Kiwi fruit storage and preservation monitoring system based on dynamic controlled atmosphere and cold chain traceability

Abstract

The application provides a kiwi fruit storage and preservation monitoring system based on dynamic air conditioning and cold chain traceability, and belongs to the technical field of fruit and vegetable cold chain storage and preservation, dynamic air conditioning regulation and control and cold chain traceability monitoring, and comprises the following steps: generating a shelf level monitoring state vector by fusing transportation position, weather, road conditions, packaging state and environmental parameters, realizing intelligent scheduling of regional dynamic air conditioning regulation and control and monitoring components; adopting machine vision to synchronously extract fruit damage features and juice consumption features, constructing a carriage abnormal distribution layer, and completing accurate quality anomaly judgment and grading disposal; generating a consortium chain trusted traceability chain based on grading record data, and dynamically optimizing cold chain warehouse inventory scheduling. The application can realize shelf level accurate preservation management and control, early abnormal identification, low power consumption operation and full link trusted traceability, significantly reduces kiwi fruit cold chain loss, and improves storage quality and supply chain management efficiency.

Description

Technical Field

[0001] This invention relates to the fields of cold chain storage and preservation of fruits and vegetables, dynamic controlled atmosphere control, and cold chain traceability and monitoring technology, and particularly to a kiwifruit storage and preservation monitoring system based on dynamic controlled atmosphere control and cold chain traceability. Background Technology

[0002] Kiwifruit is a typical climacteric fruit, exhibiting vigorous postharvest respiration. It is extremely sensitive to the temperature, humidity, and gas composition of its storage environment. Temperature fluctuations, mechanical jolting, and gas imbalances during cold chain transportation can easily lead to mechanical damage, premature respiratory climacteric change, and spoilage, resulting in significant economic losses. However, existing technologies have several drawbacks: Existing technologies employ vehicle-wide unified controlled atmosphere control and environmental monitoring, which cannot adapt to the differences in initial quality, ripeness, and packaging status of kiwifruit placed on different shelves within the vehicle. This easily leads to fruit deterioration due to incompatible controlled atmosphere parameters on some shelves, and there are numerous monitoring blind spots. Existing technologies only assess preservation risks based on real-time environmental parameters, which can only identify and address issues after obvious fruit deterioration has occurred. Furthermore, they only focus on the damage and rot characteristics of the kiwifruit itself, and cannot predict the risk of chain rot caused by microbial growth in the juice, resulting in delayed anomaly identification. Existing technologies generally... The current method of using monitoring equipment that is turned on all time and in full capacity cannot dynamically schedule monitoring resources according to the level of preservation risk. The equipment consumes a lot of power and computing power, making it unsuitable for long-distance cold chain transportation scenarios of more than 48 hours. The existing cold chain traceability technology is only a step-by-step record of transportation trajectory and environmental parameters. It does not perform graded verification of abnormal data, resulting in low reliability of traceability data. Furthermore, the traceability data is not fed back into the cold chain warehouse storage scheduling of the supply chain, making it impossible to achieve full-chain preservation control. All these factors combined lead to low preservation efficiency and low traceability efficiency of kiwifruit during cold chain transportation.

[0003] Therefore, this invention proposes a kiwifruit storage and preservation monitoring system based on dynamic controlled atmosphere and cold chain traceability. Summary of the Invention

[0004] This invention provides a kiwifruit storage and preservation monitoring system based on dynamic controlled atmosphere and cold chain traceability to solve the aforementioned technical problems.

[0005] This invention provides a kiwifruit storage and preservation monitoring system based on dynamic controlled atmosphere and cold chain traceability, comprising: The vector acquisition module is used to acquire the preset kiwifruit cold chain transportation route and the real-time location of the transport driver during the transportation process according to the cold chain transportation route. Combined with the weather state vector corresponding to the real-time location, the road segment state vector of the current road segment, the pre-transport packaging state vector of the kiwifruit, and the basic storage and preservation vector of each shelf area at the current moment, the module generates the monitoring state vector corresponding to each shelf at the real-time location. The pre-transport packaging state vector includes the initial loading and storage state of the kiwifruit on each shelf in the transport vehicle, and the basic storage and preservation vector includes various environmental parameters of the corresponding shelf area. The dynamic controlled atmosphere module is used to dynamically adjust the regional controlled atmosphere adjustment unit built into the transport vehicle according to the monitoring status vector corresponding to each shelf and the real-time storage and preservation requirements of the kiwifruit on the corresponding shelf. At the same time, it dynamically controls the start and stop of several monitoring components pre-set in the transport vehicle according to the placement position of each shelf in the transport vehicle and the monitoring status vector of the corresponding shelf. The feature and anomaly analysis module is used to extract the damage features of the kiwifruit itself and the liquid erosion features of the kiwifruit juice seeping out from the images captured by each monitoring component in the active state. Combined with the basic storage and preservation vector of the corresponding shelf area at the current moment, it analyzes the quality anomalies of the kiwifruit on the corresponding shelf relative to the initial loading state, constructs an anomaly distribution layer of kiwifruit based on the monitoring range covered by all active monitoring components, and determines whether the kiwifruit meets the set replacement criteria based on the anomaly distribution layer. If the conditions are met, the location information of the cold chain warehouse closest to the real-time location is sent to the vehicle terminal, and the inspection of the abnormal batch of kiwifruit and the corresponding full-process data are recorded once when the transport vehicle arrives at the cold chain warehouse; if the conditions are not met, when all the kiwifruit monitored within the full coverage area of ​​the cold chain transport route corresponding to the same cold chain warehouse are in a condition that does not meet the set replacement criteria, the full-process data of the batch of kiwifruit is recorded a second time. The traceability chain construction module is used to generate a complete cold chain traceability chain for the transported batch of kiwifruit based on the results of primary recording, secondary recording, real-time transportation information along the cold chain transportation route, and regional controlled atmosphere regulation data. It also dynamically regulates the kiwifruit storage in each cold chain warehouse according to the full-process storage data of the cold chain traceability chain.

[0006] Preferably, the environmental parameters include: concentration, Concentration, ethylene concentration, ambient temperature, and ambient humidity.

[0007] Preferably, each monitoring component corresponds to a monitoring area of ​​a different shelf inside the transport vehicle, and the total number of monitoring components is N, where N is a positive integer and not less than the total number of shelves in the transport vehicle.

[0008] Preferably, the vector acquisition module includes: The model analysis submodule is used to input the weather state vector, road segment state vector, pre-transport packaging state vector, and basic storage and preservation vector into the corresponding pre-trained neural network model to obtain the first factor affecting the storage and preservation effect in each vector and the storage and preservation confidence of each first factor. The mapping processing submodule is used to map each predefined indicator in the monitoring status vector to the first factor under each vector to obtain a set of candidate factors for each predefined indicator. The sorting submodule is used to obtain a second factor from the candidate factor set whose storage and preservation confidence is greater than a preset confidence level. At the same time, the candidate factor set is sorted first according to the mapping correlation coefficient and sorted second according to the storage and preservation confidence level. The cutting submodule is used to perform a first forward cut on the second sorting result to obtain several third factors, taking the first element in the first sort as the first cutting point; at the same time, it performs a second forward cut on the first sorting result to obtain several fourth factors, taking the first element in the second sort as the second cutting point. The direction matching submodule is used to determine the first preservation anomaly direction based on all second factors, the second preservation anomaly direction based on all third factors, and the third preservation anomaly direction based on all fourth factors based on the factor-direction comparison table, thus forming a state anomaly subset corresponding to the predefined indicators. The relative damage determination submodule is used to determine the relative damage factors of the kiwifruit placed on the corresponding shelf based on the basic storage and preservation vector and the standard preservation vector of the pre-transportation packaging state vector, and in combination with the cumulative transportation time from the initial transportation time to the current time, the variety attributes of the kiwifruit placed on the shelf, and the post-harvest respiratory physiological characteristics. The adjustment submodule is used to adjust the maximum and minimum abnormal directions among the first, second, and third abnormal directions of preservation under the corresponding predefined indicators according to the relative damage factors of the kiwifruit, thereby forming a state abnormal subset of the corresponding predefined indicators. The vector combination submodule is used to obtain the monitoring status vector corresponding to each placed shelf based on the abnormal status subset of all predefined indicators.

[0009] Preferably, the relative damage determination submodule includes: The spatial determination unit is used to construct a multi-dimensional post-harvest storage quality evaluation system and a respiratory aging benchmark model for kiwifruit based on the varietal attributes and post-harvest respiratory physiological characteristics of the kiwifruit placed on the corresponding shelf. The core indicators of the storage quality evaluation system are used as dimensions to determine the sampling space of preservation parameters for the corresponding kiwifruit varieties. The sample acquisition unit is used to generate a storage and preservation benchmark sample set and a real-time quality evaluation sample set for the corresponding variety of kiwifruit within the preservation parameter sampling space, and to map the storage and preservation benchmark sample set and the real-time quality evaluation sample set to the standardized preservation evaluation space respectively, thereby obtaining the benchmark standardized sample and the real-time standardized sample. The correction unit is used to calculate the baseline storage quality response value of the baseline standardized sample based on the respiratory aging baseline model. At the same time, it dynamically corrects the baseline storage quality response value by combining the cumulative transportation time from the initial transportation time to the current time and the time decay coefficient output by the respiratory aging baseline model, so as to obtain the allowable quality deviation threshold range of kiwifruit under the corresponding shelf placement and corresponding cumulative transportation time. The boundary expansion unit is used to determine the boundary expansion range corresponding to the upper and lower limits of the allowable quality deviation threshold range based on the storage quality fluctuation tolerance of the corresponding kiwifruit variety and the disturbance coefficient of the transportation route conditions, and to generate the boundary expansion range. The deviation unit is used to calculate the weighted Mahalanobis distance between the real-time standardized sample and the benchmark standardized sample, taking the core indicators of the storage quality evaluation system as the calculation dimension and combining the physiological influence weights of each core indicator on the storage quality of kiwifruit, and to generate the real-time preservation deviation vector corresponding to each core indicator. The division unit is used to continuously divide the allowable quality deviation threshold range and the boundary expansion range based on the sensitivity of kiwifruit quality deterioration. Specifically, near the upper and lower limits of the allowable quality deviation threshold range and at the junction of the boundary expansion range and the allowable range, the minimum preset step size is used to divide the sub-processing range; in the middle region of the allowable quality deviation threshold range, the maximum preset step size is used to divide the sub-processing range, ultimately obtaining several continuous sub-processing ranges with step sizes adapted to the sensitivity of deterioration. The probability calculation unit is used to perform a Gaussian-cosine composite probability transformation on the real-time preservation deviation vector of each sub-processing interval that falls within the allowable quality deviation threshold range, so as to obtain the in-interval deviation deterioration probability value of the corresponding sub-processing interval. At the same time, for the real-time preservation deviation vector that exceeds the allowable quality deviation threshold range and falls into the boundary extension range, a piecewise decaying power composite probability transformation is performed to obtain the out-of-limit deviation deterioration probability value of the corresponding sub-processing interval. The relatively defined unit is used to construct a quality degradation probability sequence of kiwifruit placed on the shelf based on the in-interval deviation degradation probability value and the out-of-limit deviation degradation probability value corresponding to each sub-processing interval, according to the arrangement order of the sub-processing intervals. Combining the deviation direction, physiological influence weight and degradation probability contribution of all real-time preservation deviation vectors corresponding to the same core indicator, the core influencing indicators and degradation dominant factors are screened out to generate the relative damage factors of kiwifruit placed on the shelf.

[0010] Preferably, the feature and anomaly analysis module includes: The grayscale mapping unit is used to perform grayscale mapping on the images captured by the monitoring component when it is in the on state, based on the prior color gamut range of the kiwi fruit and the shelf on which it is placed, to generate a high-contrast grayscale image. A boundary determination unit is used to initially locate the first initial contour boundary of the kiwi fruit and the second initial contour boundary of the juice-stained area in the high-contrast grayscale image by using a dual gradient threshold. The mask determination unit is used to perform cubic spline interpolation fitting on the pixels of the first initial contour boundary and the second initial contour boundary respectively to locate the precise contour boundary at the sub-pixel level, and generate the first precise segmentation mask of the kiwi fruit and the second precise segmentation mask of the juice-stained area. At the same time, based on the prior space size of the shelf and the fixed coordinates inside the carriage, a fixed mask of the shelf area is generated. The pixel segmentation unit is used to extract all valid pixels within the first precise segmentation mask, perform local binary pattern texture feature calculation on each valid pixel, and divide the normal fruit peel pixel set into the abnormal pixel set. The connectivity analysis unit is used to perform connectivity analysis on the abnormal pixel set. Combining the texture threshold and grayscale threshold of the normal skin of kiwifruit, it filters out the target connected components that meet the damage characteristics and extracts the pixel area, perimeter, grayscale mean, texture roughness, and boundary gradient parameters of the target connected components to generate the damage characteristics of the kiwifruit itself. The point extraction unit is used to extract all valid pixels in the overlapping area of ​​the fixed mask and the second precise segmentation mask, which are regarded as the points to be analyzed. The erosion feature unit is used to calculate the deviation of each point to be analyzed relative to the reference gray value of the shelf, generate a pixel gray-level deviation matrix, and perform boundary gradient change analysis on the pixel gray-level deviation matrix based on the diffusion characteristics of sap contamination. It locates the sub-pixel boundary of the sap contamination infiltration front, calculates the pixel coverage area, infiltration depth, and boundary diffusion rate parameters of the infiltrated area, and combines the damage characteristics of kiwifruit in the same area to match the source fruit of sap seepage, generating liquid erosion features of damaged kiwifruit sap seeping onto the corresponding shelf.

[0011] Preferably, the feature and anomaly analysis module further includes: The coefficient generation unit is used to retrieve the basic storage and preservation vector of the corresponding shelf area at the current moment, and combine it with the cumulative transportation time from the initial transportation time to the current moment to generate the environmental impact coefficient of quality deterioration of the corresponding shelf. The quality analysis unit is used to perform pixel-level benchmark matching between the extracted kiwi fruit damage characteristics and the liquid erosion characteristics of the kiwi fruit damage and oozing juice on the corresponding shelf and the initial quality benchmark library of the corresponding shelf. It distinguishes between the fruit damage characteristics and the juice erosion characteristics added during transportation, calculates the abnormal deterioration level, real-time deterioration rate and juice contamination and diffusion risk coefficient of the kiwi fruit in the corresponding shelf, and completes the quality abnormality analysis of the kiwi fruit on the corresponding shelf relative to the initial loading state. The coordinate construction unit is used to construct a unified two-dimensional spatial coordinate system within the transport compartment based on the spatial arrangement coordinates of each shelf in the transport compartment and the monitoring coverage area boundaries of all activated monitoring components, with the pre-calibrated unique fixed verification code of each shelf as the spatial positioning reference anchor point. The parameter mapping unit is used to map the abnormal deterioration level, real-time deterioration rate, juice contamination and diffusion risk coefficient, and quality deterioration environmental impact coefficient of each placed shelf to the corresponding position in the two-dimensional spatial coordinate system, thereby generating a single shelf abnormal attribute node. The correction unit is used to construct an abnormal risk transmission matrix between adjacent shelves based on the cross-shelf risk transmission characteristics of kiwifruit rotten juice contamination and ethylene gas diffusion. It performs adjacent correction on the risk coefficient of each single shelf abnormal attribute node, and generates a kiwifruit abnormal distribution layer with spatial positioning, abnormal level, deterioration trend and diffusion risk attributes based on all corrected single shelf abnormal attribute nodes and the monitoring coverage boundary of each activated monitoring component.

[0012] Preferably, the initial quality benchmark library includes initial damage characteristic parameters of kiwifruit in the corresponding shelf, initial skin texture benchmark, initial cleanliness benchmark of the corresponding shelf, and postharvest respiration deterioration benchmark model of the kiwifruit variety.

[0013] Compared with the prior art, the beneficial effects of this application are: Achieve precise shelf-level control and eliminate monitoring blind spots: The granularity of monitoring and control is refined from the whole vehicle level to the individual shelf level. For each shelf, the initial state of the kiwifruit, the real-time environment, and external disturbance factors are used to generate a monitoring status vector. At the same time, independent regional controlled atmosphere control and dynamic scheduling of monitoring resources are realized to adapt to the differentiated preservation needs of different shelves and completely eliminate the monitoring blind spots of whole vehicle-level control. Multi-dimensional risk pre-assessment enables early intervention: By integrating four types of multi-source vectors—weather conditions, road conditions, initial packaging status, and real-time controlled atmosphere environment—and using a neural network model for coupled analysis, pre-assessment of preservation risks is achieved, allowing for early prediction of preservation risks during transportation. This provides a precise basis for dynamic controlled atmosphere control and monitoring scheduling, addressing the shortcomings of existing technologies in risk assessment. Dual-dimensional anomaly feature recognition significantly reduces the false negative rate: Simultaneously extracting the damage features of the kiwifruit itself and the liquid erosion features of the juice on the shelf, it can not only identify the fruit damage that has already occurred, but also predict the risk of secondary chain rot in advance through the juice contamination features, bringing the anomaly identification node forward by an average of more than 12 hours, and reducing the false negative rate to less than 2%. Monitoring resources are scheduled on demand to reduce equipment power consumption: Based on the preservation risk level of a single shelf, the start-up and stop of monitoring components and the frame rate of data acquisition are dynamically controlled. High-risk areas are monitored in a key manner, while low-risk areas operate with low power consumption, avoiding the waste of resources from full-time and full-volume monitoring. The average power consumption of monitoring equipment is reduced by more than 70%. Tiered and reliable traceability enables end-to-end control: Tiered primary and secondary verification records are set up for abnormal / normal scenarios. Combined with blockchain on-chain evidence storage, the credibility of traceability data is improved. At the same time, the storage of kiwifruit in each cold chain warehouse is dynamically adjusted based on the end-to-end traceability data, realizing the deep integration of freshness monitoring and supply chain management, and forming end-to-end control. Deterioration judgment is adapted to physiological characteristics, and the accuracy is greatly improved: A respiratory aging benchmark model is constructed based on the postharvest respiratory physiological characteristics of kiwifruit. Combined with the time-series decay effect of transportation time, the quality abnormality judgment threshold is dynamically adjusted to adapt to the fruit deterioration pattern at different transportation stages, thereby reducing the false judgment rate and false negative rate.

[0014] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0015] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0016] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a structural diagram of a kiwifruit storage and preservation monitoring system based on dynamic controlled atmosphere and cold chain traceability in an embodiment of the present invention; Figure 2 This is a structural diagram of the relative damage determination submodule in an embodiment of the present invention. Detailed Implementation

[0017] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0018] This invention provides a kiwifruit storage and preservation monitoring system based on dynamic controlled atmosphere and cold chain traceability, such as... Figure 1 As shown, it includes: The vector acquisition module is used to acquire the preset kiwifruit cold chain transportation route and the real-time location of the transport driver during the transportation process according to the cold chain transportation route. Combined with the weather state vector corresponding to the real-time location, the road segment state vector of the current road segment, the pre-transport packaging state vector of the kiwifruit, and the basic storage and preservation vector of each shelf area at the current moment, the module generates the monitoring state vector corresponding to each shelf at the real-time location. The pre-transport packaging state vector includes the initial loading and storage state of the kiwifruit on each shelf in the transport vehicle, and the basic storage and preservation vector includes various environmental parameters of the corresponding shelf area. The dynamic controlled atmosphere module is used to dynamically adjust the regional controlled atmosphere adjustment unit built into the transport vehicle according to the monitoring status vector corresponding to each shelf and the real-time storage and preservation requirements of the kiwifruit on the corresponding shelf. At the same time, it dynamically controls the start and stop of several monitoring components pre-set in the transport vehicle according to the placement position of each shelf in the transport vehicle and the monitoring status vector of the corresponding shelf. The feature and anomaly analysis module is used to extract the damage features of the kiwifruit itself and the liquid erosion features of the kiwifruit juice seeping out from the images captured by each monitoring component in the active state. Combined with the basic storage and preservation vector of the corresponding shelf area at the current moment, it analyzes the quality anomalies of the kiwifruit on the corresponding shelf relative to the initial loading state, constructs an anomaly distribution layer of kiwifruit based on the monitoring range covered by all active monitoring components, and determines whether the kiwifruit meets the set replacement criteria based on the anomaly distribution layer. If the conditions are met, the location information of the cold chain warehouse closest to the real-time location is sent to the vehicle terminal, and the inspection of the abnormal batch of kiwifruit and the corresponding full-process data are recorded once when the transport vehicle arrives at the cold chain warehouse; if the conditions are not met, when all the kiwifruit monitored within the full coverage area of ​​the cold chain transport route corresponding to the same cold chain warehouse are in a condition that does not meet the set replacement criteria, the full-process data of the batch of kiwifruit is recorded a second time. The traceability chain construction module is used to generate a complete cold chain traceability chain for the transported batch of kiwifruit based on the results of primary recording, secondary recording, real-time transportation information along the cold chain transportation route, and regional controlled atmosphere regulation data. It also dynamically regulates the kiwifruit storage in each cold chain warehouse according to the full-process storage data of the cold chain traceability chain.

[0019] This embodiment applies to a long-distance cold chain transport vehicle for kiwifruit. The interior of the cold chain transport vehicle is equipped with multi-layered, multi-row independent storage shelves. Each storage shelf is an independent sealed storage unit. Each storage unit is equipped with an independent zoned controlled atmosphere unit, environmental parameter acquisition sensors, and at least one set of monitoring components. The vehicle is equipped with an on-board central control system for executing the method steps of this invention. The on-board central control system is wirelessly connected to the cloud management platform and the on-board terminal.

[0020] In this embodiment, each independent storage compartment unit corresponding to the shelf is equipped with an independent zoned controlled atmosphere unit, which includes: a sealed storage compartment cabinet, a circulating air pump, and a nitrogen generator module. The system includes a removal module, an ethylene removal module, an electromagnetic air inlet valve, an electromagnetic exhaust valve, and a parameter acquisition component. The sealed storage cabinet is made of food-grade 304 stainless steel, with EPDM sealing strips on the doors. The internal airtightness of the cabinet is ≤0.02MPa / h, enabling the construction of an independent controlled atmosphere environment for each shelf area. The parameter acquisition component includes: Concentration sensor Concentration sensors, ethylene concentration sensors, PT100 temperature sensors, and capacitive humidity sensors are all installed inside the storage cabinet at the return air vent. These sensors collect various parameters of the basic storage and preservation vector for the corresponding shelf area in real time, with a sampling frequency of once every 10 seconds. The nitrogen generation module... The ethylene removal module and the ethylene removal module are connected to the internal air circuit of the storage compartment via a circulating air pump. Electromagnetic intake valves and electromagnetic exhaust valves are installed at the intake and exhaust ends of the air circuit, respectively. The vehicle-mounted central control system dynamically adjusts the operating power and valve opening of each module using an incremental PID control algorithm based on the monitored state vector, achieving closed-loop regulation of the corresponding storage compartment's internal atmosphere parameters. For kiwifruit storage and preservation, in this embodiment, the parameter control reference range for the regional atmosphere regulation unit is: Volume concentration 2-5%, Volume concentration 3-5%, ethylene concentration The ambient temperature is 0±0.5℃, and the relative humidity is 90-95%RH. The parameter baselines for different kiwifruit varieties can be adjusted according to their post-harvest physiological characteristics; for example, the suitable parameters for Hongyang kiwifruit... The volume concentration is 3-4%. A concentration of 2-3% is suitable for Xu Xiang kiwifruit. The volume concentration is 2-3%. The volume concentration is 4-5%.

[0021] Each storage compartment corresponding to a shelf is equipped with at least one monitoring component. The monitoring component uses an industrial-grade RGB wide dynamic range camera with a resolution of 2 megapixels or higher. It is installed at the top center of the storage compartment, with the lens facing the shelf surface. The acquisition area completely covers all the kiwis and shelf surface of the corresponding shelf, achieving monitoring without blind spots. The start-up, shutdown, and frame rate of the monitoring component are dynamically controlled by the vehicle-mounted central control system based on the monitoring status vector of the corresponding shelf. The total number of monitoring components is N, where N is a positive integer and not less than the total number of shelves in the transport vehicle. In this embodiment, each shelf corresponds to one monitoring component, and N is equal to the total number of shelves.

[0022] The pre-set cold chain transportation route is obtained through the APP map open platform, which includes the origin, destination, route, estimated transportation time, service areas and compliant cold chain warehouses along the way, and is pre-stored in the vehicle central control system; The real-time location is obtained through the Beidou / GPS dual-mode positioning module built into the transport vehicle, with a positioning accuracy of ≤5m and a sampling frequency of 1 time / minute; The weather state vector is obtained from the National Meteorological Data Open Platform using hourly meteorological data for the corresponding geographic area in real time. This data includes ambient temperature, ambient humidity, rainfall, light intensity, and wind speed. The data is processed to the [0,1] interval using an extreme value normalization formula to form the weather state vector: W=[T_w,H_w,R,L,P], where T_w is ambient temperature, H_w is ambient humidity, R is rainfall, L is light intensity, and P is wind speed. The extreme values ​​of the weather state vector parameters are shown in Table 1. Table 1 Extreme Values ​​of Weather State Vector Parameters The road segment state vector is obtained by acquiring real-time operating data of the corresponding road segment through the map open platform and the vehicle-mounted triaxial accelerometer, including road surface smoothness, slope, bump vibration acceleration, speed limit, and congestion level. This data is processed to the [0,1] interval using an extreme value normalization formula, forming the road segment state vector R=[S,G,A,V,C], where S is road surface smoothness, G is slope, A is bump vibration acceleration, V is speed limit, and C is congestion level. The extreme values ​​of the road segment state vector parameters are shown in Table 2. Table 2 Extreme Values ​​of Road Segment State Vector Parameters The pre-transport packaging state vector is generated by non-destructive testing and calibration of each kiwifruit placed on a shelf before transportation. This includes the initial loading and storage state of the kiwifruit on the corresponding shelf, specifically: initial maturity (titerable acid content, unit: %), initial damage (breakage rate, unit: %), packaging airtightness parameters (unit: MPa / h), initial skin texture benchmark (LBP texture mean), and initial shelf cleanliness benchmark (grayscale mean). These parameters are then processed to the [0,1] interval using an extreme value normalization formula to form the pre-transport packaging state vector E=[M,D]. [,Q,T0,J0], where M is the initial maturity, D is the initial damage condition, Q is the packaging airtightness parameter, T0 is the initial skin texture benchmark, and J0 is the initial shelf cleanliness benchmark; based on the packaging state vector before transportation, an initial quality benchmark library is constructed for each batch of kiwifruit placed on the shelf. The initial quality benchmark library contains the initial damage characteristic parameters of the kiwifruit in the corresponding shelf, the initial skin texture benchmark, the initial cleanliness benchmark of the corresponding shelf, and the postharvest respiration deterioration benchmark model of the kiwifruit variety, which is pre-stored in the vehicle central control system; The basic storage and preservation vector is acquired in real time through the parameter acquisition component of the controlled atmosphere control unit in each shelf area, including various environmental parameters of the corresponding shelf area, specifically: concentration, Concentration, ethylene concentration, ambient temperature, and ambient humidity are processed to the [0,1] interval using an extreme value normalization formula to form the basic storage and preservation vector Env=[ , , ,T,H].

[0023] In this embodiment, the dynamic control method of the zoned controlled atmosphere unit is as follows: Control objective: To adjust the basic storage and preservation vector of the corresponding shelf to the standard preservation vector range, and suppress the abnormal development of high-risk dimensions in the monitored state vector; An incremental PID closed-loop control algorithm is adopted, and the incremental PID formula is: ,in, For the first The control increment for each sample; This is the proportionality coefficient; The integral coefficient; These are the differential coefficients; For the first The deviation between the measured and target values ​​of the parameters in the second sampling; This represents the change in deviation between two consecutive samples.

[0024] In this embodiment, the differentiated control rules are shown in Table 3: Table 3 Differentiated Regulation Rules Under normal operating conditions, the control cycle is 10 minutes. Under high-risk operating conditions (any dimension of the monitoring state vector ≥ 0.7), the control cycle is shortened to 1 minute, achieving rapid closed-loop control.

[0025] In this embodiment, the monitoring component is dynamically started and stopped: Monitoring component configuration: Each shelf corresponds to one 2-megapixel industrial-grade RGB wide dynamic range camera, installed at the top center of the storage cabinet, with the lens facing the shelf surface, and the collection area completely covers all the kiwis and shelf surface of the corresponding shelf; The start / stop and frame rate scheduling rules are dynamically adjusted based on the single-shelf preservation risk level (determined by the maximum dimension value of the monitoring state vector), as shown in Table 4: Table 4 Start / Stop and Frame Rate Scheduling Rules A lightweight YOLOv8n model is used for real-time image processing. The model has only 3.2M parameters and the computing power of the vehicle terminal is ≤20%, which meets the real-time requirements of cold chain transportation.

[0026] In this embodiment, the cold chain traceability chain construction method is as follows: Underlying architecture: It adopts the Hyperledger Fabric consortium blockchain architecture, with five core node types including production cooperatives, cold chain transportation companies, cold chain warehousing companies, wholesale markets in sales areas, and regulatory departments, to achieve distributed data storage, immutability, and traceability.

[0027] On-chain rules: Each transportation batch corresponds to a unique traceability ID. The data is encrypted using the SHA-256 hash algorithm. The on-chain data includes: ① basic batch information (variety, origin, harvest time, maturity, initial quality test report); ② real-time transportation information throughout the entire process (transportation trajectory, driver information, vehicle information, road conditions); ③ full time-series data of regional controlled atmosphere regulation; ④ full data of primary / secondary records; ⑤ anomaly handling and inspection reports.

[0028] Traceability: Each batch of kiwifruit corresponds to a unique traceability QR code. Consumers and managers can scan the code to view the entire process of transportation, storage environment, quality testing, and abnormal handling information for that batch of kiwifruit, achieving full traceability.

[0029] Methods for dynamic control of cold chain storage capacity: Regulation Model: A multi-objective optimization model is adopted, with the optimization objectives being to minimize the loss rate of kiwifruit in cold chain warehouses nationwide, maximize the supply-demand matching degree, and minimize transportation costs. The constraints include cold chain warehouse capacity, transportation capacity, market supply and demand, and preservation capacity.

[0030] Control Rules: Based on the cold chain traceability data of each batch of kiwifruit, the cloud management platform statistically analyzes the storage loss rate, preservation effect, and market supply and demand of kiwifruit from different origins, varieties, and transportation routes, and dynamically optimizes the storage capacity of kiwifruit in cold chain warehouses in various regions. For transportation routes and cold chain warehouses with good preservation effect and low loss rate (<2%), the storage volume of kiwifruit should be appropriately increased to enhance the supply guarantee capacity of the region. For routes and cold chain warehouses with high loss rates (>5%) and frequent abnormal situations, optimize transportation plans and storage parameters to reduce storage volume and minimize losses; For regions with strong market demand and insufficient inventory, priority will be given to allocating batches of kiwifruit with good preservation to ensure market supply. Based on traceability data, a database of optimal storage periods and transportation routes for different varieties of kiwifruit is established to continuously optimize the overall scheduling plan.

[0031] In this embodiment, the quantification rules for setting the replacement standard are shown in Table 5: Table 5. Quantitative Rules for Setting Replacement Standards This embodiment sets up a control group and an experimental group. The control group uses the existing technology of whole-vehicle static modified atmosphere + all-time monitoring + conventional bad fruit identification. The experimental group uses the method described above in this invention. The transported object is Xu Xiang kiwifruit, and the transportation route, transportation time, loading amount and initial quality are completely consistent. After transportation, the preservation effect of the fruit is tested by a third party. The results are shown in Table 6: Table 6 Comparison of Results As can be seen from the above results, the method of the present invention can significantly improve the preservation effect of kiwifruit cold chain transportation, greatly reduce fruit loss rate, equipment power consumption and abnormality missed detection rate, and significantly advance the abnormality identification node, realizing the early intervention of preservation risks.

[0032] This invention provides a kiwifruit storage and preservation monitoring system based on dynamic controlled atmosphere and cold chain traceability. The vector acquisition module includes: The model analysis submodule is used to input the weather state vector, road segment state vector, pre-transport packaging state vector, and basic storage and preservation vector into the corresponding pre-trained neural network model to obtain the first factor affecting the storage and preservation effect in each vector and the storage and preservation confidence of each first factor. The mapping processing submodule is used to map each predefined indicator in the monitoring status vector to the first factor under each vector to obtain a set of candidate factors for each predefined indicator. The sorting submodule is used to obtain a second factor from the candidate factor set whose storage and preservation confidence is greater than a preset confidence level. At the same time, the candidate factor set is sorted first according to the mapping correlation coefficient and sorted second according to the storage and preservation confidence level. The cutting submodule is used to perform a first forward cut on the second sorting result to obtain several third factors, taking the first element in the first sort as the first cutting point; at the same time, it performs a second forward cut on the first sorting result to obtain several fourth factors, taking the first element in the second sort as the second cutting point. The direction matching submodule is used to determine the first preservation anomaly direction based on all second factors, the second preservation anomaly direction based on all third factors, and the third preservation anomaly direction based on all fourth factors based on the factor-direction comparison table, thus forming a state anomaly subset corresponding to the predefined indicators. The relative damage determination submodule is used to determine the relative damage factors of the kiwifruit placed on the corresponding shelf based on the basic storage and preservation vector and the standard preservation vector of the pre-transportation packaging state vector, and in combination with the cumulative transportation time from the initial transportation time to the current time, the variety attributes of the kiwifruit placed on the shelf, and the post-harvest respiratory physiological characteristics. The adjustment submodule is used to adjust the maximum and minimum abnormal directions among the first, second, and third abnormal directions of preservation under the corresponding predefined indicators according to the relative damage factors of the kiwifruit, thereby forming a state abnormal subset of the corresponding predefined indicators. The vector combination submodule is used to obtain the monitoring status vector corresponding to each placed shelf based on the abnormal status subset of all predefined indicators.

[0033] In this embodiment, the neural network model architecture employs four independent 3-layer BP neural network models, each corresponding to one of the four types of input vectors. Each model includes an input layer, a hidden layer, and an output layer. Taking the BP neural network model corresponding to the weather state vector as an example: the input layer has 5 nodes, corresponding to the 5 parameters of the weather state vector; the hidden layer has 10 nodes, and the ReLU activation function is used; the output layer has 2 nodes, outputting the first factor affecting storage and preservation, and the storage and preservation confidence level of the first factor (value range 0-1). The architecture of the other three models is consistent with the above model, with the number of input layer nodes matching the number of parameters of the corresponding vector, the number of hidden layer nodes being twice the number of input layer nodes, and the output layer having 2 nodes each.

[0034] Model Training and Validation: The training samples consist of historical full-process data from 1200 batches of cold chain transportation of Xuxiang, Hongyang, and Cuixiang kiwifruit from 2022 to 2025, totaling 1.5 million data points. 80% of this data is used as the training set, 10% as the validation set, and 10% as the test set. During training, the mean squared error (MSE) was used as the loss function, and the Adam optimizer was employed for iterative optimization. The learning rate was set to 0.001, and the number of iterations was 1500. After training, the model's prediction accuracy was ≥95.2%, and the generalization error was ≤3.8%, meeting the usage requirements. The model was pre-stored in the vehicle's central control system, with an inference frequency of 1 time per minute.

[0035] First factor selection rule: Among the influencing factors output by the model, the factors with a confidence level of ≥0.3 for storage and preservation are selected as the first factors to ensure coverage of all core influencing factors.

[0036] The predefined indicators for the monitoring status vector are fixed at 5: temperature anomaly indicator, gas environment anomaly indicator, mechanical damage risk indicator, fruit deterioration risk indicator, and secondary pollution risk indicator.

[0037] The linear correlation between each first factor and its corresponding predefined index is calculated using the Pearson correlation coefficient formula to obtain the mapping correlation coefficient, which ranges from [0,1]. ,in, Mapping correlation coefficient, The measured value of the first factor. For the measured values ​​of the corresponding predefined indicators, for The mean of the sequence, for The mean of the sequence, This represents the number of samples.

[0038] Candidate factor set generation: For each predefined indicator, select all first factors with a mapping correlation coefficient ≥ 0.2 to form the candidate factor set for that indicator.

[0039] In this embodiment, the preset confidence level is a threshold for screening the core factors affecting the storage and preservation of kiwifruit. Its base fixed value is 0.6, with a range of 0.5 to 0.7. This threshold is set based on the validation results of the neural network model for predicting the cold chain storage and preservation risks of kiwifruit. Only factors with a storage and preservation confidence level ≥ this threshold are retained as secondary factors. This covers more than 95% of the core deterioration factors affecting the storage and preservation effect of kiwifruit, while filtering out low-confidence interference factors, ensuring the accuracy and computational efficiency of risk prediction. In long-distance cold chain scenarios with a transportation time exceeding 72 hours, this threshold can be lowered to 0.5 to improve the coverage of potential deterioration risks; in short-distance cold chain scenarios with a transportation time ≤ 24 hours, this threshold can be raised to 0.7 to further reduce computational resource consumption.

[0040] First sort: Sort in descending order of mapping correlation coefficients, and the sorting result is denoted as Seq1.

[0041] Second sort: Sort in descending order of storage and preservation confidence, and the sorting result is denoted as Seq2.

[0042] In this embodiment, the first forward cutting rule is as follows: taking the first element of Seq1 as the reference point, all elements in Seq2 with a storage and preservation confidence difference of ≤0.2 from the reference point are cut forward, and the cut result is the third factor.

[0043] The second forward cutting rule: Taking the first element of Seq2 as the reference point, cut forward in Seq1 all elements whose mapping correlation coefficient difference with the reference point is ≤0.2. The cut result is the fourth factor.

[0044] The core mapping relationship of the factor-direction comparison table is shown in Table 7. The table also includes the physiological influence weight of each factor, providing a basis for subsequent risk calculation. Table 7 Factor-Direction Comparison Table Composition of abnormal status subsets: Each subset contains five core fields: influencing factor ID, storage and preservation confidence, mapping correlation coefficient, abnormal direction, and physiological impact weight, which provide a basis for subsequent abnormal risk prediction.

[0045] In this embodiment, the adjustment rules of the adjustment submodule are as follows: if the dominant factor of deterioration relative to the damage factor corresponds to a certain abnormal direction, the risk level of the abnormal direction is increased by 1 level, while the risk level of abnormal directions unrelated to the abnormal direction is decreased by 1 level, so as to ensure that the abnormal subset of the state accurately matches the actual deterioration trend.

[0046] In this embodiment, based on the abnormal subset of all predefined indicators, a comprehensive abnormal risk value for each predefined indicator is calculated, forming a monitoring status vector for the corresponding shelf. The formula for calculating the comprehensive abnormal risk value is as follows: ,in, For the first The comprehensive abnormal risk value of a predefined indicator, with a value range of [0,1]; For the first The mapping correlation coefficient of each influencing factor Weighting for physiological effects, This represents the probability of degradation. This represents the number of influencing factors in the abnormal subset of the indicator's state.

[0047] In this embodiment, the monitoring status vector is generated by using the comprehensive abnormal risk value of 5 predefined indicators as elements to form a monitoring status vector V=[Risk1,Risk2,Risk3,Risk4,Risk5]. Each shelf corresponds to a unique monitoring status vector, and the update frequency is 1 time / minute.

[0048] The beneficial effects of the above technical solution are: through neural network analysis, multi-dimensional factor mapping and sorting and segmentation processing, the key factors and abnormal directions affecting freshness are accurately extracted, making the monitoring status vector more in line with the actual risks, and providing a scientific basis for dynamic atmosphere control and monitoring scheduling.

[0049] This invention provides a kiwifruit storage and preservation monitoring system based on dynamic controlled atmosphere and cold chain traceability. The relative damage determination submodule, such as... Figure 2 As shown, it includes: The spatial determination unit is used to construct a multi-dimensional post-harvest storage quality evaluation system and a respiratory aging benchmark model for kiwifruit based on the varietal attributes and post-harvest respiratory physiological characteristics of the kiwifruit placed on the corresponding shelf. The core indicators of the storage quality evaluation system are used as dimensions to determine the sampling space of preservation parameters for the corresponding kiwifruit varieties. The sample acquisition unit is used to generate a storage and preservation benchmark sample set and a real-time quality evaluation sample set for the corresponding variety of kiwifruit within the preservation parameter sampling space, and to map the storage and preservation benchmark sample set and the real-time quality evaluation sample set to the standardized preservation evaluation space respectively, thereby obtaining the benchmark standardized sample and the real-time standardized sample. The correction unit is used to calculate the baseline storage quality response value of the baseline standardized sample based on the respiratory aging baseline model. At the same time, it dynamically corrects the baseline storage quality response value by combining the cumulative transportation time from the initial transportation time to the current time and the time decay coefficient output by the respiratory aging baseline model, so as to obtain the allowable quality deviation threshold range of kiwifruit under the corresponding shelf placement and corresponding cumulative transportation time. The boundary expansion unit is used to determine the boundary expansion range corresponding to the upper and lower limits of the allowable quality deviation threshold range based on the storage quality fluctuation tolerance of the corresponding kiwifruit variety and the disturbance coefficient of the transportation route conditions, and to generate the boundary expansion range. The deviation unit is used to calculate the weighted Mahalanobis distance between the real-time standardized sample and the benchmark standardized sample, taking the core indicators of the storage quality evaluation system as the calculation dimension and combining the physiological influence weights of each core indicator on the storage quality of kiwifruit, and to generate the real-time preservation deviation vector corresponding to each core indicator. The division unit is used to continuously divide the allowable quality deviation threshold range and the boundary expansion range based on the sensitivity of kiwifruit quality deterioration. Specifically, near the upper and lower limits of the allowable quality deviation threshold range and at the junction of the boundary expansion range and the allowable range, the minimum preset step size is used to divide the sub-processing range; in the middle region of the allowable quality deviation threshold range, the maximum preset step size is used to divide the sub-processing range, ultimately obtaining several continuous sub-processing ranges with step sizes adapted to the sensitivity of deterioration. The probability calculation unit is used to perform a Gaussian-cosine composite probability transformation on the real-time preservation deviation vector of each sub-processing interval that falls within the allowable quality deviation threshold range, so as to obtain the in-interval deviation deterioration probability value of the corresponding sub-processing interval. At the same time, for the real-time preservation deviation vector that exceeds the allowable quality deviation threshold range and falls into the boundary extension range, a piecewise decaying power composite probability transformation is performed to obtain the out-of-limit deviation deterioration probability value of the corresponding sub-processing interval. The relatively defined unit is used to construct a quality degradation probability sequence of kiwifruit placed on the shelf based on the in-interval deviation degradation probability value and the out-of-limit deviation degradation probability value corresponding to each sub-processing interval, according to the arrangement order of the sub-processing intervals. Combining the deviation direction, physiological influence weight and degradation probability contribution of all real-time preservation deviation vectors corresponding to the same core indicator, the core influencing indicators and degradation dominant factors are screened out to generate the relative damage factors of kiwifruit placed on the shelf.

[0050] In this embodiment, a multi-dimensional storage quality evaluation system is implemented, with core indicators divided into three categories: physiological quality indicators, environmental regulation indicators, and physical damage indicators. The weights of each indicator are determined using the analytic hierarchy process (AHP), as shown in Table 8. Table 8 Multi-dimensional Storage Quality Evaluation System Respiratory aging baseline model: A dynamic model of postharvest respiration rate in kiwifruit was constructed, with the following formula: ,in, The respiration rate of the fruit at transport time t, in units of R0 is the initial respiratory rate, obtained from measurements taken before transport. The activation energy is shown in Table 9 for different varieties of kiwifruit; R is the gas constant, with values... ; t represents the thermodynamic temperature of the storage environment at time t, in K; k is the aging coefficient, which is determined by the kiwi fruit variety, see Table 9; t is the cumulative transportation time from the initial transportation time to the present time, in h; Table 9 Core Parameters of Different Kiwi Fruit Varieties Preservation parameter sampling space: Using the 12 secondary indicators of the above quality evaluation system as dimensions, the value range of each indicator is the suitable range for the storage and preservation of this variety of kiwifruit. A 12-dimensional sampling space is constructed to provide a benchmark for subsequent sample generation.

[0051] In this embodiment, Sobol low-bias sequence is used for quasi-random sampling. The sample size of the storage and preservation benchmark sample set is 5000 sets, and the sample size of the real-time quality assessment sample set is 1000 sets. The sampling results uniformly cover the entire preservation parameter sampling space, avoiding the clustering defects of random sampling. Standardization mapping method: First, extreme value normalization is used to map the values ​​of each indicator to the [0,1] interval to eliminate the influence of dimensions. Then, Z-score standardization is used to transform the data into standard normal distribution data with a mean of 0 and a variance of 1, finally obtaining standardized samples. The formula is as follows: Extreme value normalization: Z-score standardization: ,in, The original data, , These are the minimum and maximum values ​​of the indicator, respectively. The sample mean. This represents the sample standard deviation.

[0052] In this embodiment, the time-series decay coefficient is calculated based on the respiratory aging baseline model, and the formula is: ,in, This is the aging coefficient; the longer the transportation time, the higher the aging coefficient. The larger the value, the narrower the allowable quality deviation threshold range, and the stricter the judgment standard, which is suitable for the aging and deterioration characteristics of fruits as transportation time increases. In this embodiment, ,in, The allowable quality deviation threshold range; This is the baseline storage quality response value.

[0053] The basic boundary expansion range is 20% of the interval width, that is, the boundary expansion interval is [ ×0.8, ×1.2]; If the bump and vibration acceleration of the transport section is ≥0.5g, the boundary expansion range is reduced to 10%, that is, the boundary expansion range is [ ×0.9, ×1.1], improves the accuracy of boundary determination and adapts to the quality fluctuation characteristics under bumpy working conditions.

[0054] In this embodiment, the formula for calculating the weighted Mahalanobis distance is: ,in, For real-time standardized samples, Using the standardized sample as the baseline, W is the diagonal matrix of weights for each indicator. Let be the covariance matrix of the sample. It is the inverse of the covariance matrix; the weighted Mahalanobis distance takes into account the weight of each indicator and the correlation between indicators, and the calculation result is more consistent with the actual quality deviation of the fruit.

[0055] Real-time preservation deviation vector: The real-time preservation deviation vector is constructed using the weighted Mahalanobis distance of each core indicator as its elements, and the dimension is consistent with the number of core indicators.

[0056] In this embodiment, the minimum preset step size is fixed at 1% of the total width of the interval, with a value range of 0.5% to 2% of the total width of the interval. This is suitable for the vicinity of the upper and lower limits of the allowable quality deviation threshold interval, and the junction of the boundary expansion interval and the allowable interval. These areas are the regions with the highest sensitivity to quality deterioration of kiwifruit. Using the minimum step size for dense division can improve the boundary judgment accuracy to over 98%. The maximum preset step size is fixed at 5% of the total width of the interval, with a value range of 3% to 8% of the total width of the interval. This is suitable for the middle region of the allowable quality deviation threshold interval. These areas are the stable regions with the lowest sensitivity to quality deterioration of kiwifruit. Using the maximum step size for division can reduce the computational load by more than 60% while ensuring the accuracy of the judgment.

[0057] Degradation sensitivity determination: The closer to the interval boundary, the higher the quality degradation sensitivity. Smaller step size is used to encrypt the partition to improve the accuracy of boundary determination. The degradation sensitivity in the middle of the interval is low. Larger step size is used to reduce the amount of computation.

[0058] In this embodiment, the adaptive Gaussian-cosine composite probability transformation formula is as follows: The optimal storage benchmark point is the midpoint of the allowable quality deviation threshold range (membership probability = 1, degradation probability = 0). The closer to the range boundary, the higher the degradation probability becomes non-linearly, as shown in the following formula: ,in, This is the probability value of quality deterioration within the interval, which is dimensionless and ranges from [0,1]. The closer the value is to 1, the higher the probability of quality deterioration of the kiwifruit under the corresponding deviation. The value range is [0.3, 0.8], with a base value of 0.8. Its value is dynamically adjusted based on the cumulative transportation time t and the physiological impact weight of the corresponding indicator. The longer the cumulative transportation time, the higher the weight of the physiological impact of the indicator. The smaller the value, the higher the proportion of the cosine function, and the faster the rate of increase in the probability of deterioration at the boundary, which is suitable for the aging and deterioration characteristics of kiwifruit as transportation time increases; d is the weighted Mahalanobis distance of the real-time preservation deviation vector, which is the result of calculating the weighted Mahalanobis distance between the real-time standardized sample and the baseline standardized sample. The larger the value, the greater the deviation between the real-time quality and the baseline quality. Here, represents the baseline distance corresponding to the midpoint of the interval; represents the weighted Mahalanobis distance baseline value corresponding to the midpoint of the allowable quality deviation threshold interval. This point is the optimal storage baseline point for kiwifruit, corresponding to a deterioration probability of 0. The unit of time is 1 hour; The standard deviation of the Gaussian function is fixed at 1 / 6 of the total width of the allowable quality deviation threshold range, ensuring a smooth distribution of the Gaussian function within the allowable range, which is consistent with the non-linear characteristics of kiwifruit quality deterioration. The total width of the allowable quality deviation threshold range, and the difference between the upper and lower limits of the allowable quality deviation threshold range. It should be noted that this only applies to... .

[0059] The piecewise decay power-law composite probability transformation formula is applicable to real-time preservation deviation vectors that exceed the allowable quality deviation threshold range but fall within the boundary expansion range. It divides the boundary expansion range into a near-boundary gradual change zone (0-50% exceeding the allowable range width) and a far-boundary abrupt change zone (50%-100% exceeding the allowable range width), and calculates the excess deviation degradation probability value for each corresponding sub-processing range. Gradually changing zone near the boundary: ; The region of dramatic change at the far boundary: ,in, , Let be the probability values ​​of exceeding the limit deviation for the near-boundary gradual change zone and the far-boundary drastic change zone, respectively. They are dimensionless and range from [0,1]. The closer the value is to 1, the higher the probability of the kiwifruit undergoing quality deterioration under the limit deviation. The baseline distance is the boundary of the allowable quality deviation threshold range, and the weighted Mahalanobis distance baseline value is the upper and lower limits of the allowable quality deviation threshold range. , , These are model constants. A fixed value of 0.1 is used, with a range of 0.05 to 0.2, to adapt to the gradually changing characteristics of the degradation probability near the boundary region. A fixed value of 0.8, with a range of 0.5 to 1.0, is used to adjust the rising slope of the probability of degradation in the near-boundary region. A fixed value of 2, ranging from 1.5 to 3.0, is used to regulate the rate at which the degradation probability in the far-boundary region converges to 1. This ensures that the greater the deviation from the allowable range, the faster the degradation probability approaches 1, aligning with the quality degradation pattern of kiwifruit after exceeding the limit. A larger d value results in a degradation probability closer to 1, also aligning with the degradation pattern of kiwifruit after exceeding the limit. It should be noted that the near-boundary slow-change zone... ; far-boundary dramatic change zone .

[0060] In this embodiment, the contribution of a certain indicator to the probability of deterioration = the physiological impact weight of the indicator × the probability value of deterioration corresponding to the indicator; the top 3 indicators in terms of contribution are determined as the dominant factors of deterioration, that is, the relative damage factors of kiwifruit.

[0061] The beneficial effects of the above technical solution are: a quality evaluation and aging model is constructed based on the physiological characteristics of kiwifruit, and the risk of deterioration is accurately determined by weighted Mahalanobis distance and probability transformation, so that the identification of deterioration is more in line with the true state of the fruit and the misjudgment and missed judgment are greatly reduced.

[0062] This invention provides a kiwifruit storage and preservation monitoring system based on dynamic controlled atmosphere and cold chain traceability. The feature and anomaly analysis module includes: The grayscale mapping unit is used to perform grayscale mapping on the images captured by the monitoring component when it is in the on state, based on the prior color gamut range of the kiwi fruit and the shelf on which it is placed, to generate a high-contrast grayscale image. A boundary determination unit is used to initially locate the first initial contour boundary of the kiwi fruit and the second initial contour boundary of the juice-stained area in the high-contrast grayscale image by using a dual gradient threshold. The mask determination unit is used to perform cubic spline interpolation fitting on the pixels of the first initial contour boundary and the second initial contour boundary respectively to locate the precise contour boundary at the sub-pixel level, and generate the first precise segmentation mask of the kiwi fruit and the second precise segmentation mask of the juice-stained area. At the same time, based on the prior space size of the shelf and the fixed coordinates inside the carriage, a fixed mask of the shelf area is generated. The pixel segmentation unit is used to extract all valid pixels within the first precise segmentation mask, perform local binary pattern texture feature calculation on each valid pixel, and divide the normal fruit peel pixel set into the abnormal pixel set. The connectivity analysis unit is used to perform connectivity analysis on the abnormal pixel set. Combining the texture threshold and grayscale threshold of the normal skin of kiwifruit, it filters out the target connected components that meet the damage characteristics and extracts the pixel area, perimeter, grayscale mean, texture roughness, and boundary gradient parameters of the target connected components to generate the damage characteristics of the kiwifruit itself. The point extraction unit is used to extract all valid pixels in the overlapping area of ​​the fixed mask and the second precise segmentation mask, which are regarded as the points to be analyzed. The erosion feature unit is used to calculate the deviation of each point to be analyzed relative to the reference gray value of the shelf, generate a pixel gray-level deviation matrix, and perform boundary gradient change analysis on the pixel gray-level deviation matrix based on the diffusion characteristics of sap contamination. It locates the sub-pixel boundary of the sap contamination infiltration front, calculates the pixel coverage area, infiltration depth, and boundary diffusion rate parameters of the infiltrated area, and combines the damage characteristics of kiwifruit in the same area to match the source fruit of sap seepage, generating liquid erosion features of damaged kiwifruit sap seeping onto the corresponding shelf.

[0063] In this embodiment, the prior color gamut range is: kiwi fruit (R channel 100-220, G channel 50-180, B channel 30-150); 304 stainless steel white shelf (R channel 180-255, G channel 180-255, B channel 180-255). The color gamut weighted grayscale mapping formula is: Gray=0.6×R+0.3×G+0.1×B. Based on the color gamut characteristics of kiwifruit and the shelf, the weight of the R channel is increased to enhance the contrast between the fruit and the shelf and the juice.

[0064] In this embodiment, a dual-gradient threshold is used: the high threshold is set to 80 and the low threshold is set to 30, which effectively filters out false edges caused by noise while preserving weak edges of fruit and juice. The subpixel edge positioning accuracy can reach 0.1 pixels, which can accurately capture the blurred boundaries of early traces of sap, greatly improving recognition accuracy.

[0065] In this embodiment, the local binary mode uses the LBP operator, which is a circular operator with a radius of 1 and a neighborhood of 8. The LBP feature value range of normal fruit skin pixels is [10, 50], and pixels outside this range are judged as abnormal pixels.

[0066] In this embodiment, the damage characteristics of the kiwifruit itself include: the number of damaged fruits, the total damaged area, the average damaged perimeter, the damage grade, the texture roughness, and the average grayscale value. The damage grade is divided into: minor damage (damaged area < 0.5 mm). Minor damage (0.5 mm) ≤damaged area<2 Severe damage (damaged area ≥ 2) ).

[0067] In this embodiment, the characteristics of liquid erosion include: total area of ​​sap infection, location of the infiltration front, diffusion rate, infection level, and location of the source fruit. The infection level is divided into: trace infection (area < 10... ), mild infection (10) Area < 50 Severe infection (area ≥ 50) ).

[0068] In this embodiment, the 8-neighborhood connectivity rule is used, combined with the texture threshold and grayscale threshold of the normal kiwifruit skin, to filter out target connected regions that meet the damage characteristics. The pixel-physical size mapping relationship is calibrated by the shelf fixed verification code, and the conversion formula is: actual size (cm) = number of pixels × calibration coefficient (cm / pixel).

[0069] In this embodiment, the formula for calculating the immersion depth is as follows: ,in, This represents the deviation between the point to be analyzed and the reference grayscale value of the shelf. The conversion factor is calibrated, with the dimension being mm / grayscale value.

[0070] Formula for calculating boundary diffusion rate: ,in, unit of time The change in the area of ​​internal infection; This refers to the total area of ​​the shelf surface where the corresponding shelves will be placed.

[0071] The beneficial effects of the above technical solution are: through high-precision image segmentation, texture analysis and juice contamination feature extraction, it can simultaneously identify potential fruit damage and juice erosion risks, predict the risk of chain rot in advance, and identify anomalies earlier and with higher accuracy.

[0072] This invention provides a kiwifruit storage and preservation monitoring system based on dynamic controlled atmosphere and cold chain traceability. The feature and anomaly analysis module further includes: The coefficient generation unit is used to retrieve the basic storage and preservation vector of the corresponding shelf area at the current moment, and combine it with the cumulative transportation time from the initial transportation time to the current moment to generate the environmental impact coefficient of quality deterioration of the corresponding shelf. The quality analysis unit is used to perform pixel-level benchmark matching between the extracted kiwi fruit damage characteristics and the liquid erosion characteristics of the kiwi fruit damage and oozing juice on the corresponding shelf and the initial quality benchmark library of the corresponding shelf. It distinguishes between the fruit damage characteristics and the juice erosion characteristics added during transportation, calculates the abnormal deterioration level, real-time deterioration rate and juice contamination and diffusion risk coefficient of the kiwi fruit in the corresponding shelf, and completes the quality abnormality analysis of the kiwi fruit on the corresponding shelf relative to the initial loading state. The coordinate construction unit is used to construct a unified two-dimensional spatial coordinate system within the transport compartment based on the spatial arrangement coordinates of each shelf in the transport compartment and the monitoring coverage area boundaries of all activated monitoring components, with the pre-calibrated unique fixed verification code of each shelf as the spatial positioning reference anchor point. The parameter mapping unit is used to map the abnormal deterioration level, real-time deterioration rate, juice contamination and diffusion risk coefficient, and quality deterioration environmental impact coefficient of each placed shelf to the corresponding position in the two-dimensional spatial coordinate system, thereby generating a single shelf abnormal attribute node. The correction unit is used to construct an abnormal risk transmission matrix between adjacent shelves based on the cross-shelf risk transmission characteristics of kiwifruit rotten juice contamination and ethylene gas diffusion. It performs adjacent correction on the risk coefficient of each single shelf abnormal attribute node, and generates a kiwifruit abnormal distribution layer with spatial positioning, abnormal level, deterioration trend and diffusion risk attributes based on all corrected single shelf abnormal attribute nodes and the monitoring coverage boundary of each activated monitoring component.

[0073] In this embodiment, a fixed verification code is used: a unique fixed verification code is set at a preset position in the upper left corner of the shelf surface. The fixed verification code is a QR code, 50mm×50mm in size, printed on a corrosion-resistant and stain-resistant food-grade PVC substrate. The fixed verification code stores the unique ID of the corresponding shelf, its physical size, and its relative fixed coordinates in the vehicle compartment. Before transportation, the fixed verification codes are pre-calibrated, including the physical size of the verification code, the pixel-to-actual physical size mapping relationship, and the relative coordinates of the verification code in the vehicle compartment. A priori verification database for the corresponding shelf is generated and stored in the vehicle's central control system. In this embodiment, the method for constructing the abnormal risk transmission matrix between adjacent shelves is as follows: With the target shelf as the center, the closer the distance between adjacent shelves, the higher the transmission coefficient; the higher the anomaly level of the target shelf, the greater the transmission coefficient; the transmission coefficient ranges from 0 to 1, and the formula is: ,in, Let i1 be the risk transmission coefficient from shelf i1 to shelf j1. The spatial distance between shelf i1 and shelf j1 The maximum diagonal length of the carriage. The abnormality level of shelf i1, This is the highest level of abnormality, with a fixed value of 3.

[0074] In this embodiment, the adjacent correction formula is as follows: ,in, The original risk coefficient for shelf j1; This is the corrected risk coefficient.

[0075] In this embodiment, the formula for calculating the environmental impact coefficient of quality deterioration is: , , , , These are the baseline values ​​for the parameters corresponding to the standard preservation vector.

[0076] In this embodiment, the difference between the pixel grayscale value and LBP texture feature value of the real-time image and the corresponding pixel value in the initial benchmark library is calculated. Pixels whose difference exceeds a preset threshold are identified as newly added abnormal pixels. The preset threshold for grayscale difference is fixed at 20, with a value range of 15~30. When the absolute value of the difference between the grayscale value of the real-time image pixel and the benchmark grayscale value of the corresponding pixel in the initial quality benchmark library is ≥ the threshold, it is identified as a newly added abnormal pixel. The preset threshold for texture feature difference is fixed at 10, with a value range of 5~15. When the absolute value of the difference between the LBP texture feature value of the real-time image pixel and the benchmark texture value of the corresponding pixel in the initial quality benchmark library is ≥ the threshold, it is identified as a newly added abnormal pixel. The above thresholds are set based on measured data of the normal texture fluctuation range of kiwi fruit skin and the benchmark fluctuation range of stainless steel shelf cleanliness. This can effectively filter out normal fluctuations in pixel values ​​caused by changes in light and slight vibrations during transportation, while accurately identifying real abnormalities caused by fruit damage and juice contamination.

[0077] In this embodiment, the formula for calculating the real-time degradation rate is as follows: ,in, unit of time The change in the area of ​​internal damage; This represents the total surface area of ​​all the fruit on the corresponding shelf; Formula for calculating the risk factor of sap contamination and spread: ,in, This represents the total area affected.

[0078] The beneficial effects of the above technical solution are: by calculating the environmental impact coefficient, mapping spatial coordinates and correcting cross-shelf risk transmission, an anomaly distribution layer with spatial positioning is generated, realizing global risk visualization and accurate judgment, and improving the efficiency of anomaly handling.

[0079] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A kiwifruit storage and preservation monitoring system based on dynamic controlled atmosphere and cold chain traceability, characterized in that, include: The vector acquisition module is used to acquire the preset kiwifruit cold chain transportation route and the real-time location of the transport driver during the transportation process according to the cold chain transportation route. Combined with the weather state vector corresponding to the real-time location, the road segment state vector of the current road segment, the pre-transport packaging state vector of the kiwifruit, and the basic storage and preservation vector of each shelf area at the current moment, the module generates the monitoring state vector corresponding to each shelf at the real-time location. The pre-transport packaging state vector includes the initial loading and storage state of the kiwifruit on each shelf in the transport vehicle, and the basic storage and preservation vector includes various environmental parameters of the corresponding shelf area. The dynamic controlled atmosphere module is used to dynamically adjust the regional controlled atmosphere adjustment unit built into the transport vehicle according to the monitoring status vector corresponding to each shelf and the real-time storage and preservation requirements of the kiwifruit on the corresponding shelf. At the same time, it dynamically controls the start and stop of several monitoring components pre-set in the transport vehicle according to the placement position of each shelf in the transport vehicle and the monitoring status vector of the corresponding shelf. The feature and anomaly analysis module is used to extract the damage features of the kiwifruit itself and the liquid erosion features of the kiwifruit juice seeping out from the images captured by each monitoring component in the active state. Combined with the basic storage and preservation vector of the corresponding shelf area at the current moment, it analyzes the quality anomalies of the kiwifruit on the corresponding shelf relative to the initial loading state, constructs an anomaly distribution layer of kiwifruit based on the monitoring range covered by all active monitoring components, and determines whether the kiwifruit meets the set replacement criteria based on the anomaly distribution layer. If the conditions are met, the location information of the cold chain warehouse closest to the real-time location is sent to the vehicle terminal, and the inspection of the abnormal batch of kiwifruit and the corresponding full-process data are recorded once when the transport vehicle arrives at the cold chain warehouse; if the conditions are not met, when all the kiwifruit monitored within the full coverage area of ​​the cold chain transport route corresponding to the same cold chain warehouse are in a condition that does not meet the set replacement criteria, the full-process data of the batch of kiwifruit is recorded a second time. The traceability chain construction module is used to generate a complete cold chain traceability chain for the transported batch of kiwifruit based on the results of primary recording, secondary recording, real-time transportation information along the cold chain transportation route, and regional controlled atmosphere regulation data. It also dynamically regulates the kiwifruit storage in each cold chain warehouse according to the full-process storage data of the cold chain traceability chain.

2. The kiwifruit storage and preservation monitoring system based on dynamic controlled atmosphere and cold chain traceability according to claim 1, characterized in that, The environmental parameters include: concentration, Concentration, ethylene concentration, ambient temperature, and ambient humidity.

3. The kiwifruit storage and preservation monitoring system based on dynamic controlled atmosphere and cold chain traceability according to claim 1, characterized in that, Each monitoring component corresponds to a monitoring area inside the transport vehicle where different shelves are placed. The total number of monitoring components is N, where N is a positive integer and not less than the total number of shelves placed in the transport vehicle.

4. The kiwifruit storage and preservation monitoring system based on dynamic controlled atmosphere and cold chain traceability according to claim 1, characterized in that, The vector acquisition module includes: The model analysis submodule is used to input the weather state vector, road segment state vector, pre-transport packaging state vector, and basic storage and preservation vector into the corresponding pre-trained neural network model to obtain the first factor affecting the storage and preservation effect in each vector and the storage and preservation confidence of each first factor. The mapping processing submodule is used to map each predefined indicator in the monitoring status vector to the first factor under each vector to obtain a set of candidate factors for each predefined indicator. The sorting submodule is used to obtain a second factor from the candidate factor set whose storage and preservation confidence is greater than a preset confidence level. At the same time, the candidate factor set is sorted first according to the mapping correlation coefficient and sorted second according to the storage and preservation confidence level. The cutting submodule is used to perform a first forward cut on the second sorting result to obtain several third factors, taking the first element in the first sort as the first cutting point; at the same time, it performs a second forward cut on the first sorting result to obtain several fourth factors, taking the first element in the second sort as the second cutting point. The direction matching submodule is used to determine the first preservation anomaly direction based on all second factors, the second preservation anomaly direction based on all third factors, and the third preservation anomaly direction based on all fourth factors based on the factor-direction comparison table, thus forming a state anomaly subset corresponding to the predefined indicators. The relative damage determination submodule is used to determine the relative damage factors of the kiwifruit placed on the corresponding shelf based on the basic storage and preservation vector and the standard preservation vector of the pre-transportation packaging state vector, and in combination with the cumulative transportation time from the initial transportation time to the current time, the variety attributes of the kiwifruit placed on the shelf, and the post-harvest respiratory physiological characteristics. The adjustment submodule is used to adjust the maximum and minimum abnormal directions among the first, second, and third abnormal directions of preservation under the corresponding predefined indicators according to the relative damage factors of the kiwifruit, thereby forming a state abnormal subset of the corresponding predefined indicators. The vector combination submodule is used to obtain the monitoring status vector corresponding to each placed shelf based on the abnormal status subset of all predefined indicators.

5. The kiwifruit storage and preservation monitoring system based on dynamic controlled atmosphere and cold chain traceability according to claim 4, characterized in that, The relative damage determination submodule includes: The spatial determination unit is used to construct a multi-dimensional post-harvest storage quality evaluation system and a respiratory aging benchmark model for kiwifruit based on the varietal attributes and post-harvest respiratory physiological characteristics of the kiwifruit placed on the corresponding shelf. The core indicators of the storage quality evaluation system are used as dimensions to determine the sampling space of preservation parameters for the corresponding kiwifruit varieties. The sample acquisition unit is used to generate a storage and preservation benchmark sample set and a real-time quality evaluation sample set for the corresponding variety of kiwifruit within the preservation parameter sampling space, and to map the storage and preservation benchmark sample set and the real-time quality evaluation sample set to the standardized preservation evaluation space respectively, thereby obtaining the benchmark standardized sample and the real-time standardized sample. The correction unit is used to calculate the baseline storage quality response value of the baseline standardized sample based on the respiratory aging baseline model. At the same time, it dynamically corrects the baseline storage quality response value by combining the cumulative transportation time from the initial transportation time to the current time and the time decay coefficient output by the respiratory aging baseline model, so as to obtain the allowable quality deviation threshold range of kiwifruit under the corresponding shelf placement and corresponding cumulative transportation time. The boundary expansion unit is used to determine the boundary expansion range corresponding to the upper and lower limits of the allowable quality deviation threshold range based on the storage quality fluctuation tolerance of the corresponding kiwifruit variety and the disturbance coefficient of the transportation route conditions, and to generate the boundary expansion range. The deviation unit is used to calculate the weighted Mahalanobis distance between the real-time standardized sample and the benchmark standardized sample, taking the core indicators of the storage quality evaluation system as the calculation dimension and combining the physiological influence weights of each core indicator on the storage quality of kiwifruit, and to generate the real-time preservation deviation vector corresponding to each core indicator. The division unit is used to continuously divide the allowable quality deviation threshold range and the boundary expansion range based on the sensitivity of kiwifruit quality deterioration. Specifically, near the upper and lower limits of the allowable quality deviation threshold range and at the junction of the boundary expansion range and the allowable range, the minimum preset step size is used to divide the sub-processing range; in the middle region of the allowable quality deviation threshold range, the maximum preset step size is used to divide the sub-processing range, ultimately obtaining several continuous sub-processing ranges with step sizes adapted to the sensitivity of deterioration. The probability calculation unit is used to perform a Gaussian-cosine composite probability transformation on the real-time preservation deviation vector of each sub-processing interval that falls within the allowable quality deviation threshold range, so as to obtain the in-interval deviation deterioration probability value of the corresponding sub-processing interval. At the same time, for the real-time preservation deviation vector that exceeds the allowable quality deviation threshold range and falls into the boundary extension range, a piecewise decaying power composite probability transformation is performed to obtain the out-of-limit deviation deterioration probability value of the corresponding sub-processing interval. The relatively defined unit is used to construct a quality degradation probability sequence of kiwifruit placed on the shelf based on the in-interval deviation degradation probability value and the out-of-limit deviation degradation probability value corresponding to each sub-processing interval, according to the arrangement order of the sub-processing intervals. Combining the deviation direction, physiological influence weight and degradation probability contribution of all real-time preservation deviation vectors corresponding to the same core indicator, the core influencing indicators and degradation dominant factors are screened out to generate the relative damage factors of kiwifruit placed on the shelf.

6. The kiwifruit storage and preservation monitoring system based on dynamic controlled atmosphere and cold chain traceability according to claim 1, characterized in that, The feature and anomaly analysis module includes: The grayscale mapping unit is used to perform grayscale mapping on the images captured by the monitoring component when it is in the on state, based on the prior color gamut range of the kiwi fruit and the shelf on which it is placed, to generate a high-contrast grayscale image. A boundary determination unit is used to initially locate the first initial contour boundary of the kiwi fruit and the second initial contour boundary of the juice-stained area in the high-contrast grayscale image by using a dual gradient threshold. The mask determination unit is used to perform cubic spline interpolation fitting on the pixels of the first initial contour boundary and the second initial contour boundary respectively to locate the precise contour boundary at the sub-pixel level, and generate the first precise segmentation mask of the kiwi fruit and the second precise segmentation mask of the juice-stained area. At the same time, based on the prior space size of the shelf and the fixed coordinates inside the carriage, a fixed mask of the shelf area is generated. The pixel segmentation unit is used to extract all valid pixels within the first precise segmentation mask, perform local binary pattern texture feature calculation on each valid pixel, and divide the normal fruit peel pixel set into the abnormal pixel set. The connectivity analysis unit is used to perform connectivity analysis on the abnormal pixel set. Combining the texture threshold and grayscale threshold of the normal skin of kiwifruit, it filters out the target connected components that meet the damage characteristics and extracts the pixel area, perimeter, grayscale mean, texture roughness, and boundary gradient parameters of the target connected components to generate the damage characteristics of the kiwifruit itself. The point extraction unit is used to extract all valid pixels in the overlapping area of ​​the fixed mask and the second precise segmentation mask, which are regarded as the points to be analyzed. The erosion feature unit is used to calculate the deviation of each point to be analyzed relative to the reference gray value of the shelf, generate a pixel gray-level deviation matrix, and perform boundary gradient change analysis on the pixel gray-level deviation matrix based on the diffusion characteristics of sap contamination. It locates the sub-pixel boundary of the sap contamination infiltration front, calculates the pixel coverage area, infiltration depth, and boundary diffusion rate parameters of the infiltrated area, and combines the damage characteristics of kiwifruit in the same area to match the source fruit of sap seepage, generating liquid erosion features of damaged kiwifruit sap seeping onto the corresponding shelf.

7. The kiwifruit storage and preservation monitoring system based on dynamic controlled atmosphere and cold chain traceability according to claim 1, characterized in that, The feature and anomaly analysis module also includes: The coefficient generation unit is used to retrieve the basic storage and preservation vector of the corresponding shelf area at the current moment, and combine it with the cumulative transportation time from the initial transportation time to the current moment to generate the environmental impact coefficient of quality deterioration of the corresponding shelf. The quality analysis unit is used to perform pixel-level benchmark matching between the extracted kiwi fruit damage characteristics and the liquid erosion characteristics of the kiwi fruit damage and oozing juice on the corresponding shelf and the initial quality benchmark library of the corresponding shelf. It distinguishes between the fruit damage characteristics and the juice erosion characteristics added during transportation, calculates the abnormal deterioration level, real-time deterioration rate and juice contamination and diffusion risk coefficient of the kiwi fruit in the corresponding shelf, and completes the quality abnormality analysis of the kiwi fruit on the corresponding shelf relative to the initial loading state. The coordinate construction unit is used to construct a unified two-dimensional spatial coordinate system within the transport compartment based on the spatial arrangement coordinates of each shelf in the transport compartment and the monitoring coverage area boundaries of all activated monitoring components, with the pre-calibrated unique fixed verification code of each shelf as the spatial positioning reference anchor point. The parameter mapping unit is used to map the abnormal deterioration level, real-time deterioration rate, juice contamination and diffusion risk coefficient, and quality deterioration environmental impact coefficient of each placed shelf to the corresponding position in the two-dimensional spatial coordinate system, thereby generating a single shelf abnormal attribute node. The correction unit is used to construct an abnormal risk transmission matrix between adjacent shelves based on the cross-shelf risk transmission characteristics of kiwifruit rotten juice contamination and ethylene gas diffusion. It performs adjacent correction on the risk coefficient of each single shelf abnormal attribute node, and generates a kiwifruit abnormal distribution layer with spatial positioning, abnormal level, deterioration trend and diffusion risk attributes based on all corrected single shelf abnormal attribute nodes and the monitoring coverage boundary of each activated monitoring component.

8. The kiwifruit storage and preservation monitoring system based on dynamic controlled atmosphere and cold chain traceability according to claim 7, characterized in that, The initial quality benchmark library includes the initial damage characteristic parameters of the kiwifruit in the corresponding shelf, the initial skin texture benchmark, the initial cleanliness benchmark of the corresponding shelf, and the postharvest respiration deterioration benchmark model of this variety of kiwifruit.

Images