservation adjustments. The control and decision-making process provides comprehensive and accurate multi-dimensional data. Specifically, it first uses real-time temperature, humidity, and gas concentration data collected in the cold storage, as well as precise data on the temperature, humidity, and gas concentration of the blueberry packaging microenvironment, as the core foundation. Combined with the relevant laws of microbial growth and reproduction during blueberry preservation and historical monitoring data, a microbial prediction model that fits the actual blueberry storage scenario is constructed. This model can accurately fit the correlation between changes in environmental parameters and the growth of microbial numbers on the blueberry surface. Subsequently, real-time monitoring data of the cold storage and packaging microenvironment are continuously input into the microbial prediction model. Through analysis and calculation of dynamic changes in environmental parameters, the model simulates and extrapolates the growth and reproduction status of microorganisms on the blueberry surface under different environmental conditions, and then accurately predicts the dynamic change trend of the number of microorganisms on the blueberry surface during the subsequent storage period. This allows for early understanding of the potential risk of excessive microbial growth, providing a scientific basis for timely implementation of targeted preservation and control measures and ensuring the quality of blueberry storage.

[0032] In one embodiment, by combining computational fluid dynamics simulation models to analyze the temperature, humidity, and gas concentration data within the storage chamber, the optimal operating parameters for the variable frequency chiller and the guide fan are determined, including: Data on temperature, humidity, and gas concentration within the cold storage, along with spatial structure parameters and blueberry loading volume, are input into a computational fluid dynamics (CFD) simulation model to construct a three-dimensional simulation model of airflow organization and temperature field distribution within the cold storage. Pre-cooling curve data and final pre-cooling temperature data are obtained synchronously during the blueberry pre-cooling stage. These data are then combined to iteratively refine the three-dimensional simulation model, predicting the dynamic trends of the cold storage temperature field under different refrigeration capacities and fan speeds. Based on these predicted dynamic trends, and with the uniformity of the temperature field within the cold storage as the core objective, while also considering refrigeration energy consumption and equipment operating losses, the optimal output power of the variable frequency refrigeration unit and the differentiated speed parameters of each guide fan are calculated.

[0033] Specifically, firstly, real-time data on temperature, humidity, and gas concentration within the cold storage, along with the inherent spatial structure parameters of the cold storage and the actual amount of blueberries entering the cold storage, are integrated and uniformly input into a computational fluid dynamics simulation model. Leveraging the model's 3D modeling and simulation capabilities, a 3D simulation model capable of realistically reproducing the airflow organization and temperature field distribution within the cold storage is constructed. Secondly, pre-cooling curve data and final pre-cooling temperature data collected synchronously during the blueberry pre-cooling stage are acquired and integrated into the 3D simulation model for iterative correction. This allows the model's simulation calculations to better reflect the actual heat transfer between the blueberries and the cold storage environment, significantly improving the model's simulation accuracy. Subsequently, based on the iteratively corrected 3D simulation model, simulations and pre-cooling data are performed. The study measured the spatiotemporal dynamic changes of the temperature field within a cold storage facility under different output power levels and varying speed combinations of the various guide fans. Finally, with the uniformity of the temperature field within the cold storage facility as the core objective, a multi-objective optimization system was formed, incorporating the operating energy consumption of the refrigeration unit and the operating losses of the refrigeration and fan equipment. Combining the temperature field dynamic changes predicted by the model, the optimal output power of the variable frequency refrigeration unit was obtained through iterative calculations using a computational fluid dynamics simulation model, while ensuring a uniform temperature distribution throughout the cold storage facility and meeting the temperature requirements for blueberry preservation. Furthermore, considering the temperature field differences in different areas of the cold storage facility due to variations in spatial location and heat load distribution, the differentiated speed parameters of each guide fan were determined.

[0034] In one embodiment, precooling curve data and precooling final temperature data synchronized during the blueberry precooling stage are acquired. Combining these data, the three-dimensional simulation model is iteratively corrected to predict the dynamic change trend of the cold storage temperature field under different refrigeration capacities and fan speeds. Acquire synchronized precooling curve data and precooling final temperature data during the blueberry precooling stage, and extract temperature drop characteristic data from the precooling curve data and temperature difference characteristic data from the precooling final temperature data as correction factors. Integrate the correction factors into the 3D simulation model to dynamically calibrate the temperature transfer coefficient and fruit heat exchange coefficient of the 3D simulation model, completing the first iterative correction of the 3D simulation model. Based on the corrected 3D simulation model, substitute historical refrigeration control data of the cold storage to verify the deviation between the temperature field simulation results of the 3D simulation model under different refrigeration powers and different fan speeds and the actual monitoring results. If the deviation exceeds the preset range, perform a second correction on the airflow organization calculation parameters of the 3D simulation model according to the deviation value to achieve iterative convergence of the model. Based on the 3D simulation model that has completed iterative correction, predict the dynamic change trend of the cold storage temperature field under different refrigeration powers and different fan speeds.

[0035] Specifically, precooling curve data and final precooling temperature data were acquired synchronously during the blueberry precooling stage. Temperature drop characteristic data, such as the rate of temperature decrease and the duration of temperature stabilization, were extracted from the precooling curve data. Temperature difference characteristic data between the fruit and the precooling environment was extracted from the final precooling temperature data. These characteristic data were then integrated as correction factors. Subsequently, this correction factor was incorporated into the constructed 3D simulation model. Precise dynamic calibration was performed on the temperature conduction coefficient (which affects heat transfer calculations) and the fruit heat exchange coefficient (which reflects the characteristics of blueberry fruit), thus completing the first iterative correction of the 3D simulation model. This ensured that the model's core computational parameters better reflected the actual heat exchange patterns between the blueberry fruit and the cold storage environment. Next, based on the 3D simulation model that underwent the first iterative correction, historical refrigeration control data under different refrigeration control conditions in the cold storage were substituted to simulate the cold storage temperature under different combinations of refrigeration power and fan speeds. The simulation results are compared with the actual temperature field monitoring results under the corresponding operating conditions to calculate the deviation value between the two, thereby verifying the simulation accuracy of the model. If the calculated deviation value exceeds the preset range, the airflow organization calculation parameters affecting the airflow calculation in the 3D simulation model are specifically corrected based on the magnitude of the deviation value, the distribution pattern, and the key reasons for the deviation. Through multiple rounds of calibration, the model achieves iterative convergence, making the simulation results of the model highly consistent with the actual operating state of the cold storage. Finally, based on the 3D simulation model that has completed iterative correction, the spatiotemporal dynamic change trend of the cold storage temperature field under different combinations of refrigeration power and different fan speeds is simulated and predicted. The change law, homogenization effect, and achievement time of the temperature field under each control condition are clarified, providing accurate and reliable trend basis for solving the optimal operating parameters of the variable frequency refrigeration unit and the guide fan.

[0036] In one embodiment, processing surface image data using an image analysis algorithm to quantify the integrity of blueberry bloom and determining humidity correction parameters based on bloom integrity includes: Based on surface image data, the peel area of ​​a single fruit is extracted using image analysis algorithms, removing interference from the stem, background, and packaging texture. Based on the diffuse reflection characteristics of fruit powder to specific wavelengths of light, a fruit powder feature vector is constructed using gray-level co-occurrence matrix and texture entropy algorithms to calculate the fruit powder shedding rate of the peel area, thus quantifying the integrity of the blueberry fruit powder. If the fruit powder shedding rate exceeds a preset humidity threshold, the humidity baseline value is lowered to obtain a humidity correction parameter. After humidity correction, the changing trend of the fruit powder shedding rate is continuously tracked using image analysis algorithms. If the fruit powder shedding rate falls back to within the preset humidity threshold, the humidification device is driven to restore the baseline humidity.

[0037] Specifically, firstly, based on the collected blueberry surface image data, image analysis algorithms are used to extract targets and remove backgrounds from the images, extracting the peel area of ​​individual fruits and effectively removing irrelevant information such as fruit stems, backgrounds, and packaging textures to ensure the accuracy of subsequent fruit powder analysis. Secondly, based on the diffuse reflection characteristics of fruit powder to specific wavelengths of light, gray-level co-occurrence matrix and texture entropy algorithms are used to perform feature analysis on the peel area images, constructing a fruit powder feature vector that can accurately represent the state of fruit powder, and calculating the fruit powder shedding rate of the peel area through feature vector analysis, thereby achieving a quantitative representation of the integrity of blueberry fruit powder. Then, the actual bloom shedding rate is compared with the preset humidity threshold. If the bloom shedding rate exceeds the threshold, the humidity baseline value is immediately lowered, and corresponding humidity correction parameters are generated to reduce bloom shedding through dehumidification. Finally, after completing the humidity correction, the changing trend of blueberry bloom shedding rate in the area is continuously tracked through image analysis algorithms. If the bloom shedding rate falls back to within the preset humidity threshold, the humidification equipment is promptly driven to restore the humidity to the baseline value, forming a closed-loop control of "image analysis - bloom quantification - humidity correction - trend tracking - parameter restoration", which takes into account both the humidity requirements for blueberry preservation and the requirements for bloom protection.

[0038] In one embodiment, based on the diffuse reflection characteristics of fruit powder to light of a specific wavelength, a fruit powder feature vector is constructed using a gray-level co-occurrence matrix and texture entropy algorithm to calculate the fruit powder shedding rate in the peel area, thereby quantifying the integrity of blueberry fruit powder, including: A specific near-infrared band with the most significant diffuse reflection characteristics of the fruit bloom was selected as the image acquisition light source. The extracted fruit peel area was then subjected to directional imaging to obtain a grayscale image of the characteristic band of the peel area. Based on the grayscale image of the characteristic band, multiple pixel sub-regions were divided, and a grayscale co-occurrence matrix was constructed for each sub-region to extract texture feature parameters. The texture entropy of each sub-region was calculated based on the texture feature parameters. Combined with the texture entropy difference threshold between the bloom-covered and unbloomed areas, a fruit bloom feature vector was constructed. Based on the fruit bloom feature vector, the fruit bloom shedding rate of the peel area was calculated to quantify the integrity of the blueberry bloom.

[0039] Specifically, firstly, a specific near-infrared band with the most significant diffuse reflection characteristics of the fruit powder is selected as the dedicated image acquisition light source. Directional imaging is then performed on the extracted blueberry peel area to obtain a grayscale image of the peel area's characteristic bands that reflects the distribution of the fruit powder, allowing for a clear distinction between the presence and density of the fruit powder in the image. Secondly, based on the acquired grayscale image of the characteristic bands, the peel area is divided into multiple pixel sub-regions. A corresponding grayscale co-occurrence matrix is ​​constructed for each sub-region, and core texture feature parameters are extracted from the grayscale co-occurrence matrix. Then, the texture entropy of each sub-region is calculated based on the extracted texture feature parameters. Combined with the previously calibrated texture entropy difference threshold between the powder-covered and powder-free areas, a fruit powder feature vector that characterizes the fruit powder coverage state is constructed, establishing a precise mapping relationship between texture features and fruit powder state. Finally, each pixel in the peel area is classified and identified point-by-point according to this mapping relationship, accurately distinguishing between powder-covered pixels and powder-shedded pixels. By statistically analyzing the proportion of powder-shedded pixels to the total number of pixels in the peel area, the fruit powder shedding rate of the peel area is calculated.

[0040] In one embodiment, data on temperature, humidity, and gas concentration within the storage facility, along with data on the packaging microenvironment, are input into a pre-defined shelf-life prediction model to predict the remaining shelf life of the blueberries. A shelf-life prediction model integrating the XGBoost algorithm and time series analysis was constructed. Initial quality data and varietal characteristics of blueberries were used as basic input dimensions, while temperature, humidity, and gas concentration data within the storage facility, as well as packaging microenvironment data, were used as dynamic input dimensions. Simultaneously, the physiological stage assessment results of blueberries were incorporated as a correlation correction dimension. The shelf-life prediction model was trained and validated using historical data from multiple batches of blueberries throughout their entire shelf-life cycle, and the model's hyperparameters were optimized using a grid search method. Real-time collected temperature, humidity, and gas concentration data within the storage facility, as well as packaging microenvironment data, were input into the trained shelf-life prediction model. Based on time series trend extrapolation, the model outputs the predicted remaining shelf-life value for the corresponding batch of blueberries.

[0041] Specifically, firstly, a shelf-life prediction model integrating the XGBoost algorithm and time series analysis was constructed. Initial quality data and varietal characteristic data of blueberries were used as the basic input dimensions of the model, reflecting the basic preservation characteristics of blueberries themselves. Temperature, humidity, and gas concentration data within the storage facility, as well as packaging microenvironment data, were used as dynamic input dimensions to reflect the impact of environmental changes during storage on shelf life. Simultaneously, the results of blueberry physiological stage assessments were introduced as a correlation correction dimension, allowing the model to closely match the real-time metabolic state of blueberries and achieve comprehensive integration of multi-dimensional data. Secondly, historical data from multiple batches of blueberries throughout their entire preservation cycle were used to predict the shelf life. The predictive model was trained and validated, and its hyperparameters were optimized using a grid search method to enable the model to accurately simulate the nonlinear mapping relationship between various input data and the shelf life of blueberries, thereby improving the model's prediction accuracy. Finally, real-time collected data on temperature, humidity, and gas concentration in the storage facility, as well as packaging microenvironment data, were input into the trained and optimized shelf-life prediction model. Relying on the trend extrapolation capabilities of time series analysis and the precise fitting capabilities of the XGBoost algorithm, the model dynamically analyzes the storage status of blueberries and ultimately outputs the predicted remaining shelf life of the corresponding batch of blueberries, providing accurate trend data for preservation control.

[0042] In one embodiment, upon blueberries being stored, initial quality data and varietal characteristic data of the blueberries are detected and acquired to generate an initial preservation parameter plan for the corresponding batch, thereby determining the initial control benchmark, including: Upon blueberries entering the warehouse, a full-area non-destructive scanning inspection is performed to extract initial quality data, which is then standardized and graded. RFID technology is used to collect blueberry varietal characteristic data, distinguishing between climacteric and non-climacteric varieties. Combined with the grading results of the initial quality data, a pre-set database of preservation parameters is retrieved. The parameters in this database are then individually adjusted to generate an initial preservation parameter plan for the corresponding batch of blueberries, thus establishing the initial control baseline.

[0043] Specifically, firstly, during the blueberry warehousing process, a full-area non-destructive scanning detection technology is used to accurately extract core initial quality data such as sugar content, acidity, firmness, and surface defects. Simultaneously, the quality data undergoes standardized grading to clearly define the quality grade of different batches of blueberries, overcoming the limitations of traditional testing methods that damage the fruit and produce incomplete data. Secondly, RFID radio frequency identification technology is used to quickly collect varietal characteristic data such as blueberry variety, origin, harvest time, and harvest maturity, achieving rapid, non-contact collection of blueberry characteristic data. Based on this varietal characteristic data, it accurately distinguishes between climacteric and non-climacteric blueberries. For each batch of blueberries, a pre-set database of preservation parameters is retrieved based on the quality grading results. Finally, according to the physiological and metabolic characteristics of different varieties and the preservation requirements of different quality grades, the parameters such as temperature, humidity, gas concentration, and fan operation in the database are adjusted in a personalized way. This generates a unique initial preservation parameter plan for each batch of blueberries, clearly defining differentiated initial control benchmarks. This breaks the traditional model of uniform parameter setting for preservation, allowing subsequent dynamic control to have a precise starting point that fits the characteristics of blueberries themselves. This improves the adaptability and accuracy of control from the source of preservation, laying the foundation for high-quality preservation throughout the entire process.

[0044] In one embodiment, such as Figure 3 As shown, an intelligent preservation parameter control device for a blueberry cold storage is provided, comprising: a parameter reference determination module 10, a data acquisition module 20, a data analysis module 30, a control command generation module 40, and a control command execution module 50, wherein: The parameter baseline determination module 10 is used to detect and obtain the initial quality data and variety characteristic data of blueberries when they are put into storage, and generate the initial preservation parameter plan for the corresponding batch to determine the initial control baseline. The data acquisition module 20 is used to collect data on the respiration intensity of blueberries, ethylene release, surface image data, packaging microenvironment data, and temperature, humidity and gas concentration data in the cold storage in real time through a multi-source sensor array arranged in the cold storage. The data analysis module 30 is used to comprehensively analyze the collected data, determine the dynamic adjustment ratio of oxygen and carbon dioxide, the optimal operating parameters of the variable frequency refrigeration unit and the guide fan, the humidity correction parameters, and predict the remaining shelf life of blueberries and the trend of surface microbial quantity changes. The control command generation module 40 is used to generate the optimal control command based on the dynamic adjustment ratio, optimal operating parameters, humidity correction parameters, remaining shelf life and the trend of surface microbial quantity changes through a preset decision model. The control command execution module 50 is used to send the optimal control command to the fuzzy PID controller, and drive each actuator through the fuzzy PID controller to dynamically and closed-loop regulate the preservation parameters of the cold storage.

[0045] In one embodiment, the data analysis module 30 is further configured to analyze respiration intensity data and ethylene release data based on a machine learning model to determine the physiological stage of the blueberries and determine the dynamic adjustment ratio of oxygen and carbon dioxide based on the physiological stage; analyze the temperature, humidity and gas concentration data in the storage area using a computational fluid dynamics simulation model to determine the optimal operating parameters of the variable frequency refrigeration unit and the guide fan; process the surface image data using an image analysis algorithm to quantify the integrity of the blueberry bloom and determine the humidity correction parameters based on the bloom integrity; input the temperature, humidity and gas concentration data in the storage area and the packaging microenvironment data into a preset shelf life prediction model to predict the remaining shelf life of the blueberries; and construct a microbial prediction model based on the temperature, humidity and gas concentration data in the storage area and the packaging microenvironment data to predict the trend of microbial quantity changes on the blueberry surface.

[0046] In one embodiment, the data analysis module 30 is further used to input the temperature, humidity and gas concentration data in the cold storage, the spatial structure parameters of the cold storage, and the blueberry loading data into the computational fluid dynamics simulation model to construct a three-dimensional simulation model of the airflow organization and temperature field distribution in the cold storage; to obtain the pre-cooling curve data and pre-cooling final temperature data synchronously during the blueberry pre-cooling stage, and to iteratively correct the three-dimensional simulation model by combining the pre-cooling curve data and pre-cooling final temperature data, predicting the dynamic change trend of the cold storage temperature field under different refrigeration powers and different fan speeds; based on the predicted dynamic change trend, with the uniformity of the temperature field in the cold storage as the core objective, while taking into account refrigeration energy consumption and equipment operating losses, to solve for the optimal output power of the variable frequency refrigeration unit and the differentiated speed parameters of each guide fan.

[0047] In one embodiment, the data analysis module 30 is further used to acquire precooling curve data and precooling final temperature data synchronized during the blueberry precooling stage, and extract temperature drop characteristic data and temperature difference characteristic data from the precooling curve data and the precooling final temperature data as correction factors; integrate the correction factors into the three-dimensional simulation model, dynamically calibrate the temperature transfer coefficient and fruit heat exchange coefficient of the three-dimensional simulation model, and complete the first iterative correction of the three-dimensional simulation model; based on the corrected three-dimensional simulation model, substitute historical refrigeration control data of the cold storage, and verify the deviation between the temperature field simulation results of the three-dimensional simulation model under different refrigeration powers and different fan speeds and the actual monitoring results; if the deviation value exceeds the preset range, the airflow organization calculation parameters of the three-dimensional simulation model are corrected a second time according to the deviation value to achieve iterative convergence of the model, and based on the three-dimensional simulation model that has completed iterative correction, predict the dynamic change trend of the cold storage temperature field under different refrigeration powers and different fan speeds.

[0048] In one embodiment, the data analysis module 30 is further configured to extract the peel area of ​​a single fruit based on surface image data using an image analysis algorithm, and remove interference from fruit stems, background, and packaging texture; based on the diffuse reflection characteristics of fruit powder to specific wavelengths of light, a fruit powder feature vector is constructed using a gray-level co-occurrence matrix and texture entropy algorithm to calculate the fruit powder shedding rate of the peel area, thereby quantifying the integrity of blueberry fruit powder; if the fruit powder shedding rate exceeds a preset humidity threshold, the humidity baseline value is lowered to obtain a humidity correction parameter; after humidity correction, the image analysis algorithm continuously tracks the changing trend of the fruit powder shedding rate, and if the fruit powder shedding rate falls back to within the preset humidity threshold, the humidification device is driven to restore the baseline humidity.

[0049] In one embodiment, the data analysis module 30 is further configured to select a specific near-infrared band with the most significant diffuse reflection characteristics of the fruit powder as the image acquisition light source, perform directional imaging on the extracted fruit peel area, and obtain a grayscale image of the characteristic band of the fruit peel area; based on the grayscale image of the characteristic band, divide multiple pixel sub-regions and construct a grayscale co-occurrence matrix for each sub-region, and extract texture feature parameters; calculate the texture entropy of each sub-region based on the texture feature parameters, and construct a fruit powder feature vector by combining the texture entropy difference threshold between the fruit powder-covered area and the non-fruit powder area; and calculate the fruit powder shedding rate of the fruit peel area based on the fruit powder feature vector to quantify the integrity of the blueberry fruit powder.

[0050] In one embodiment, the data analysis module 30 is also used to construct a shelf-life prediction model that integrates the XGBoost algorithm and time series analysis. Initial quality data and varietal characteristic data of blueberries are used as basic input dimensions, while temperature, humidity, and gas concentration data within the storage facility and packaging microenvironment data are used as dynamic input dimensions. Simultaneously, the physiological stage judgment results of blueberries are introduced as a correlation correction dimension. The shelf-life prediction model is trained and validated using historical data from multiple batches of blueberries throughout their preservation cycle, and the model's hyperparameters are optimized using a grid search method. Real-time collected temperature, humidity, and gas concentration data within the storage facility and packaging microenvironment data are input into the trained shelf-life prediction model, which outputs the predicted remaining shelf-life value for the corresponding batch of blueberries based on time series trend extrapolation.

[0051] In one embodiment, the parameter benchmark determination module 10 is also used to perform full-area non-destructive scanning detection when blueberries are put into storage, extract the initial quality data of blueberries, and perform standardized grading processing on the initial quality data; collect the variety characteristic data of blueberries through RFID radio frequency identification technology, and distinguish between climacteric and non-climacteric varieties based on the variety characteristic data, and retrieve the preset preservation parameter base database in combination with the grading results of the initial quality data; perform personalized adaptation and adjustment of the parameters in the preservation parameter base database, and generate the initial preservation parameter plan for the corresponding batch of blueberries to determine the initial control benchmark.

[0052] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A method for intelligent control of preservation parameters in a blueberry cold storage facility, characterized in that, include: When blueberries are put into storage, the initial quality data and varietal characteristic data of blueberries are obtained by testing, and the initial preservation parameter plan for the corresponding batch is generated to determine the initial control benchmark. The system uses a multi-source sensor array deployed inside the cold storage to collect real-time data on blueberry respiration intensity, ethylene release, surface image data, packaging microenvironment data, and temperature, humidity, and gas concentration data within the storage. The collected data were comprehensively analyzed to determine the dynamic adjustment ratio of oxygen and carbon dioxide, the optimal operating parameters of the variable frequency refrigeration unit and the guide fan, the humidity correction parameters, and to predict the remaining shelf life of blueberries and the trend of surface microbial growth. Based on the dynamic adjustment ratio, the optimal operating parameters, the humidity correction parameters, the remaining shelf life, and the trend of surface microbial quantity changes, the optimal control command is generated through a preset decision model. The optimal control command is sent to the fuzzy PID controller, which drives each actuator to dynamically control the preservation parameters of the cold storage in a closed loop.

2. The intelligent preservation parameter control method for a blueberry cold storage according to claim 1, characterized in that, The process of comprehensively analyzing the collected data to determine the dynamic adjustment ratio of oxygen and carbon dioxide, the optimal operating parameters of the variable frequency refrigeration unit and the guide fan, the humidity correction parameters, and predicting the remaining shelf life of blueberries and the changing trend of surface microbial counts includes: The respiratory intensity data and ethylene release data are analyzed based on a machine learning model to determine the physiological stage of the blueberry and to determine the dynamic regulation ratio of oxygen and carbon dioxide based on the physiological stage. By combining computational fluid dynamics simulation models to analyze the temperature, humidity and gas concentration data in the warehouse, the optimal operating parameters of the variable frequency refrigeration unit and the guide fan are determined. The surface image data is processed by an image analysis algorithm to quantify the integrity of the blueberry bloom, and humidity correction parameters are determined based on the bloom integrity. The temperature, humidity and gas concentration data in the warehouse, as well as the packaging microenvironment data, are input into a preset shelf-life prediction model to predict the remaining shelf life of the blueberries. A microbial prediction model is constructed based on the temperature, humidity, and gas concentration data in the storage room and the packaging microenvironment data. The microbial prediction model is used to predict the trend of microbial quantity changes on the surface of blueberries.

3. The intelligent preservation parameter control method for a blueberry cold storage according to claim 2, characterized in that, The analysis of temperature, humidity, and gas concentration data within the storage facility using a computational fluid dynamics simulation model to determine the optimal operating parameters for the variable frequency refrigeration unit and the guide fan includes: The temperature, humidity and gas concentration data inside the cold storage, the spatial structure parameters of the cold storage, and the blueberry loading data are input into the computational fluid dynamics simulation model to construct a three-dimensional simulation model of airflow organization and temperature field distribution inside the cold storage. Acquire the precooling curve data and precooling final temperature data synchronized during the blueberry precooling stage, and combine the precooling curve data and precooling final temperature data to iteratively correct the three-dimensional simulation model, and predict the dynamic change trend of the cold storage temperature field under different refrigeration power and different fan speed. Based on the predicted dynamic change trend, with the uniformity of the temperature field inside the cold storage as the core objective, and taking into account both refrigeration energy consumption and equipment operating losses, the optimal output power of the variable frequency refrigeration unit and the differentiated speed parameters of each guide fan are obtained.

4. The intelligent preservation parameter control method for a blueberry cold storage according to claim 3, characterized in that, The process of acquiring synchronized precooling curve data and precooling final temperature data during the blueberry precooling stage, and combining this data to iteratively correct the three-dimensional simulation model to predict the dynamic change trend of the cold storage temperature field under different refrigeration powers and different fan speeds includes: Acquire the precooling curve data and precooling final temperature data synchronized during the blueberry precooling stage, and extract the temperature drop characteristic data in the precooling curve data and the temperature difference characteristic data in the precooling final temperature data as correction factors. The correction factor is incorporated into the three-dimensional simulation model to dynamically calibrate the temperature conduction coefficient and fruit heat exchange coefficient of the three-dimensional simulation model, thus completing the first iterative correction of the three-dimensional simulation model. Based on the corrected three-dimensional simulation model, historical refrigeration control data of the cold storage were substituted to verify the deviation between the temperature field simulation results of the three-dimensional simulation model under different refrigeration power and different fan speed and the actual monitoring results. If the deviation value exceeds the preset range, the airflow organization calculation parameters of the three-dimensional simulation model are corrected a second time according to the deviation value to achieve iterative convergence of the model. Based on the three-dimensional simulation model that has completed the iterative correction, the dynamic change trend of the cold storage temperature field under different refrigeration power and different fan speed is predicted.

5. The intelligent preservation parameter control method for a blueberry cold storage according to claim 2, characterized in that, The process of processing the surface image data using an image analysis algorithm to quantify the integrity of the blueberry bloom and determining humidity correction parameters based on the bloom integrity includes: Based on the surface image data, the peel area of ​​a single fruit is extracted using an image analysis algorithm, and interference from the fruit stem, background, and packaging texture is removed. Based on the diffuse reflection characteristics of fruit powder to light of a specific wavelength, a fruit powder feature vector is constructed using gray-level co-occurrence matrix and texture entropy algorithm, and the fruit powder shedding rate of the fruit peel area is calculated to quantify the integrity of blueberry fruit powder. If the fruit powder shedding rate exceeds the preset humidity threshold, the humidity baseline value is lowered to obtain the humidity correction parameter; After humidity correction, the image analysis algorithm continuously tracks the trend of the fruit powder shedding rate. If the fruit powder shedding rate falls back to within the preset humidity threshold, the humidification device is driven to restore the baseline humidity.

6. The intelligent preservation parameter control method for a blueberry cold storage according to claim 5, characterized in that, Based on the diffuse reflection characteristics of fruit powder to specific wavelengths of light, a fruit powder feature vector is constructed using a gray-level co-occurrence matrix and texture entropy algorithm. The fruit powder shedding rate of the peel region is then calculated to quantify the integrity of blueberry fruit powder, including: A specific near-infrared band with the most significant diffuse reflection characteristics of fruit powder was selected as the image acquisition light source. The extracted fruit peel area was subjected to directional imaging to obtain the characteristic band grayscale image of the fruit peel area. Based on the grayscale image of the feature band, multiple pixel sub-regions are divided and grayscale co-occurrence matrices of each sub-region are constructed to extract texture feature parameters. Based on the texture feature parameters, the texture entropy of each sub-region is calculated, and the texture entropy difference threshold between the powder-covered area and the powder-free area is combined to construct the powder feature vector. Based on the fruit bloom feature vector, the fruit bloom shedding rate of the fruit peel region is calculated to quantify the integrity of the blueberry fruit bloom.

7. The intelligent preservation parameter control method for a blueberry cold storage according to claim 2, characterized in that, The step of inputting the temperature, humidity, and gas concentration data of the storage area, as well as the packaging microenvironment data, into a preset shelf-life prediction model to predict the remaining shelf life of blueberries includes: A shelf-life prediction model integrating the XGBoost algorithm and time series analysis was constructed. The initial quality data and varietal characteristic data of the blueberries were used as the basic input dimensions, and the temperature, humidity and gas concentration data in the warehouse and the packaging microenvironment data were used as dynamic input dimensions. At the same time, the results of the blueberry physiological stage judgment were introduced as the correlation correction dimension. The shelf life prediction model was trained and validated using historical data from multiple batches of blueberries throughout their entire shelf life cycle, and the model hyperparameters were optimized using a grid search method. The real-time collected data on temperature, humidity, and gas concentration in the warehouse, as well as the packaging microenvironment data, are input into the trained shelf-life prediction model. The shelf-life prediction model outputs the remaining shelf-life prediction value for the corresponding batch of blueberries based on time series trend extrapolation.

8. The intelligent preservation parameter control method for a blueberry cold storage according to claim 1, characterized in that, The process of detecting and acquiring initial quality data and varietal characteristic data of blueberries upon their arrival at the warehouse, and generating an initial preservation parameter plan for the corresponding batch to determine the initial control benchmark includes: When blueberries are received into the warehouse, a full-area non-destructive scanning inspection is performed to extract the initial quality data of the blueberries, and the initial quality data is then standardized and graded. Blueberry varietal characteristic data are collected using RFID radio frequency identification technology. Based on the varietal characteristic data, climacteric and non-climacteric varieties are distinguished. Combined with the grading results of the initial quality data, a preset preservation parameter database is retrieved. The parameters in the preservation parameter database are individually adapted and adjusted to generate an initial preservation parameter plan for the corresponding batch of blueberries, so as to determine the initial control benchmark.

9. An intelligent preservation parameter control device for a blueberry cold storage, characterized in that, include: The parameter baseline determination module is used to detect and obtain the initial quality data and variety characteristic data of blueberries when they are put into storage, and generate the initial preservation parameter plan for the corresponding batch to determine the initial control baseline. The data acquisition module is used to collect real-time data on blueberry respiration intensity, ethylene release, surface image data, packaging microenvironment data, and temperature, humidity and gas concentration data inside the cold storage through a multi-source sensor array arranged inside the cold storage. The data analysis module is used to comprehensively analyze various types of data collected, determine the dynamic adjustment ratio of oxygen and carbon dioxide, the optimal operating parameters of the variable frequency refrigeration unit and the guide fan, the humidity correction parameters, and predict the remaining shelf life of blueberries and the trend of surface microbial quantity changes. The control command generation module is used to generate optimal control commands based on the dynamic adjustment ratio, the optimal operating parameters, the humidity correction parameters, the remaining shelf life, and the trend of surface microbial quantity changes through a preset decision model. The control command execution module is used to send the optimal control command to the fuzzy PID controller, and drive each actuator through the fuzzy PID controller to dynamically control the preservation parameters of the cold storage in a closed loop.

Images