Intelligent temperature control method and system for cold storage

Abstract

The application discloses an intelligent temperature control method and system for cold storage, and relates to the technical field of intelligent cold chain control, and is suitable for the storage management of fruits such as durians and other fruits that need to be controlled. The method collects cold storage environment parameters and fruit state parameters through a multi-modal sensor network, identifies the ripening stage of fruits in each region of the cold storage, constructs a ripening state distribution map, combines target storage strategies, a rule base and dynamic data processing, determines a unified control stage, and generates a temperature control trajectory of the whole cold storage including the control target temperature and humidity, the control duration and the actual temperature change rate. The temperature control trajectory describes the temperature and humidity change process with time in the form of a temperature change path function corresponding to each stage, and is dynamically corrected combined with real-time sensing data to realize closed-loop control. The method can improve the accuracy and flexibility of cold storage temperature control, improve the ripening consistency and terminal flavor quality of fruits, and significantly improve the intelligent level and effect of storage management.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent cold chain control, and in particular to an intelligent temperature variable control method and system for cold storage. Background Art

[0002] Tropical fruits like durian have a strong tendency to ripen during storage. Their quality and flavor at the end of the fruit's life are significantly influenced not only by temperature control but also by the dynamic changes in the fruit's ripening stages. In traditional cold storage, environmental parameters such as temperature and humidity are often controlled using constant or simple preset curves. This lacks the ability to perceive and respond to the fruit's actual ripening state, making it difficult to effectively cover multiple critical storage stages, from pre-cooling, preservation, ripening induction, to final preservation.

[0003] Although some existing studies have attempted to introduce image acquisition or gas detection methods to monitor the status of fruits, a systematic fusion recognition and temperature control linkage mechanism has not yet been formed. Especially when faced with the uneven spatial distribution of fruit maturity status in cold storage, problems such as temperature control strategy mismatch, uneven ripening or flavor deterioration often occur, which seriously restricts the storage quality and market value of fruits.

[0004] In addition, current cold storage temperature control strategies are mostly based on empirical settings, lacking the ability to intelligently adapt to specific storage targets (such as target maturity structure, delivery time, and terminal flavor). Control strategies cannot be dynamically adjusted based on real-time monitoring data, resulting in problems such as difficulty in compressing the storage cycle, large fluctuations in flavor output, and inaccurate control of the maturation rate. Summary of the Invention

[0005] In response to the above problems, the present invention proposes an intelligent variable temperature control method and system for cold storage, which integrates the intelligent variable temperature control of fruit maturity stage identification and multi-stage temperature and humidity regulation to achieve precise and flexible environmental control of the entire process from pre-cooling to terminal preservation, thereby improving the uniformity, stability and terminal flavor performance of fruit storage.

[0006] To achieve the above objectives, in a first aspect, the present invention provides an intelligent temperature control method for cold storage, comprising the following steps:

[0007] Through the deployment of a multimodal sensor network, the environmental parameters and fruit status parameters in the cold storage are collected in real time. The environmental parameters include temperature, humidity and gas concentration, and the fruit status parameters include appearance image features and volatile substance concentration.

[0008] Identify the maturity stages of fruits in each area of ​​the cold storage and construct a maturity distribution map based on the collected fruit status data. This map is used to reflect the maturity stage distribution of fruits in the spatial dimension and the proportion of each maturity stage.

[0009] According to the maturity state distribution map and the target storage strategy, combined with the preset rule base and dynamic data processing, the unified control stage that the current cold storage should execute is determined, and the temperature control trajectory of the entire cold storage is generated according to the unified control stage;

[0010] Dynamically adjust cold storage environmental parameters according to the temperature control trajectory;

[0011] According to the continuous monitoring results of sensors in the multimodal sensor network, the current environmental parameters are dynamically compared with the target values ​​of the temperature control trajectory, the subsequent temperature control trajectory is corrected in real time, and the cold storage environmental parameters are adjusted synchronously to achieve closed-loop control.

[0012] In a second aspect, the present invention provides an intelligent temperature control system for cold storage, comprising:

[0013] A multimodal sensor network module is used to collect real-time environmental parameters and fruit status parameters in the cold storage, the environmental parameters including temperature, humidity and gas concentration, and the fruit status parameters including appearance image features and volatile substance concentration;

[0014] The fusion recognition module is used to identify the maturity stage of fruits in each area of ​​the cold storage based on the collected fruit status data, and to construct a maturity status distribution map to reflect the maturity stage distribution of fruits in the spatial dimension and the proportion of each maturity stage;

[0015] The temperature control trajectory generation module is used to determine the unified control stage that the current cold storage should execute based on the maturity state distribution map and the target storage strategy, combined with the preset rule library and dynamic data processing, and generate the temperature control trajectory of the entire cold storage according to the unified control stage;

[0016] An execution control module is used to dynamically adjust the environmental parameters in the cold storage according to the temperature control trajectory;

[0017] The closed-loop feedback module is used to dynamically compare the current environmental parameters with the target value of the temperature control trajectory based on the continuous monitoring results of the sensor, correct the subsequent temperature control trajectory in real time, and synchronously adjust the cold storage environmental parameters to achieve closed-loop control.

[0018] One or more technical solutions provided in the present invention have at least the following technical effects or advantages: Targeting the staged temperature control management needs of fruits such as durian that need to be ripened, by collecting environmental parameters and fruit state parameters, identifying the fruit ripening stage, combining storage strategies and rule bases to generate the overall temperature control trajectory of the cold storage, and continuously correcting based on closed-loop feedback to achieve intelligent temperature and humidity control throughout the entire process. Specifically, a multimodal sensor network is used to collect temperature, humidity, gas concentration and fruit status parameters in the cold storage, comprehensively perceive environmental and fruit changes, and improve storage status perception capabilities; by extracting images and gas features, the spatial distribution of fruit maturity stages in the cold storage is recognized, supporting the subsequent formulation of precise control strategies; based on the maturity state distribution map and target storage strategy, the matching rule library and dynamic data processing are used to determine the unified control stage that the cold storage should currently execute, ensuring that the temperature control strategy meets the fruit maturity requirements; calling the environmental parameter template and the target storage strategy, combined with the actual state of the previous stage, calculate the control target temperature and humidity, control duration and actual temperature change rate of each stage, construct a temperature change path function corresponding to each stage that describes the change of temperature and humidity over time, and output the overall temperature change control trajectory of the cold storage, realizing dynamic management of temperature and humidity throughout the entire process from pre-cooling, preservation, ripening control to terminal maintenance; based on the real-time monitoring data of the sensor, the current environmental parameters and trajectory targets are dynamically compared, the subsequent temperature and humidity control path is corrected, and a closed-loop feedback mechanism is constructed to improve the real-time and stability of environmental control. Compared with the existing technology, a multimodal fruit state recognition mechanism is introduced to realize dynamic and spatial perception of the maturity stage; the fusion of recognition results, target strategies and rule bases are used for intelligent decision-making and stage control judgment; a temperature change path function corresponding to each stage is proposed based on the control target temperature and humidity, control duration and actual temperature change rate to achieve flexible and precise temperature control; full-link closed-loop control from collection, identification, prediction to execution and feedback is realized to improve storage efficiency and fruit quality stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0020] Figure 1 A flow chart of an intelligent temperature control method for cold storage provided in an embodiment of the present application;

[0021] Figure 2 A schematic diagram of the structure of an intelligent temperature control system for cold storage provided in an embodiment of the present application;

[0022] Figure 3A call flow chart between the preset rule library and the temperature control trajectory generation provided in the embodiment of the present application.

[0023] Explanation of reference numerals: multimodal sensor network module 11 , fusion recognition module 12 , variable temperature control trajectory generation module 13 , execution control module 14 , closed-loop feedback module 15 . DETAILED DESCRIPTION

[0024] The present invention introduces multimodal perception, fusion recognition and predictive control mechanisms, and proposes an intelligent variable temperature control method and system for cold storage, which solves the problems of traditional cold storage in insufficient recognition of fruit maturity status, lack of dynamic adaptability of temperature control strategies, and unstable ripening effect and flavor quality.

[0025] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only some of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0026] It should be noted that the terms "first", "second", etc. in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] Example 1, as Figure 1 As shown, the present application provides an intelligent temperature control method for cold storage, which is characterized by comprising the following steps:

[0028] S1. Through the deployment of a multimodal sensor network, real-time collection of cold storage environmental parameters and fruit status parameters, the environmental parameters include temperature, humidity and gas concentration, the fruit status parameters include appearance image features and volatile substance concentration;

[0029] Specifically, in this embodiment, a multimodal sensor network is deployed in the cold storage to collect environmental parameters and fruit status parameters in real time. The sensor network includes the following types of sensor nodes:

[0030] Environmental parameter collection unit:

[0031] Temperature sensors: Use high-precision thermocouples or digital temperature probes, distributed and installed in key areas of the cold storage, such as air inlets, return air outlets, the central fruit stacking area, and the air outlet duct, to monitor the temperature distribution in the space;

[0032] Humidity sensor: A capacitive humidity probe is used, which is placed in the same position as the temperature sensor to monitor local humidity fluctuations;

[0033] Gas concentration sensor: Electrochemical or metal oxide gas sensors are used to detect gas components closely related to fruit ripening, such as ethylene, carbon dioxide and oxygen, to reflect changes in fruit respiration intensity and ripening status.

[0034] Fruit status collection unit:

[0035] Image acquisition equipment: Install a high-definition visible light camera or industrial camera above the fruit storage area, and use it with an LED fill light device to regularly capture images of the fruit surface to extract appearance features such as fruit color, glossiness, and spots.

[0036] Volatile substance concentration detector: Install gas sampling devices and MEMS gas sensors in some typical fruit concentration areas to monitor VOCs released by fruits, such as esters, alcohols or sulfides, to reflect the changing trends of fruit flavor.

[0037] Communication and data transmission systems:

[0038] Each sensor node communicates with the edge gateway or central control system through RS485 wired network or LoRa / Wi-Fi wireless network;

[0039] The data collection frequency can be configured according to the cold storage operation settings. In cases where key parameters fluctuate greatly, the collection frequency can be appropriately increased to ensure the continuity and stability of the data.

[0040] For example, during a cold storage operation cycle, collected data showed that the ethylene concentration in the fruit pile area in the second row and third column increased rapidly from 6 ppm to 15 ppm in a short period of time, exceeding the preset fluctuation threshold. At the same time, the imaging equipment in this area captured the surface color of some fruits changing from green to yellow. Combined with the detected local spot characteristics, this reflected a rapid increase in maturity. Based on this comprehensive judgment, the fruits in this area entered the ripening stage. The subsequent control process uses this result as input.

[0041] Through the coordinated deployment and real-time data collection of the above-mentioned multimodal sensor network, the system can obtain comprehensive, continuous, and quantifiable data on the cold storage environment and fruit ripening status, providing high-quality data support for subsequent ripening identification, control strategy generation, and trajectory execution.

[0042] S2. Identify the maturity stage of the fruit in each area of ​​the cold storage and, based on the collected fruit status data, construct a maturity status distribution map. This map reflects the spatial distribution of the fruit's maturity stages and the proportion of each maturity stage.

[0043] Specifically, based on the collected fruit appearance images and volatile substance concentration data, image features and gas features are extracted in sequence. The image features include the color, gloss, spot distribution, etc. of the fruit surface, and the gas features include the concentration change trend of key volatile substances closely related to fruit maturity. Through feature fusion and classification judgment, the current maturity stage of the fruit in each fruit area is identified; the recognition results are spatially annotated in units of regions, and then a maturity state map reflecting the distribution of fruit maturity stages inside the cold storage is constructed.

[0044] The maturity state distribution map is a spatial distribution model constructed by this application for the maturity of fruits in cold storage, which is used to reflect the distribution of the maturity stages of fruits in each area of ​​the cold storage and the changes in their proportions. The map is based on the fruit state data collected by multimodal sensors, integrates image features and gas concentration information, classifies the maturity stages of fruits, and maps them in combination with spatial coordinate information to form a comprehensive visual expression of the maturity state. The maturity state distribution map can be expressed in the form of a two-dimensional table using a matrix of the proportion of area × maturity stage, or it can be expressed in a graphical form using a heat map, a bar chart, or a regional distribution diagram, or it can be expressed in the form of a dynamic curve or a dynamic map combined with time evolution.

[0045] In order to facilitate the textual description and understanding of the maturity distribution map, this implementation process classifies the fruit maturity stage into four stages, and uses a two-dimensional table to show the proportion of fruits at each maturity stage in different areas, so as to achieve a quantitative description of the differences in maturity progress in different areas. For example, at a certain time point in the operation of the cold storage, the results of the identified maturity state can be used to construct a maturity distribution map as shown in Table 1.

[0046] Table 1 shows the distribution of mature states.

[0047]

[0048] It is worth noting that in actual application, the division of fruit maturity stages is not limited to the four stages shown in the maturity state distribution map constructed in this implementation process.

[0049] Furthermore, in this implementation, the identification method for the mature stage includes:

[0050] A convolutional neural network is used to extract visual features closely related to the ripening process from the fruit surface image, including color features: such as the color offset value of the transition from green to yellow, texture features: texture statistics such as skin roughness and detail level changes, and shape features: structural contour information such as local spots, cracks, and changes in fruit edge shape. The collected visual features are then subjected to Gaussian filtering to remove image noise, and the Z-Score normalization method is used to adjust the feature values ​​to a uniform scale range. Canny edge detection is used to extract the contour information of the fruit in the image to achieve the purposes of denoising, normalization, and edge detection. Finally, encoding is performed to generate an image feature vector representing the visual state of the fruit.

[0051] The data collected by the gas concentration sensor is processed to extract key time series features related to the fruit ripening state. Specifically, the concentration change trends of gases such as ethylene, carbon dioxide, and oxygen are tracked. The sliding window mechanism is used to process the time series and calculate indicators such as concentration growth rate, fluctuation frequency, and peak position. These indicators reflect the dynamic indicators of the physiological process of fruit ripening. Finally, a set of gas perception feature vectors that can be used to represent the physiological process of ripening are encoded and generated.

[0052] The image feature vector and the gas sensing feature vector are concatenated in the feature dimension to form a unified fused feature vector. This fused feature vector is then classified and learned using a lightweight convolutional neural network algorithm. The feature patterns of different maturity stages are learned based on historical data to identify the maturity stage of fruit in each area of ​​the cold storage. Finally, the maturity stage of each fruit area is obtained and used to construct a maturity state distribution map. The maturity stages of the fruit are divided into the pre-cooling period, the fresh-keeping period, the ripening induction period, and the terminal fresh-keeping period.

[0053] S3. According to the maturity state distribution map and target storage strategy, combined with the preset rule base and dynamic data processing, determine the unified control stage that the current cold storage should execute, and generate the temperature control trajectory of the entire cold storage according to the unified control stage. The calling process between the preset rule base and the temperature control trajectory generation is as follows: Figure 3 As shown;

[0054] Specifically, based on the aforementioned ripening state distribution map, the current ripening stage structure of the fruits in each area of ​​the cold storage is identified, including the proportion of each stage in spatial distribution, change trends and regional distribution patterns. Combined with the set target storage strategy, such as delaying ripening, promoting ripening or maintaining terminal flavor, the control objectives are further comprehensively judged. The current regulation stage of the cold storage as a whole is further judged. This judgment process relies on the preset rule library and dynamic data processing. After completing the above judgment, the temperature control trajectory of the cold storage as a whole is generated according to the unified regulation stage that should be executed at present. The temperature control trajectory of the cold storage as a whole describes the temperature and humidity control targets that the cold storage needs to implement in this stage. The evolution path of the control variables is usually given in the form of a time series, which has stage matching, gradual change and feasibility, and can be used to guide subsequent temperature control regulation behaviors.

[0055] Furthermore, in this implementation process, in order to achieve orderly regulation of the fruit ripening structure in the cold storage, a target storage strategy is constructed to guide the determination of the unified regulation stage and the generation of the variable temperature control trajectory of the cold storage as a whole. The target storage strategy includes: target storage cycle, expected distribution of fruit ripening stages, terminal flavor preference requirements and outbound time setting. By clarifying the regulation objectives during the operation of the cold storage and the final quality orientation of the fruit, it provides a strategic parameter basis for the subsequent temperature and humidity control path. Among them, the target storage cycle: indicates the planned duration of the fruit from storage to delivery; the expected ripening stage distribution structure: clarifies the proportion of fruits of each ripening stage that should be achieved at different time points in the storage stage; the terminal maturity preference or outbound quality requirement: specifies the maturity state that the fruit should reach when it is shipped out; the outbound time setting: specifically points to the planned outbound time point of a certain storage batch, so as to regulate the variable temperature control trajectory and the ripening process to keep pace;

[0056] Furthermore, in this implementation process, in order to realize the standardization of stage judgment logic and control behavior in the fruit storage regulation process, a preset rule base was constructed. The preset rule base includes environmental parameter templates corresponding to different fruit maturity stages and rules for determining the maturity stage of the fruit, among which,

[0057] The environmental parameter templates corresponding to different fruit maturity stages are used to provide parameter basis for the generation of variable temperature control trajectories corresponding to the unified regulation stage, and to define the control boundaries in the process of generating variable temperature control trajectories. The environmental parameter templates include template target temperature, template target humidity, recommended duration and limited temperature change rate for each stage. Template target temperature: indicates the optimal temperature point to be maintained in this stage, which is used to guide the benchmark value setting of the temperature and humidity control path; template target humidity: the environmental humidity target that is conducive to maintaining fruit quality in the corresponding stage; recommended duration: indicates the recommended duration of this stage, which is used to control the time span of the trajectory stage; limited temperature change rate: is used to constrain the maximum change speed of temperature adjustment to prevent sudden changes in the environment from causing fruit stress or abnormal maturity. Table 2 provides an example of the environmental parameter templates for each maturity stage in a typical durian cold storage, as follows:

[0058] Table 2 is a typical template of environmental parameters for each maturity stage of durian cold storage

[0059]

[0060] By setting the above templates, after the current stage is identified in the ripening state distribution map, the corresponding template can be directly called to provide clear parameter targets and boundary constraints for the generation of the temperature control trajectory for that stage, thereby ensuring that the temperature control trajectory changes both meet the physiological needs of the fruit and avoid the loss of ripening quality caused by control fluctuations.

[0061] The rules for fruit maturity stages are used to identify the fruit status in different areas. The rules are based on image features, gas concentrations, and dynamic multi-dimensional fruit status parameters of the environment. Specifically, they include:

[0062] a) Rules based on image features: Using image acquisition equipment, the system acquires image data of the fruit surface. The system analyzes the visual features contained in the image and extracts key changes related to fruit ripening, including color changes, such as the gradual change from green to yellow or brown; surface cracks, such as the appearance of small vertical and horizontal cracks or openings; and lesions and mildew, such as dark spots or mildew. Based on these image features, common pattern recognition algorithms such as support vector machines (SVM) and convolutional neural networks (CNN) are used to compare the image feature vectors with preset thresholds for each maturity stage to achieve a preliminary determination of the fruit's maturity stage.

[0063] b) Gas concentration-based rules: By monitoring the concentrations and changing trends of key gases associated with fruit ripening and combining them with the characteristic threshold ranges corresponding to each ripening stage, the ripening stage of the fruit can be determined. The changing patterns of gas concentrations at different ripening stages have typical characteristics: in the early stage of ripening, ethylene concentrations rise rapidly; during the peak respiration period, carbon dioxide concentrations rise significantly and oxygen concentrations fall; during the fresh-keeping period, gas concentrations remain at a low level of fluctuation.

[0064] c) Rules based on environmental dynamics: Combined with the temperature and humidity change trends in the cold storage, analyze the impact of the environment on the ripening speed of the fruit, and adjust the judgment point of the ripening stage accordingly.

[0065] Since temperature and humidity are important external factors affecting the ripening process of fruit, relying solely on image features or gas data is prone to timing errors, such as judging the ripening stage too early or too late, which affects the adaptability and accuracy of the control strategy. Therefore, this rule analyzes the time series data of temperature and humidity in the cold storage, extracts key indicators such as the temperature rise rate, the duration of high temperature, and the drastic fluctuation of humidity, and determines whether the current environmental conditions are likely to accelerate or delay fruit ripening. If a typical temperature-inducing ripening trend is identified, the time point for the fruit to enter the ripening or terminal preservation stage is advanced; if it is under preservation conditions such as constant temperature and low humidity, the time point of stage switching is delayed. By introducing environmental dynamic characteristics and making comprehensive corrections with the fruit status judgment results, the ripening recognition deviation is effectively reduced, and the timeliness and matching degree of ripening stage recognition are improved.

[0066] d) Rule priority and condition weight setting mechanism: When there are multiple rule matching conflicts or the fruit status recognition result is unclear, a comprehensive judgment is made according to the preset priority and weight coefficient.

[0067] During the fruit ripening stage recognition process, multiple parameter dimensions such as image features, gas concentration, and environmental dynamics may simultaneously give different stage judgment results, resulting in conflicting results or ambiguous states. To improve the stability and consistency of judgment, this implementation process introduces a rule priority and condition weight setting mechanism. The preset priority setting logic is as follows: when identifying the ripening period, the gas concentration rule has the highest priority; when identifying the terminal fresh-keeping period, the image feature rule has a higher priority; when judging the pre-cooling period and the fresh-keeping period, the environmental dynamics rule has priority. If there is a conflict between gas concentration and image features, the main rule is selected according to the preset priority corresponding to the stage, and the environmental dynamics rule is referenced for correction. The condition weight setting logic is as follows: when identifying the ripening period, the gas concentration rule has the highest weight; when identifying the terminal fresh-keeping period, the image feature rule has a higher weight; when identifying the pre-cooling period and the fresh-keeping period, the environmental dynamics rule has a higher weight. If the results of multiple rules are inconsistent, they are weighted and fused according to the preset weight coefficients of the current stage, and the stage with the highest confidence is output. When the data quality of a dimension is abnormal or missing, its weight ratio is temporarily reduced, and the remaining weight is proportionally distributed to other rule dimensions to ensure judgment stability.

[0068] Furthermore, in this implementation process, in order to achieve dynamic grasp of the fruit ripening process in the cold storage and forward-looking planning of the temperature control trajectory, the dynamic data processing method adopted includes the following steps:

[0069] a. Based on the fused feature vector, the time series prediction method is used to estimate the changes in the proportion of fruits at different maturity stages in different areas of the cold storage, and fit the proportion trend of each maturity stage in the future time period; specifically, the ARIMA (Autoregressive Integrated Moving Average) method is used in this implementation process to predict the proportion of future fruit maturity stages, including:

[0070] Establish time series data. The historical data includes the maturity progress of the fruit in each area of ​​the cold storage. These progresses change over time. Extract the proportion of fruit in each maturity stage at each time step from the historical data to form a time series data containing the proportion of each maturity stage.

[0071] Data preprocessing, detect and remove outliers in the time series data to ensure the stability and accuracy of the model. If the mean or variance of the data changes over time, the data can be made stationary by differencing (usually using the d parameter). p, d, and q are selected as the main parameters of the ARIMA model, where

[0072] p (autoregressive term): indicates how many time steps of data are used to predict the current value. The PACF graph is used to select the appropriate p value. The PACF graph shows the relationship between each time lag and the current value. The significant lag is selected as the p value.

[0073] d (number of differences): indicates how many times the difference needs to be made in order to make the data stable. The unit root test is used to determine whether the data needs to be differentiated and to determine the d value.

[0074] q (moving average term): represents the relationship between the current data point and the previous q residuals. The ACF plot is used to determine the q value. The ACF plot shows the relationship between the residuals and past data points. The significant lag is selected as the q value.

[0075] Using p, d, and q as parameters, the ARIMA model is trained on the time series data, and the optimal parameters are fitted according to the time series data to minimize the residual between the actual value and the predicted value. During the training process, the ARIMA model determines the optimal parameter combination by minimizing the sum of squares of the residuals.

[0076] Proportion trend prediction: Use the trained ARIMA model to calculate the changes in the proportion of fruits at different maturity stages in each area of ​​the cold storage in the future time step. Then, based on the prediction results of the time series data, the proportion change trends of different maturity stages in the future time are obtained. Through these trends, the dynamic changes of fruit maturity in the future period can be intuitively seen;

[0077] b. According to the obtained proportion change trend, the starting and ending time of each maturation stage is calculated, specifically, for each maturation stage, a threshold value of proportion change is set, and according to the set threshold value, the starting and ending time of each maturation stage is calculated, according to the change trend of the proportion, the duration of each maturation stage is calculated, according to the real-time collected information data related to the key gas concentration of fruit maturation, the calculated starting and ending time is adjusted in real time, to ensure that the calculated starting and ending time and duration of each maturation stage conforms to the actual environmental changes;

[0078] c. According to the prediction result, combining the target storage strategy and the environmental parameter template corresponding to each stage in the preset rule library, the temperature control trajectory of the whole cold storage is calculated, which includes the control target temperature, the control target humidity, the control duration and the actual temperature change rate, specifically;

[0079] Further, in the present embodiment, based on the change trend of the proportion of each region fruit maturation stage and the starting and ending time point, the maturation target parameter of the stage set in the target storage strategy is further called, and the environmental parameter template corresponding to the target stage in the preset rule library is combined to generate the temperature control trajectory, which includes,

[0080] a) After determining the uniform regulation stage that the current cold storage should execute, the maturation target of the stage set in the target storage strategy is called to obtain the corresponding control target temperature and control target humidity parameters;

[0081] b) Based on the actual monitoring obtained cold storage environment parameters in the last uniform regulation stage continued by the current stage, the actual temperature change rate and control duration required to achieve the target environment parameters are calculated by combining the upper limit of the temperature change rate and the recommended duration of the stage in the preset rule library;

[0082] c) The control target temperature and control target humidity parameters of each stage are combined with the calculated actual temperature change rate and control duration to construct the temperature change path function corresponding to each stage describing the change process of the environment parameters with time, and the temperature change path function corresponding to each stage is formed;

[0083] Wherein, the construction process of the temperature change path function corresponding to each stage is as follows:

[0084] 1) Control duration calculation under temperature and humidity change boundary constraints

[0085] In practical application, the response of fruit to temperature and humidity change rate has physiological boundary, if the temperature rises or falls too fast, it may cause maturation disorder or quality degradation. Therefore, the upper limit value of temperature control limited temperature change rate and the upper limit value of humidity control limited temperature change rate , as the control constraint parameter, is used to limit the temperature and humidity change rate in each control stage.

[0086] Assume that the target environmental parameters of a certain control stage are: the control target temperature is , the target humidity is controlled to ; and the actual parameters of the cold storage at the end of the previous stage are: the end temperature is , the end humidity is , the shortest duration required to complete the target control is calculated using the following formula:

[0087] ;

[0088] The time It is the lower limit of temperature and humidity regulation in the current stage, ensuring a stable and controllable environmental transition process and preventing drastic changes from damaging the fruit. The maximum value logic method is used here to determine the duration of regulation in this stage, ensuring that the temperature and humidity adjustment processes can be completed synchronously, avoiding the premature completion of one regulation and the failure of the other to achieve its goal, thereby enhancing the coordination of the regulation process and the accuracy of ripening state identification.

[0089] 2) Construction of temperature change path function

[0090] Once the control time interval is determined , calculate the linear rate of change per unit time based on the initial value and the target value:

[0091] , , based on this, the temperature change path function of this stage is constructed for

[0092] ;

[0093] It is used to express the change path function of temperature and humidity in the control stage. More specifically, it uses To express the temperature change path function in this regulation stage, we use To express the change path function of humidity in this control stage, Indicates the actual temperature change rate of temperature control in the current stage. Indicates the actual temperature change rate of humidity control in the current stage, where is a relative time variable within the current stage, used to describe the control trajectory of environmental parameters changing over time within this stage. This function has the following characteristics:

[0094] Monotonicity: Ensure that the direction of temperature and humidity changes is consistent with the target, without fluctuations or rebounds;

[0095] Differentiability: The path is a continuous differentiable curve, which is convenient for controller prediction and adjustment;

[0096] Parameter traceability: The source of each parameter in the function is clear and directly corresponds to the output and target parameters of the previous stage;

[0097] 3) Continuous splicing and management of temperature change path functions corresponding to each stage

[0098] In order to fully cover the entire storage process, this method constructs the temperature change path function corresponding to each stage in sequence according to the time sequence of the unified control stage. , each function is defined in a non-overlapping time interval , ensure the parameter continuity and derivative consistency of the function splicing points. Finally, by splicing the temperature change path functions corresponding to each stage of each control stage along the time axis, a continuous and controllable temperature change control trajectory of the entire cold storage is constructed, ensuring that the temperature and humidity changes at any moment are in line with the preset strategy and do not exceed the physiological limits. At the same time, a smooth and timely control response is achieved to avoid sudden changes in environmental parameters.

[0099] S4. Dynamically adjust cold storage environment parameters according to the temperature control trajectory;

[0100] Specifically, after completing the construction of the variable temperature control trajectory of the cold storage, the control execution phase is entered, that is, the temperature and humidity of the cold storage are dynamically adjusted according to the path of the environmental parameters set in the constructed variable temperature control trajectory over time. The specific adjustments are as follows:

[0101] First, in the current regulation stage, according to the relative time variable Extract the target temperature given by the temperature change path function corresponding to each stage and target humidity , used to guide the adjustment of environmental parameters at this moment;

[0102] Then, according to the extracted target values, the actual environmental parameters in the cold storage are adjusted to make the current temperature and humidity as close as possible to the target values. and ,This regulation can be achieved by controlling the working states of the execution equipment such as refrigeration, heating, humidification, dehumidification, etc.;

[0103] During the entire adjustment process, the temperature and humidity data inside the cold storage are continuously collected and monitored, and compared with the set value of the variable temperature control trajectory to verify the control accuracy and support the evaluation of the adjustment effect in the subsequent stages;

[0104] In addition, during the switching process between different ripening stages, the regulation rhythm is adjusted in a timely manner according to the temperature change path function corresponding to each stage set in the temperature change control trajectory to avoid environmental shock to the fruit due to excessive temperature increase or decrease, thereby ensuring the stability and softness of the entire ripening process.

[0105] S5. According to the continuous monitoring results of the sensors in the multi-modal sensor network, the current environmental parameters are dynamically compared with the target values of the variable temperature control trajectory, the subsequent variable temperature control trajectory is corrected in real time, and the cold storage environment parameters are adjusted synchronously to realize closed-loop control.

[0106] Specifically, according to the time progress in the current regulation stage, the target temperature and humidity value at the corresponding time point is obtained from the preset variable temperature path function corresponding to each stage as the control reference that should be reached at the time point; the real-time monitored environmental parameters are compared with the target value, if the deviation exceeds the preset tolerance threshold, the variable temperature control trajectory correction process is triggered, the current maturity state distribution, the target storage strategy and the boundary conditions in the rule base are combined to dynamically correct the path segment that has not been executed, and the subsequent variable temperature control trajectory is updated; the corrected variable temperature control trajectory takes effect immediately and drives the execution of the corresponding regulation instruction, so that the environmental parameter adjustment always keeps consistent with the target change path, thereby ensuring that the regulation process has self-adaptive adjustment capability and realizing closed-loop iteration of temperature and humidity control.

[0107] Based on the above intelligent variable temperature control method for cold storage, in order to further improve the regional adaptability of cold storage temperature and humidity regulation, a local fine-tuning mechanism is introduced on the basis of the overall variable temperature control trajectory.

[0108] S6. Based on the real-time identification of potential deviation and the response adjustment through the regularization strategy, the coordination and accuracy of the variable temperature process in the spatial distribution level are ensured.

[0109] Further, the local fine-tuning mechanism includes:

[0110] a) Real-time acquisition of environmental parameter data of each spatial region of the cold storage, and difference calculation of the current value of the region and the target value that should be reached in the predetermined variable temperature control trajectory to obtain the deviation of each region : wherein,

[0111] The temperature deviation value is ,

[0112] The humidity deviation value is ,

[0113] represents the region number, is the relative time variable in the current stage;

[0114] b) Fine-tuning strategy rule matching and calling

[0115] In order to avoid misadjustment or overadjustment, the local fine-tuning is not directly based on the deviation, but is combined with the maturity stage state of the fruits in the region to jointly determine whether to execute the fine-tuning, and the specific process includes: first, judging the maturity stage to which the fruits in the current region belong; then according to the current deviation 、 and the sensitivity tolerance range of the mature stage; then match the set fine-tuning strategy rules in the rule base; if the match is successful, call the corresponding fine-tuning rule for execution;

[0116] c) Local dynamic adjustment of control target parameters

[0117] According to the selected fine-tuning strategy, the control target parameters of the area, such as temperature and humidity, are slightly dynamically adjusted. The principle of the adjustment process is: do not break the continuity of the original temperature change path function in the current control stage, that is, The structure remains unchanged, and the target point is fine-tuned; the preset control boundaries such as the upper limit of the temperature change rate and the minimum control cycle are not exceeded; it is executed at the local controller level and does not interfere with the temperature change control trajectory of other areas or the entire cold storage;

[0118] Deviation persistence trend judgment and path segment update

[0119] If the deviation continues to exceed the preset range, the subsequent path segments of the area will be locally updated according to the real-time monitoring trend to enhance the adaptability of the overall temperature control trajectory of the cold storage to spatial heterogeneity. The specific local update is to analyze the deviation change trend of the area in the future period in real time; in the next control cycle, the temperature change path function of the area will be updated. , subsequent segments are updated to achieve predictive fine-tuning; to ensure that the new path segment is continuous with the original variable temperature control trajectory at the segment junction, a spline function with continuous first derivative can be used for connection;

[0120] By introducing a local fine-tuning mechanism, the present invention achieves refined and personalized adjustment of temperature and humidity control at the regional level while maintaining the overall temperature control trajectory framework of the cold storage unchanged. This mechanism effectively compensates for the regulation deviation caused by spatial heterogeneity; enhances the control strategy's ability to respond to real-time disturbances and dynamic changes in fruit ripening; and improves the consistency of fruit ripening and the stability of flavor quality.

[0121] Example 2, based on the same inventive concept as the intelligent temperature control method for cold storage in the above embodiment, Figure 2 As shown, the present application provides an intelligent temperature control system for cold storage. The system and method embodiments in the present application are based on the same inventive concept. The system includes:

[0122] The multimodal sensor network module 11 is used to collect environmental parameters and fruit status parameters in the cold storage in real time. The environmental parameters include temperature, humidity and gas concentration, and the fruit status parameters include appearance image features and volatile substance concentration;

[0123] The fusion recognition module 12 is used to identify the maturity stage of the fruits in each area of ​​the cold storage based on the collected fruit state data, and to construct a maturity state distribution map to reflect the maturity stage distribution of the fruits in the spatial dimension and the proportion of each maturity stage;

[0124] The temperature control trajectory generation module 13 is used to determine the unified control stage that the current cold storage should execute based on the maturity state distribution map and the target storage strategy, combined with the preset rule library and dynamic data processing, and generate the temperature control trajectory of the entire cold storage according to the unified control stage;

[0125] An execution control module 14 is used to dynamically adjust the environmental parameters in the cold storage according to the temperature control trajectory;

[0126] The closed-loop feedback module 15 is used to dynamically compare the current environmental parameters with the target value of the temperature control trajectory based on the continuous monitoring results of the sensor, correct the subsequent temperature control trajectory in real time and synchronously adjust the cold storage environmental parameters to achieve closed-loop control.

[0127] The intelligent variable temperature control system for cold storage provided by the embodiment of the present invention can execute the intelligent variable temperature control method for cold storage provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0128] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other and are not used to limit the scope of protection of the present invention.

[0129] The above specific embodiments do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application shall be included within the scope of protection of this application.

Claims

1. An intelligent temperature control method for cold storage, characterized in that: The following steps are included: Through the deployment of a multimodal sensor network, the environmental parameters and fruit status parameters in the cold storage are collected in real time. The environmental parameters include temperature, humidity and gas concentration, and the fruit status parameters include appearance image features and volatile substance concentration. Identify the maturity stages of fruits in each area of ​​the cold storage and construct a maturity distribution map based on the collected fruit status data. This map is used to reflect the maturity stage distribution of fruits in the spatial dimension and the proportion of each maturity stage. Based on the maturity state distribution map and the target storage strategy, combined with a preset rule base and dynamic data processing, the unified control stage that the current cold storage should execute is determined, and a temperature control trajectory of the entire cold storage is generated based on the unified control stage. The generation process of the temperature control trajectory includes: a) After determining the unified control stage that the cold storage should currently execute, call the maturity target for that stage set in the target storage strategy to obtain the corresponding control target temperature and control target humidity parameters; b) Based on the actual monitored cold storage environmental parameters in the previous unified control phase that follows the current phase, combined with the upper limit of the temperature change rate and the duration range allowed in the preset rule base for that phase, the actual temperature change rate and control duration required to achieve the target environmental parameters are calculated; c) Combine the control target temperature and control target humidity parameters of each stage with the calculated actual temperature change rate and control duration to construct a segmented temperature change path function that describes the change process of environmental parameters over time, and form the temperature change control trajectory of the entire cold storage. The construction process of the segmented temperature change path function includes the following algorithm steps: Set the control target temperature of the current stage , control target humidity , and determine the end temperature of the previous regulation stage , end humidity , combined with the maximum allowable temperature change rate given in the rule base , calculate the control duration of the current stage : ； To control the duration For the time interval , t is the relative time variable in the current stage, and the temperature and humidity path function of the current stage is constructed: ； in, , respectively, are the actual control temperature change rates of temperature and humidity in the current stage, and the path functions corresponding to multiple stages They are spliced ​​sequentially on the time axis to form a complete set of cold storage temperature and humidity control trajectories, which are used to guide subsequent dynamic environmental control; Dynamically adjust cold storage environmental parameters according to the temperature control trajectory; Based on the continuous monitoring results of sensors in the multimodal sensor network, the current environmental parameters are dynamically compared with the target values ​​of the variable temperature control trajectory, the subsequent temperature control trajectory is corrected in real time, and the cold storage environmental parameters are adjusted synchronously to achieve closed-loop control.

2. The control method according to claim 1, characterized in that: The target storage strategy includes: target storage period, expected fruit maturity stage distribution, terminal flavor preference requirements and delivery time setting. The target storage strategy is used to guide the determination of the unified regulation stage and provide parameter basis for the generation of the overall temperature control trajectory of the cold storage.

3. The control method according to claim 1, wherein: The method for identifying the mature stage includes: By collecting fruit appearance images, three types of visual features, namely color, texture and shape, are extracted from the fruit appearance images, and the visual features are processed to generate image feature vectors; By collecting the concentration changes of key gases related to fruit ripening, the time series features reflecting the fruit ripening process are extracted, and the time series features are processed to generate gas perception feature vectors; Based on the collected image features and gas concentration features, the data fusion method is used to combine the image features and gas concentration features to form a unified fusion feature vector. The classification algorithm is used to identify the maturity stage of fruits in each area of ​​the cold storage, and the maturity stage of each fruit area is obtained.

4. The control method according to claim 1, wherein: The preset rule base includes environmental parameter templates corresponding to different fruit maturity stages and rules for determining the maturity stage of the fruit, wherein: The environmental parameter template includes the template target temperature, template target humidity, recommended duration and limited temperature change rate of each maturity stage, which is used to provide parameter basis for generating the temperature change control trajectory corresponding to the unified regulation stage and define the control boundary during the trajectory generation process. The rules for the ripening stage of the fruit are constructed based on image features, gas concentration and environmental dynamic multi-dimensional fruit state parameters, specifically including: Image feature-based rules: By identifying the visual features of cracks, color changes, or mold spots on the fruit surface, the corresponding maturity stage is determined; Gas concentration-based rules: By monitoring the concentration and changing trends of key gases related to fruit ripening and combining the characteristic threshold ranges corresponding to each ripening stage, the ripening stage of the fruit can be determined; Rules based on environmental dynamics: Combined with the temperature and humidity change trends in the cold storage, the impact of the environment on the fruit ripening speed is analyzed, and the determination time of the ripening stage is adjusted accordingly; Rule priority and condition weight setting mechanism: When there are multiple rule matching conflicts or the fruit status recognition results are unclear, a comprehensive judgment is made according to the preset priority and weight coefficient.

5. The control method according to claim 1, characterized in that: The dynamic data processing method comprises the following steps: Based on the fused feature vectors, the time series prediction method is used to estimate the changes in the proportion of fruits at different maturity stages in different areas of the cold storage, and fit the proportion trend of each maturity stage in the future time period; According to the obtained percentage change trend, the start and end time of each maturity stage are estimated; According to the prediction results, combined with the target storage strategy and the environmental parameter templates corresponding to each stage in the preset rule library, the overall temperature change control trajectory of the cold storage is calculated. The temperature change control trajectory includes the control target temperature, control target humidity, control duration and actual temperature change rate.

6. The control method according to any one of claims 1 to 5, characterized in that: It further includes a local fine-tuning mechanism for fine-tuning the environmental control process of each area based on the overall temperature control trajectory of the cold storage according to the real-time deviation of the environmental parameters in each area and the ripening state of the fruit. The local fine-tuning mechanism includes: Based on the data collected by sensors, the deviation between the current environmental parameters of each area of ​​the cold storage and the target value of the corresponding temperature control trajectory is calculated; Based on the matching relationship between the current maturity stage of the fruit in the region and the environmental deviation, the applicable rules are selected and called from the pre-set fine-tuning strategy rules in the preset rule library to guide the fine-tuning of regional environmental control; Based on the fine-tuning strategy rules, the control target parameters of the area are adjusted to achieve fine-grained optimization of the overall temperature change trajectory by the local control path; If the deviation continues to exceed the preset range, the subsequent path segments in the area are locally updated according to the real-time monitoring trend to enhance the adaptability of the overall trajectory to spatial heterogeneity.

7. An intelligent temperature control system for cold storage, characterized in that: The steps for implementing the intelligent temperature control method for cold storage according to any one of claims 1 to 5, wherein the intelligent temperature control system for cold storage comprises: A multimodal sensor network module is used to collect real-time environmental parameters and fruit status parameters in the cold storage, the environmental parameters including temperature, humidity and gas concentration, and the fruit status parameters including appearance image features and volatile substance concentration; The fusion recognition module is used to identify the maturity stage of fruits in each area of ​​the cold storage based on the collected fruit status data, and to construct a maturity status distribution map to reflect the maturity stage distribution of fruits in the spatial dimension and the proportion of each maturity stage; The temperature control trajectory generation module is used to determine the unified control stage that the current cold storage should execute based on the maturity state distribution map and the target storage strategy, combined with the preset rule library and dynamic data processing, and generate the temperature control trajectory of the entire cold storage according to the unified control stage; An execution control module is used to dynamically adjust the environmental parameters in the cold storage according to the temperature control trajectory; The closed-loop feedback module is used to dynamically compare the current environmental parameters with the target value of the temperature control trajectory based on the continuous monitoring results of the sensor, correct the subsequent temperature control trajectory in real time, and synchronously adjust the cold storage environmental parameters to achieve closed-loop control.

Images