System and method for dynamically monitoring quality of harvested agricultural products and intelligently grading and pricing agricultural products

Abstract

The invention provides a postharvest quality dynamic monitoring and intelligent grading pricing system and method for agricultural products, and the method comprises the steps: extracting a high-intensity conduction region from an influence coefficient matrix, judging the quality deterioration risk level of the agricultural products, and obtaining a risk distribution mapping which is used for visualizing a potential quality reduction hotspot region in a warehouse; aiming at the risk distribution mapping, evaluating the current quality state of the agricultural products by adopting a classification algorithm, and determining classification categories for distinguishing batches of the agricultural products with different maturity degrees and fresh-keeping degrees; after the grading categories are obtained, a storage layout optimization model is updated in combination with the ethylene conduction path data, and optimized coexistence configuration is obtained and used for adjusting a goods shelf placement strategy to reduce ethylene cross contamination; and iteratively updating the local environment distribution diagram through the prevention adjustment scheme, quantifying the overall supply chain efficiency improvement effect, and determining a final grading fixed value for guiding agricultural product storage and logistics decision.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a system and method for post-harvest quality dynamic monitoring and intelligent grading and pricing of agricultural products. Background Technology

[0002] The research field of post-harvest quality dynamic monitoring and intelligent grading and pricing systems and methods for agricultural products is related to the entire process of preservation and value realization of fresh fruits and vegetables from harvest to sale. Its core importance lies in directly determining farmers' income, distribution losses, and whether consumers can buy high-quality products. In modern agricultural warehousing and supply chains, how to enable different types of agricultural products to coexist in a limited space and maintain their optimal condition as much as possible has become a key link affecting the efficiency of the entire industry.

[0003] Most current monitoring and pricing schemes are designed for the independent storage of single varieties, with little consideration for the actual situation when multiple agricultural products are stored together. This approach ignores the real interactions between different agricultural products, leading to inconsistent rates of quality change in actual warehouses. For example, some fruits release ethylene gas, which significantly accelerates the ripening or prematurely ages other fruits and vegetables in the vicinity. However, existing systems often fail to capture the gradual spread of this effect from one shelf to adjacent shelves, resulting in systemic biases in quality assessment and pricing.

[0004] While ethylene release and other gas interactions are superficial, the deeper technical challenge lies in the strong spatial dependence and transmissive characteristics of the interactions between agricultural products. Changes in the quality of one product can gradually propagate to surrounding areas through subtle gradients in local temperature, humidity, and gas concentration. The path and intensity of this transmission vary significantly depending on shelf placement, stacking density, and ventilation conditions. Existing monitoring methods struggle to accurately characterize the strength of these spatially increasing or decreasing correlations, and it is difficult to quantify the extent and distance of a negative transmission effect of a particular agricultural product on neighboring products within a specific combination. This makes it impossible for warehouse managers to predict the cascading risks associated with mixed storage.

[0005] Therefore, in real-time scenarios of mixed storage, how to identify and quantify the mutual influence between different types of agricultural products through the spatial distribution of gas, temperature and humidity, especially to capture the transmission path and specific intensity of a product's quality deterioration on neighboring products, has become a key issue for achieving accurate dynamic quality monitoring and reasonable intelligent graded pricing. Summary of the Invention

[0006] This invention provides a system and method for post-harvest quality dynamic monitoring and intelligent grading and pricing of agricultural products, mainly including: By using sensor array devices deployed in the warehouse to collect real-time data on ethylene gas concentration, ambient temperature, and air humidity at each shelf location, a local environmental distribution map of the warehouse is generated to reflect the microclimate conditions of the agricultural product storage area. Based on the local environmental distribution map, a neural network algorithm is used to analyze the gas concentration gradient change pattern, determine the starting point and propagation direction of the ethylene conduction path, and track potential sources affecting the ripening of agricultural products. If the starting point of the ethylene transmission path is located on a shelf storing specific agricultural products susceptible to ethylene, the intensity of the ethylene influence of the path on adjacent shelves is quantified by an ensemble learning algorithm to obtain an influence coefficient matrix, which is used to represent the degree of interaction between regions. High-intensity transmission regions are extracted from the influence coefficient matrix to determine the risk level of agricultural product quality deterioration and obtain a risk distribution map, which is used to visualize potential hotspots of quality decline in warehouses. Based on the aforementioned risk distribution mapping, a classification algorithm is used to assess the current quality status of agricultural products and determine grading categories to distinguish batches of agricultural products with different levels of maturity and freshness. After obtaining the classification, the warehouse layout optimization model is updated in conjunction with the ethylene transmission path data to obtain an optimized coexistence configuration, which is used to adjust the shelf placement strategy to reduce ethylene cross-contamination. Based on the optimized coexistence configuration, the propagation process of the ethylene transmission effect in the warehouse is simulated to determine potential chain risks and obtain prevention and adjustment plans, which are used to propose targeted storage environment intervention measures. By iteratively updating the local environmental distribution map through the aforementioned prevention and adjustment scheme, the overall supply chain efficiency improvement effect is quantified, and the final grading and value determination is made to guide agricultural product storage and logistics decisions.

[0007] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: This invention discloses a method for preventing and controlling quality risks in agricultural product storage based on real-time multi-dimensional environmental perception and intelligent tracking of ethylene transmission. The method involves real-time collection of ethylene concentration, temperature, and humidity data at various shelf locations using an array of sensors within the warehouse, generating a local environmental distribution map reflecting the microclimate of agricultural product storage. Based on this map, a neural network is used to analyze the ethylene concentration gradient change pattern, accurately determining the starting point and propagation direction of the ethylene transmission path, thereby tracking potential sources of mature influence. When the starting point is located on a specific agricultural product shelf susceptible to ethylene, ensemble learning is used to quantify the intensity of the path's impact on adjacent shelves, forming an influence coefficient matrix. High-intensity transmission areas are extracted from this matrix, generating a risk distribution map to visualize quality deterioration hotspots. After assessing the current quality status of each batch and determining its classification category using a classification algorithm, the warehouse layout is iteratively optimized by integrating ethylene transmission path data to obtain a shelf coexistence configuration that reduces cross-contamination. Furthermore, the future ethylene transmission process is simulated to identify cascading risks and generate preventative adjustment plans. The environmental distribution is updated iteratively through a closed-loop system of these plans, ultimately achieving overall supply chain efficiency improvement and precise classification guidance. The core of this invention lies in combining real-time ethylene gas transmission path tracking with multi-dimensional impact quantification, thereby achieving dynamic and accurate identification, source tracing, and proactive intervention and control of the risk of quality deterioration in agricultural product storage. Attached Figure Description

[0008] Figure 1 This is a flowchart of a post-harvest quality dynamic monitoring and intelligent grading and pricing system and method for agricultural products according to the present invention.

[0009] Figure 2 This is a schematic diagram of a post-harvest quality dynamic monitoring and intelligent grading and pricing system and method for agricultural products according to the present invention.

[0010] Figure 3 This is another schematic diagram of a post-harvest quality dynamic monitoring and intelligent grading and pricing system and method for agricultural products according to the present invention. Detailed Implementation

[0011] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0012] like Figures 1-3 This embodiment of a post-harvest quality dynamic monitoring and intelligent grading and pricing system and method for agricultural products may specifically include: Step S101: Real-time data on ethylene gas concentration, ambient temperature, and air humidity at each shelf location are collected using sensor array devices deployed in the warehouse to generate a local environmental distribution map inside the warehouse, which reflects the microclimate conditions of the agricultural product storage area.

[0013] The system acquires raw data on ethylene concentration, temperature, and humidity at each shelf location from a sensor array. Synchronous concentration, temperature, and humidity sequences for each location are obtained using a time-series alignment method. A three-dimensional environmental state vector for each shelf location at the current moment is calculated based on these sequences. A continuous ethylene concentration, temperature, and humidity distribution field within the warehouse is generated using the Kriging interpolation method based on the three-dimensional environmental state vectors of all shelf locations. The coordinates of areas with current concentrations exceeding a preset threshold are extracted from the ethylene concentration distribution field, resulting in a set of high-concentration areas. Spatial overlay analysis is performed on the high-concentration area set with the temperature and humidity distribution fields to determine if there are sub-regions with both high temperature and low humidity within the high-concentration areas. If such sub-regions exist, they are marked as key areas of interest, and their spatial extent is recorded. Based on the spatial extent of the key areas of interest, the concentration change sequence at the corresponding location is extracted from the ethylene concentration distribution field, and the rate of increase of this sequence over a recent period is calculated. If the rate of increase exceeds a preset threshold, the key area of ​​interest is determined to be in a state of accelerated ethylene accumulation. For key areas of concern experiencing accelerated ethylene accumulation, temperature and humidity change sequences are extracted from the temperature and humidity distribution fields. These sequences are then used to determine whether current microclimate conditions further promote ethylene release. If the determination indicates promotion of release, a local environmental anomaly alert is generated, including the location of the key area of ​​concern, ethylene concentration, temperature, humidity, and the state of accelerated accumulation.

[0014] For example, in a warehouse environment, collecting ethylene concentration, temperature, and humidity data at various shelf locations using sensor arrays is fundamental for environmental monitoring. Assuming a warehouse has 10 shelf locations, with each sensor recording data once per minute, the resulting data series would be ethylene concentration in milligrams per cubic meter, temperature in degrees Celsius, and humidity in percentage. Time series alignment methods can be used to match timestamps, ensuring data synchronization across locations and creating concentration, temperature, and humidity sequences.

[0015] Specifically, when calculating the three-dimensional environmental state vector, the current concentration, temperature, and humidity values ​​at each shelf location can be combined into a three-dimensional vector. For example, the vector for a certain location might be a concentration of 5.2, a temperature of 25.5, and a humidity of 40%. Based on these vectors, the Kriging interpolation method is used to generate a continuous distribution field within the warehouse. Kriging is a spatial statistical method that predicts unknown point values ​​using known point data, and is suitable for generating continuous distribution fields of ethylene concentration, temperature, and humidity. This method effectively smooths the data, provides a more accurate spatial distribution view, and helps identify potential anomaly areas.

[0016] In one embodiment, areas with ethylene concentrations exceeding a preset threshold, such as 4.0 mg / m³, are extracted from the ethylene concentration distribution field to obtain a set of high-concentration areas. Assume a certain area in a warehouse has a concentration of 5.5 mg / m³, located between shelves 3 and 5. Through spatial overlay analysis, this area is compared with temperature and humidity distribution fields. If the temperature in this area is found to be above 28 degrees Celsius and the humidity below 35%, it is marked as an area of ​​high concern. This analysis helps identify microclimate conditions that may accelerate the spoilage of fruits and vegetables.

[0017] For example, for key areas of concern, concentration change sequences are extracted from the concentration distribution field. If, for instance, the concentration rises from 4.5 mg / m³ to 5.5 mg / m³ in the past hour, exceeding a preset threshold of 0.8 mg / m³ / hour, it is determined to be a state of accelerated ethylene accumulation. This determination can provide timely warnings of potential risks. Further, change sequences are extracted from the temperature and humidity distribution fields. If the temperature continues to rise (e.g., from 26 to 29 degrees Celsius) and the humidity continues to decrease (e.g., from 40% to 33%), it is determined that the current microclimate conditions are promoting ethylene release. Based on this, an anomaly alert is generated, including location, concentration of 5.5 mg / m³, temperature of 29 degrees Celsius, humidity of 33%, and the state of accelerated accumulation.

[0018] In one embodiment, the technical benefits of this multi-dimensional analysis are significant. By comprehensively monitoring concentration, temperature, and humidity, problem areas can be accurately located, allowing for timely ventilation or cooling measures to reduce fruit and vegetable losses. The combination of spatial overlay analysis and change sequence analysis enables a comprehensive assessment of environmental conditions from static distribution to dynamic trends, enhancing the accuracy and timeliness of early warnings and providing a scientific basis for warehouse management.

[0019] Step S102: Based on the local environmental distribution map, a neural network algorithm is used to analyze the gas concentration gradient change pattern to determine the starting point and propagation direction of the ethylene conduction path, which is used to track potential sources affecting the ripening of agricultural products.

[0020] The locations of monitoring points and corresponding ethylene gas concentrations in the local environmental distribution map are obtained. Based on the locations of each monitoring point and the ethylene gas concentration data, a neural network algorithm is used to calculate the concentration difference between adjacent points, obtaining a gas concentration gradient field. By analyzing the direction and magnitude of the gradient vector in the gas concentration gradient field, the reverse direction of the local maximum gradient at each point is determined, yielding the initial direction of ethylene conduction paths. The region with the highest gradient vector continuity is selected from the initial ethylene conduction path directions to obtain a set of candidate ethylene conduction paths. For each path in the candidate ethylene conduction path set, the path is traced back point by point from its end along the reverse direction of the gradient to determine whether the concentration continues to increase. If the concentration continues to increase, the path is retained; if the concentration decreases, the path is discarded, resulting in a selected ethylene conduction path. The starting endpoint where the concentration gradient vector converges is found in the selected ethylene conduction path, and this endpoint is determined as the starting position of the ethylene conduction path. Using the starting position of the ethylene conduction path as the source point, the direction of ethylene propagation and the location of potential sources affecting agricultural product ripening are obtained.

[0021] For example, in a cold storage warehouse for fruits and vegetables, the location coordinates of each monitoring point on the shelves and their corresponding ethylene concentration values ​​are acquired in real time. Assume a warehouse has twenty sensor points, where the concentration at the center of shelf area A is 0.8 ppm, and at the edge of adjacent area B it is 0.3 ppm. Based on this discrete data, a neural network learns the concentration differences between adjacent points and outputs a concentration gradient field between them. The gradient field shows the direction and magnitude of the vector from low-concentration areas to high-concentration areas; regions with larger gradient values ​​indicate drastic changes in ethylene concentration.

[0022] Specifically, in a gas concentration gradient field, the gradient vector points to the path of the fastest increase in concentration, and the opposite direction is the direction of the fastest decrease in concentration.

[0023] In one possible implementation, the system uses the reverse direction of the local maximum gradient at each monitoring point as the initial ethylene conduction path direction.

[0024] For example, if the gradient vector at a point in shelf area C points northeast and has a magnitude of 0.4 ppm / m, then the opposite direction, southwest, is the initial direction of the ethylene conduction path. By integrating the directional information of all points, the region with the highest gradient vector continuity is selected, forming a set of several candidate ethylene conduction paths. These paths typically span multiple shelves, forming potential ethylene diffusion channels.

[0025] Preferably, for each candidate path, backtracking is performed point by point from the end along the reverse direction of the gradient. Assuming a path has a concentration of 0.2 ppm at the end, rises to 0.5 ppm at the penultimate point, and to 1.1 ppm at the third point, with the concentration continuously increasing throughout the backtracking process, this path is retained. If a path experiences a concentration drop from 1.2 ppm to 0.9 ppm midway through, the path is discarded. This screening process yields selected ethylene conduction paths that reflect the true trend of ethylene diffusion from the source outwards.

[0026] In one embodiment, the endpoint where multiple gradient vectors converge in the selected path is determined as the starting position.

[0027] For example, the gradient vectors of all three paths point to an apple shelf area in the southwest corner of the warehouse, which is marked as the starting point of the ethylene transmission path. This location often corresponds to the area of ​​ripe fruit where local ethylene release is most intense. By using this starting point as the source and extending forward along the selected path, the main direction of ethylene propagation and the potential range of impact can be inferred.

[0028] For example, the concentration remained above 0.7 ppm even after extending about 4 meters northeast, indicating that the path may have affected nearby banana and mango shelves.

[0029] It's important to note that this gradient field-based transmission path analysis can pinpoint the source of ethylene release in advance, preventing a large-scale increase in concentration throughout the warehouse. By identifying the starting point and direction of propagation, managers can prioritize ventilating, relocating, or removing the source shelves early, effectively mitigating the ripening effect of ethylene on surrounding agricultural products and extending their overall shelf life. This method, from discrete point data to a logical chain of continuous path inference, provides precise micro-level traceability capabilities, facilitating proactive control of ethylene risks within warehouses.

[0030] Step S103: If the starting point of the ethylene conduction path is located on a shelf storing specific agricultural products susceptible to ethylene, then the intensity of the ethylene influence of the path on adjacent shelves is quantified by an ensemble learning algorithm to obtain an influence coefficient matrix, which is used to represent the degree of interaction between regions.

[0031] Real-time ethylene concentration data for each shelf in the shelving layout is acquired. An ethylene transmission path is determined based on a preset concentration difference. If the concentration difference between adjacent shelves exceeds a preset threshold, the path is confirmed and the starting shelf position is recorded. For the starting shelf of a confirmed ethylene transmission path, the sequence of its downstream neighboring shelves is extracted to form a path-related shelf group. An ensemble learning algorithm is used to train the ethylene concentration sequence of the path-related shelf group to obtain the influence coefficients of each path on neighboring shelves. A coefficient matrix is ​​constructed using these influence coefficients, where the matrix elements represent the intensity of the ethylene effect of the starting shelf on the target shelf. The intensity distribution of interaction relationships between different shelf areas is determined based on the positions of the non-zero elements in the coefficient matrix. A threshold filtering is performed on the coefficient matrix, retaining interaction relationships with intensity higher than a preset threshold, forming the final regional interaction intensity characterization result.

[0032] For example, in the field of agricultural product shelving management, obtaining real-time ethylene concentration data for each shelf in the shelving layout is a crucial foundational task. Suppose a cold chain warehouse has 10 shelves, each equipped with an ethylene concentration sensor. Real-time monitoring data shows that the concentration on shelf 1 is 5.0 ppm, shelf 2 is 3.5 ppm, shelf 3 is 2.0 ppm, and the concentrations on the remaining shelves are lower. This data allows for a preliminary assessment of the uneven distribution of ethylene concentration, laying the foundation for subsequent analysis.

[0033] Specifically, to determine the ethylene conduction path based on concentration differences, a preset threshold of 1.5 ppm can be set. When the concentration difference between shelf 1 and shelf 2 is 1.5 ppm, reaching the threshold, a conduction path is determined to exist between them, and shelf 1 is recorded as the starting position. The concentration difference between shelf 2 and shelf 3 is also 1.5 ppm, satisfying the condition, so the path continues to extend. This method intuitively reflects the possibility of ethylene diffusing from a high-concentration area to a low-concentration area.

[0034] For example, when extracting downstream adjacent shelf sequences, shelf 1 can be used as the starting point, with shelf 2 and shelf 3 as its downstream adjacent shelves, forming a path-related shelf group. This grouping method helps to focus on the core areas that may be affected by ethylene, thus narrowing the scope of the analysis.

[0035] Specifically, when training the concentration sequence of path-associated shelf groups using an ensemble learning algorithm, the influence coefficients of shelf 1 on shelves 2 and 3 can be analyzed using historical data. Assuming the training results show that the influence coefficient of shelf 1 on shelf 2 is 0.8 and on shelf 3 is 0.6, it indicates that shelf 1 has a stronger effect on shelf 2. This coefficient quantifies the degree of influence of ethylene diffusion, providing data support for subsequent matrix construction.

[0036] For example, when constructing the coefficient matrix, shelf 1, shelf 2, and shelf 3 are used as the rows and columns of the matrix, and the matrix elements are filled with influence coefficients. For instance, the element value of shelf 1 to shelf 2 is 0.8, the element value of shelf 1 to shelf 3 is 0.6, and the element value of other unrelated shelves is 0. This matrix clearly represents the distribution of ethylene influence intensity among the shelves, facilitating further analysis of the interaction relationships between regions.

[0037] Specifically, when determining the intensity distribution of interaction relationships, the positions and values ​​of non-zero elements in the matrix reveal that shelf 1 is the primary source of influence, while shelves 2 and 3 are directly affected areas. This distribution analysis helps identify key nodes in ethylene diffusion, providing a basis for precise management.

[0038] For example, when performing threshold screening, assuming a preset intensity threshold of 0.5, the interaction relationship between shelf 1 and shelves 2 and 3 is retained, while other relationships below the threshold are eliminated, ultimately forming the regional interaction intensity characterization results. This screening ensures that the analysis results focus on the most significant ethylene impact pathways, improving the targeting of management. Through these methods, not only can potential ethylene diffusion pathways between shelves be effectively identified, but data support can also be provided for optimizing shelf layout and adjusting storage strategies, thereby extending the shelf life of agricultural products.

[0039] Step S104: Extract high-intensity transmission regions from the influence coefficient matrix, determine the risk level of agricultural product quality deterioration, and obtain a risk distribution map for visualizing potential hotspots of quality decline within the warehouse.

[0040] High-intensity transmission areas are identified from the influence coefficient matrix. Based on the transmission intensity values ​​at each location within these high-intensity transmission areas, regions with transmission intensity exceeding a preset threshold are identified, resulting in high-risk transmission sub-regions. For these high-risk sub-regions, K-means clustering is used to spatially cluster their locations, yielding quality degradation clusters. The risk transmission intensity at each location within each quality degradation cluster is ranked to determine the cluster's risk level, resulting in high-risk and medium-risk clusters. A corresponding risk score grid is generated based on the spatial coordinates of the high-risk and medium-risk clusters. The risk level intervals are defined by the score values ​​of each grid point, resulting in a risk distribution matrix. This risk distribution matrix is ​​then mapped to the warehouse spatial layout coordinate system to obtain a visual distribution of quality degradation hotspots within the warehouse.

[0041] For example, in the ethylene management scenario of fruit and vegetable warehouses, extracting high-intensity transmission areas from the influence coefficient matrix is ​​a key step in identifying potential quality risks.

[0042] Specifically, when the influence coefficient of a starting shelf in the matrix on multiple adjacent shelves all exceed 0.7, these shelves collectively constitute a high-intensity transmission zone. One possible implementation is to preset a transmission intensity threshold of 0.65; any area exceeding this value is identified as a high-risk transmission sub-region.

[0043] In one embodiment, K-means clustering is used to process the spatial coordinates of high-risk transmission sub-regions. Assuming the warehouse is divided into a grid layout, a high-risk sub-region includes four locations: shelves A12, A13, A14, and B12. After K-means clustering based on spatial distance, a quality deterioration cluster is formed containing these four shelves. This cluster indicates that ethylene rapidly diffuses from shelf A12 to adjacent locations, leading to localized concentration accumulation.

[0044] Preferably, the risk transmission intensity at each location within the quality deterioration cluster is ranked.

[0045] For example, A12 has a source intensity of 0.92, A13 0.85, A14 0.78, and B12 0.71, which determines the overall risk level of the clusters. Clusters with a mean intensity above 0.80 are marked as high-risk clusters, while those with a mean intensity between 0.65 and 0.80 are marked as medium-risk clusters. This classification helps to prioritize interventions at the most severe sources of spread.

[0046] Understandably, a risk score grid is further generated based on the spatial coordinates of high-risk and medium-risk clusters. One implementation involves assigning the highest score to the center point of each cluster and interpolating outwards using a distance-decay method. For example, the center score is set to 100, and the score decreases by 15 points for each additional grid unit, ultimately forming a continuous risk score grid. Points with scores greater than 80 are classified as high-risk, 50 to 80 as medium-risk, and below 50 as low-risk, thus obtaining a risk distribution matrix.

[0047] Specifically, by mapping the risk distribution matrix back to the actual spatial layout coordinate system of the warehouse, the visual distribution of hotspots of quality degradation can be clearly displayed.

[0048] For example, in an apple storage area, an elliptical high-risk hotspot forms in the middle of shelf A, centered near A12, gradually weakening into a medium-risk zone towards shelves B and both sides of A. This distribution clearly shows that ethylene transmission has formed a localized concentrated deterioration trend.

[0049] It should be noted that the hotspots identified through the above methods can guide warehouse managers to accurately apply ethylene absorbent or prioritize adjusting ventilation paths, effectively blocking high-intensity transmission chains and reducing the overall fruit and vegetable spoilage rate.

[0050] Preferably, in actual operation, high-risk clusters can trigger automatic alarms, while medium-risk clusters are included in the key inspection points for the next day, thereby achieving a tiered response and significantly improving the pertinence and timeliness of warehouse quality control.

[0051] Step S105: For the risk distribution mapping, a classification algorithm is used to assess the current quality status of agricultural products and determine the grading category to distinguish batches of agricultural products with different maturity and freshness levels.

[0052] By retrieving relevant data from the storage system, a preliminary analysis of the quality status of agricultural product batches is initiated, yielding initial distribution data. Based on this initial distribution data, a classification algorithm is used to extract features related to the maturity and freshness of the agricultural product batches, determining the status assessment results for each batch. If the maturity of a particular agricultural product batch is below a preset threshold, it is classified into a pending-processing category, and detailed distribution data for this category is obtained. A secondary analysis of this detailed distribution data is performed to determine the differences in freshness among the pending-processing batches, providing specific criteria for category division. Based on these criteria, if the freshness of a pending-processing batch meets a preset standard, it is adjusted to the normal category, determining the final batch category. After obtaining the final batch category, corresponding quality status records are generated for both the normal and pending-processing batches, completing the classification process. By organizing the quality status records, the classification results for each batch are stored in a unified format, and the completeness of the storage is assessed to obtain the final analysis archive.

[0053] For example, in the business scenario of agricultural product quality status analysis, the step of retrieving batch data from the storage system can be understood as extracting information such as the origin, harvest time, and preliminary test data of each batch from the warehouse management system. Suppose there are three batches of data: Batch A comes from a southern orchard, harvested 10 days ago, and preliminary testing shows no obvious surface damage; Batch B comes from a northern orchard, harvested 15 days ago, and has slight surface damage; Batch C was harvested 5 days ago and is in good condition. This data provides a foundation for subsequent analysis and helps to quickly locate batches that may have problems.

[0054] For example, during the initial analysis to obtain preliminary distribution data, the collected data can be organized to generate a map showing the location and preliminary status of each batch in the warehouse. Assuming batch A is concentrated on the east side of the warehouse, batch B on the west side, and batch C in the central area, the preliminary status distribution suggests that batch B may have potential quality issues. This distribution data provides a clear reference for subsequent classification, improving analytical efficiency.

[0055] For example, in the feature extraction stage using classification algorithms, key features can be extracted by quantitatively analyzing indicators of ripeness and freshness. Assuming ripeness is measured by fruit color depth and firmness, and freshness by surface moisture content and storage temperature, batch A has a ripeness of 80% and a freshness of 75%; batch B has a ripeness of 60% and a freshness of 50%; and batch C has a ripeness of 90% and a freshness of 85%. Through algorithmic classification, batch B is marked as a potentially problematic batch. This method helps to accurately identify batches with abnormal conditions, providing a basis for subsequent processing.

[0056] For example, in the process of classifying batches with a maturity level below a preset threshold into the pending category, assuming the preset threshold is 70%, batch B, with a maturity level of 60%, would be classified into the pending category. Detailed distribution data might show that batch B is distributed across multiple shelves on the west side of the warehouse. This classification method facilitates centralized management of problematic batches and avoids omissions.

[0057] For example, analyzing the differences in freshness levels among batches to be processed allows for further examination of whether their storage environments meet standards. Suppose that some shelves in batch B have excessively high temperatures, leading to decreased freshness, while other shelves have normal temperatures and meet freshness standards; in this case, they can be categorized into two groups. This secondary analysis helps to more accurately adjust the categories, avoiding a one-size-fits-all approach.

[0058] For example, during the process of adjusting batch categories, if a portion of batch B meets the freshness standard, such as reaching 80% or more, it is moved to the normal category, while the remaining portion remains in the pending category. This dynamic adjustment method ensures the rationality of classification and improves resource utilization efficiency.

[0059] For example, when generating quality status records, a record file containing location, category, and status details can be generated for each batch. For instance, batch A might be recorded as normal and stored on the east side; batch B might be partially adjusted to the normal category, while some parts remain in the pending category. This recording method facilitates subsequent review and management.

[0060] For example, in the process of storing classification results and verifying their completeness, all batch records can be stored in a database in a uniform format, and the system can verify that no data is missing. If the verification finds that some data in batch C is missing, it needs to be supplemented. This approach ensures the reliability of the analysis archives and provides comprehensive support for subsequent decision-making.

[0061] Step S106: After obtaining the classification category, update the warehouse layout optimization model in conjunction with the ethylene transmission path data to obtain an optimized coexistence configuration, which is used to adjust the shelf placement strategy to reduce ethylene cross-contamination.

[0062] An initial coexistence constraint matrix is ​​constructed using hierarchical categories and ethylene transmission path data. The transmission strength sequence between each cargo pair is calculated based on the initial coexistence constraint matrix and path data. A genetic algorithm is used to iteratively optimize the transmission strength sequence, resulting in an optimized cargo coexistence configuration table. Recommended cargo sets for each shelf unit are extracted from the optimized cargo coexistence configuration table. A shelf adjustment instruction sequence is generated based on the recommended cargo sets and the current shelf occupancy status. After executing the shelf adjustment instruction sequence, the actual occupancy record of the warehouse layout is updated. Based on the updated actual occupancy record, hierarchical categories and ethylene transmission path data are re-acquired, forming a closed-loop update.

[0063] For example, in the business field of agricultural product storage management, the processing of grading and ethylene transmission path data can begin with constructing an initial coexistence constraint matrix. The core of this matrix is ​​to reflect the relationship between ethylene release and sensitivity among different batches of agricultural products. Suppose there are three batches of apples, bananas, and pears, labeled A, B, and C, respectively. Apple batch A has a high ethylene release, banana batch B is highly sensitive to ethylene, while pear batch C is relatively stable. Using grading data, batch A is classified as high ethylene release, and batch B as highly sensitive. The matrix will then indicate a strong constraint relationship between A and B, meaning they should not be stored together to avoid accelerating the ripening and spoilage of the bananas.

[0064] For example, when calculating the conduction intensity sequence, the potential impact intensity between cargo pairs can be analyzed based on ethylene conduction path data. Assuming a warehouse environment with multiple racking units, and path data indicating strong airflow between rack 1 and rack 2, if batch A is placed on rack 1 and batch B on rack 2, the conduction intensity value may be high, indicating a significant risk of ethylene impact. By performing similar analysis on all cargo pairs, a complete intensity sequence is formed, providing a basis for subsequent optimization.

[0065] For example, the iterative optimization process of genetic algorithms can be understood as an optimization method that simulates biological evolution. In practice, the initial goods placement scheme may be random, such as batches A, B, and C being placed on shelves 1, 2, and 3 respectively. Then, through algorithmic simulation of "selection" and "mutation," the placement combinations are continuously adjusted to gradually reduce the probability of high-intensity transmission pairs, ultimately resulting in an optimized goods coexistence configuration table, in which batches A and B are arranged in different areas with better ventilation and isolation.

[0066] For example, when retrieving recommended placement sets of goods, the optimized configuration table might suggest placing batches A and C on shelf 1 because there is no significant ethylene interference between them, while placing batch B alone on shelf 2. This recommended placement set effectively reduces mutual interference between goods and extends shelf life.

[0067] For example, when generating a shelf adjustment instruction sequence, considering the current shelf occupancy status, assuming that shelf 1 already contains other low-ethylene-release goods, the instruction sequence will prioritize coexisting with batch A and instruct batch B to be moved to the vacant shelf 2. This adjustment instruction can quickly respond to the actual storage situation and ensure a reasonable layout.

[0068] For example, after executing an adjustment instruction, the actual occupancy record is updated. The system can record the latest goods list for each shelf, such as shelf 1 for batches A and C, and shelf 2 for batch B. This record update provides an accurate data foundation for subsequent closed-loop management.

[0069] For example, in the closed-loop update process, when re-acquiring the classification and ethylene transfer path data, it might be discovered that the ethylene release of batch A has decreased due to changes in environmental humidity. In this case, the transfer intensity can be recalculated, the constraint matrix adjusted, and a dynamically optimized storage layout formed. This closed-loop mechanism can continuously adapt to environmental changes, ensuring optimal preservation of agricultural products while reducing losses and improving the overall efficiency of storage management.

[0070] Step S107: Based on the optimized coexistence configuration, simulate the propagation process of the future ethylene transmission effect in the warehouse, determine potential chain risks, obtain prevention and adjustment plans, and propose targeted storage environment intervention measures.

[0071] By constructing a data model of the warehouse environment, the initial distribution of ethylene transmission is obtained, and its concentration variation trend in different areas is determined. Based on the concentration variation trend, a spatial grid division method is used to simulate the propagation path of ethylene within the warehouse environment, obtaining the dynamic diffusion range of each area. For the dynamic diffusion range, combined with the layout data of the coexistence configuration, the trigger points of potential threats and cascading risks are analyzed to determine the location distribution of high-risk areas. If the location distribution of high-risk areas exceeds a preset threshold, the risk assessment module calculates the cascading risk probability of each area, identifying key areas for priority intervention. Storage environment parameters of key areas are obtained, and combined with the simulation analysis results, targeted adjustment plans are generated, resulting in an optimized layout configuration. Based on the optimized layout configuration, specific intervention measures are generated to adjust environmental parameters targeting weak points in ethylene transmission, and the final prevention and control effect is assessed. Based on the data feedback on the prevention and control effect, real-time monitoring data of the warehouse environment is updated to continuously track changes in potential threats and obtain long-term prevention strategies.

[0072] By constructing a data model of the warehouse environment, the initial distribution state of ethylene conduction can be obtained. For example.

[0073] In one embodiment, the system first collects ethylene concentration readings at each monitoring point in the warehouse to form an initial concentration field distribution map, and then determines the gradient direction of the concentration from high to low.

[0074] In one possible implementation, the initial distribution shows a concentration of 0.8 ppm in the fruit zone and only 0.2 ppm in the vegetable zone, a difference that directly reflects the difference in release rates between the different goods.

[0075] Specifically, a spatial grid partitioning method was used to simulate the ethylene propagation path. The warehouse was divided into a uniform grid of 50cm × 50cm, and the ethylene diffusion increment was calculated for each grid point based on wind direction, wind speed, and the location of the cargo release source.

[0076] Preferably, simulation results show that the concentration in three adjacent grids near the apple shelf can accumulate to over 1.2 ppm within 24 hours, while the increase in grids farther from the source is less than 0.1 ppm, thus obtaining the dynamic diffusion range of each area. Based on the dynamic diffusion range and combined with current cargo coexistence configuration data, potential threats and trigger points for chain risks are analyzed.

[0077] For example, when apples and bananas are placed next to each other, the ethylene conduction path is the shortest, and the trigger point is concentrated in the aisle area between the two, which is only one row of shelves apart.

[0078] Understandably, if the concentration at such trigger points exceeds the standard, it will cause bananas to ripen faster, creating a chain reaction of ripening and rapidly expanding the threat to the surrounding four shelf units. If the distribution of high-risk areas exceeds a preset threshold, for example, if the proportion of high-risk grids exceeds 18% of the total warehouse area, the risk assessment module will calculate the probability of chain risks in each area.

[0079] Specifically, the risk probability of the apple-banana combination is as high as 0.75, while that of the apple-leafy vegetable combination is only 0.12. Therefore, the priority area for intervention is the densely stored area of ​​high-release fruit products. Storage environment parameters of the key areas, including temperature, humidity, and ventilation rate, are obtained, and targeted adjustment plans are generated based on simulation analysis results.

[0080] For example, by lowering the temperature in key areas from 13°C to 10°C and increasing local exhaust volume by 30%, an optimized layout is achieved. This involves moving some high-release fruits to a separate shelf area near the ventilation openings. Based on this optimized layout, specific intervention measures are generated to adjust environmental parameters in areas with weak ethylene transfer.

[0081] For example, installing mobile activated carbon adsorption devices in high-risk passageways, operating for 8 hours daily, can achieve an ethylene removal rate of over 65%. In another embodiment, adjusting the shelf spacing to increase by 20cm reduces overlapping transmission paths, significantly mitigating the cumulative concentration effect. Based on feedback from control effectiveness data, such as a decrease in concentration in key areas from 1.2 ppm to 0.4 ppm after intervention, real-time monitoring data of the warehouse environment is updated to continuously track changes in potential threats.

[0082] It should be noted that this closed-loop tracking mechanism can promptly identify new sources of ethylene release or ventilation blind spots, thereby forming a long-term prevention strategy. This includes regularly rotating the locations of high-release goods or dynamically adjusting ventilation plans to ensure that the overall risk of ethylene cross-contamination remains at a controllable level. The beneficial effects are that, through the aforementioned multi-level analysis and intervention, not only are the main ethylene transmission pathways effectively blocked, but the spoilage rate can also be reduced by approximately 15%-25%, extending the shelf life of ripe fruits and vegetables and improving the overall economic efficiency of warehousing.

[0083] Step S108: Iteratively update the local environment distribution map through the aforementioned prevention and adjustment scheme, quantify the overall supply chain efficiency improvement effect, determine the final graded value, and use it to guide agricultural product storage and logistics decisions.

[0084] Obtain an initial local environment distribution map. Iterate and update the local environment distribution map using a preventative adjustment plan to obtain an updated local environment distribution map. Calculate the agricultural product storage suitability score for each region based on the updated local environment distribution map, obtaining the storage suitability distribution. Match the storage suitability distribution with logistics path distance data to obtain a supply chain end-to-end efficiency score sequence. Use a random forest algorithm to perform regression analysis on the supply chain end-to-end efficiency score sequence to obtain a quantitative value for efficiency improvement. If the quantitative value for efficiency improvement reaches a preset threshold, retain the current graded value; otherwise, continue executing the preventative adjustment plan to update the local environment distribution map for the next iteration. Determine the final graded value and output it to the agricultural product storage location allocation module and the logistics path planning module.

[0085] For example.

[0086] In one possible implementation, an initial local environmental distribution map is first obtained. This map reflects key parameters such as the current ethylene concentration, temperature, and humidity of each grid cell within the warehouse, forming the initial spatial pattern of ethylene conduction.

[0087] Specifically, suppose that the concentration of ethylene released by apples in the central area of ​​a fruit cold storage reaches 0.8 ppm, while the concentration in the outer area is only 0.1 ppm. The high concentration in the center shows a clear gradient distribution, which provides a benchmark for subsequent intervention.

[0088] Preferably, the initial distribution map is iteratively updated using a preventative adjustment plan. Adjustment measures include locally increasing ventilation rates, moving high-ethylene-releasing cargo to the edges, or adding ethylene absorbent modules. The updated local environmental distribution map shows that the concentration in the central area has decreased to 0.45 ppm, the overall gradient is becoming gentler, and the area of ​​the high-concentration peak region has shrunk by approximately 35%, indicating that ethylene conduction has been initially suppressed. Based on the updated local environmental distribution map, the suitability score for agricultural product storage in each area is calculated. Suitability comprehensively considers the sensitivity threshold of ethylene concentration to different types of fruits and vegetables; for example, bananas score 95 points when the concentration is below 0.3 ppm, while apples can tolerate concentrations up to 0.6 ppm, scoring 85 points. Calculations show that the northwest area of ​​the warehouse, due to its low concentration and stable temperature and humidity, achieves a suitability score of 92 points, while the southeast area, due to higher residual ethylene levels, scores only 68 points, thus forming a clear distribution pattern of storage suitability.

[0089] In one embodiment, the distribution of warehouse suitability is matched with logistics path distance data for calculation. Distance data is derived from goods entry and exit channels, main forklift routes, and loading / unloading area locations. For example, high suitability areas have shorter path distances if they are close to main channels, resulting in a weighted increase in score. After matching, a supply chain efficiency score sequence is obtained, with the Northwest region having the highest combined score of 0.89, while the Southeast region only has 0.62. The sequence as a whole shows a spatial decreasing characteristic from high to low.

[0090] Specifically, a random forest algorithm was used to perform regression analysis on the above efficiency score sequence. Input features included suitability, path distance, handling frequency, and cargo turnover rate. The output was the predicted overall efficiency value. The analysis results showed that the efficiency improvement under the current configuration was quantified as 18.7%.

[0091] It should be noted that if this value exceeds the preset threshold of 15%, the current layout is considered to have reached a better state, and this graded value can be retained.

[0092] In one embodiment, if the efficiency improvement quantification value does not reach the threshold, the preventive adjustment plan continues for the next iteration. For example, the absorbent dosing density is further refined or the shelf spacing is adjusted. The distribution map is updated and the calculation is repeated until the quantification value stably exceeds the threshold. Finally, a graded value is determined. For example, the warehouse is divided into three categories: preferred area, sub-preferred area, and restricted area. This value is output to the agricultural product storage location allocation module and the logistics route planning module to achieve precise placement of goods and optimization of transportation routes.

[0093] Understandably, the above process significantly reduces the risk of ethylene-induced ripening through iterative optimization, while improving the utilization rate of storage space and the efficiency of logistics turnover, resulting in a beneficial effect of reducing the loss rate of goods by about 12% to 20%, and extending the overall shelf life of fruits and vegetables.

[0094] If the technical solution of this application involves the acquisition of personal information, the product using this solution has clearly informed the user of the processing rules and obtained the user's consent before processing. If sensitive personal information is involved, the user's individual consent has been obtained and the "express consent" requirement has been met. For example, a clear sign is placed at the collection device to indicate the collection scope, and the user's voluntary entry is considered as consent; or authorization is obtained through pop-up windows, user uploads, etc. The processing rules include the processor, purpose, method, and type of information.

[0095] The above embodiments are merely one of the preferred embodiments of the present invention and should not be used to limit the scope of protection of the present invention. Any modifications or refinements made to the main design concept and spirit of the present invention that are not of substantial significance, but solve the same technical problem as the present invention, should be included within the scope of protection of the present invention.

Claims

1. A system and method for dynamic monitoring of post-harvest quality and intelligent grading and pricing of agricultural products, characterized in that, The method includes: By using sensor array devices deployed in the warehouse to collect real-time data on ethylene gas concentration, ambient temperature, and air humidity at each shelf location, a local environmental distribution map of the warehouse is generated to reflect the microclimate conditions of the agricultural product storage area. Based on the local environmental distribution map, a neural network algorithm is used to analyze the gas concentration gradient change pattern, determine the starting point and propagation direction of the ethylene conduction path, and track potential sources affecting the ripening of agricultural products. If the starting point of the ethylene transmission path is located on a shelf storing specific agricultural products susceptible to ethylene, the intensity of the ethylene influence of the path on adjacent shelves is quantified by an ensemble learning algorithm to obtain an influence coefficient matrix, which is used to represent the degree of interaction between regions. High-intensity transmission regions are extracted from the influence coefficient matrix to determine the risk level of agricultural product quality deterioration and obtain a risk distribution map, which is used to visualize potential hotspots of quality decline in warehouses. Based on the aforementioned risk distribution mapping, a classification algorithm is used to assess the current quality status of agricultural products and determine grading categories to distinguish batches of agricultural products with different levels of maturity and freshness. After obtaining the classification, the warehouse layout optimization model is updated in conjunction with the ethylene transmission path data to obtain an optimized coexistence configuration, which is used to adjust the shelf placement strategy to reduce ethylene cross-contamination. Based on the optimized coexistence configuration, the propagation process of the ethylene transmission effect in the warehouse is simulated to determine potential chain risks and obtain prevention and adjustment plans, which are used to propose targeted storage environment intervention measures. By iteratively updating the local environmental distribution map through the aforementioned prevention and adjustment scheme, the overall supply chain efficiency improvement effect is quantified, and the final grading and value determination is made to guide agricultural product storage and logistics decisions.

2. The post-harvest quality dynamic monitoring and intelligent grading and pricing system and method for agricultural products according to claim 1, characterized in that, The system uses a sensor array deployed within the warehouse to collect real-time data on ethylene gas concentration, ambient temperature, and air humidity at each shelf location, generating a local environmental distribution map of the warehouse to reflect the microclimate conditions of the agricultural product storage area, including: Acquire raw data on ethylene concentration, temperature, and humidity at each shelf location collected by the sensor array; The concentration, temperature, and humidity sequences at each location were synchronized using a time series alignment method. Calculate the three-dimensional environmental state vector of each shelf location at the current moment based on the concentration sequence, temperature sequence, and humidity sequence; The Kriging interpolation method is used to generate continuous ethylene concentration, temperature, and humidity distribution fields inside the warehouse based on the three-dimensional environmental state vectors of all shelf locations; Extract the location coordinates of regions where the current concentration is higher than a preset threshold from the ethylene concentration distribution field to obtain a set of high-concentration regions; Spatial overlay analysis is performed on the high-concentration region set with the temperature and humidity distribution fields to determine whether there are sub-regions with high temperature and low humidity within the high-concentration region. If there is a sub-region with high temperature and low humidity within the high concentration area, then mark the sub-region as an area of ​​key concern and record its spatial range; Based on the spatial range of the key concern area, extract the concentration change sequence of the corresponding location from the ethylene concentration distribution field, and calculate the rate of increase of the sequence in the recent period. If the rate of increase exceeds the preset rate threshold, then the key area of ​​concern is determined to be in a state of accelerated ethylene accumulation. For key areas of concern that are in a state of accelerated ethylene accumulation, temperature change sequences and humidity change sequences at corresponding locations are extracted from the temperature distribution field and humidity distribution field; Determine whether current microclimate conditions further promote ethylene release based on temperature and humidity change sequences; If the judgment result is to promote release, a local environmental anomaly alert will be generated, which includes the location of the key concern area, ethylene concentration value, temperature value, humidity value, and accelerated accumulation status.

3. The post-harvest quality dynamic monitoring and intelligent grading and pricing system and method for agricultural products according to claim 1, characterized in that, The step involves analyzing the gas concentration gradient change pattern using a neural network algorithm based on the local environmental distribution map to determine the starting point and propagation direction of the ethylene conduction path, used to track potential sources affecting agricultural product ripening, including: Obtain the location of each monitoring point and the corresponding ethylene gas concentration data in the local environmental distribution map; Based on the location of each monitoring point and the ethylene gas concentration data, a neural network algorithm is used to calculate the concentration difference between adjacent points to obtain the gas concentration gradient field. By determining the direction and magnitude of the gradient vector in the gas concentration gradient field, the reverse direction of the local maximum gradient at each point can be determined, thus obtaining the preliminary direction of the ethylene conduction path. The region with the highest gradient vector continuity is selected from the initial ethylene conduction path directions to obtain a set of candidate ethylene conduction paths; For each path in the candidate ethylene conduction pathway set, backtrack point by point from the end along the opposite direction of the gradient to determine whether the concentration continues to rise. If the concentration continues to rise, the path is retained; if the concentration decreases, the path is discarded, thus obtaining the selected ethylene conduction pathways. In the selected ethylene conduction pathway, find the starting endpoint where the concentration gradient vector converges, and determine the position of this endpoint as the starting position of the ethylene conduction pathway; By taking the starting point of the ethylene transmission path as the source point and extending along the selected ethylene transmission path, the direction of ethylene propagation and the location of potential sources of influence on agricultural product ripening can be obtained.

4. The post-harvest quality dynamic monitoring and intelligent grading and pricing system and method for agricultural products according to claim 1, characterized in that, If the starting point of the ethylene transmission path is located on a shelf storing specific agricultural products susceptible to ethylene, then an ensemble learning algorithm is used to quantify the intensity of the path's ethylene influence on adjacent shelves, resulting in an influence coefficient matrix to represent the degree of interaction between regions, including: Obtain real-time ethylene concentration data for each shelf in the shelving layout; The existence of an ethylene transmission path is determined by a preset concentration difference. If the concentration difference between adjacent shelves exceeds a preset threshold, the existence of the path is confirmed and the starting shelf position is recorded. For the starting shelf of the identified ethylene transmission path, extract the sequence of its downstream adjacent shelves to form a path-related shelf group; An ensemble learning algorithm was used to train the ethylene concentration sequence of path-related shelf groups to obtain the influence coefficient of each path on adjacent shelves. A coefficient matrix is ​​constructed using influence coefficients, where the matrix elements represent the intensity of the ethylene effect of the starting shelf on the target shelf; The distribution of interaction intensity between different shelf areas is determined based on the positions of non-zero elements in the coefficient matrix. A threshold screening is performed on the coefficient matrix to retain interaction relationships with an intensity higher than a preset threshold, thus forming the final regional interaction intensity characterization result.

5. The post-harvest quality dynamic monitoring and intelligent grading and pricing system and method for agricultural products according to claim 1, characterized in that, The process of extracting high-intensity transmission regions from the influence coefficient matrix, determining the risk level of agricultural product quality deterioration, and obtaining a risk distribution map are used to visualize potential hotspots of quality decline within the warehouse, including: High-intensity conduction regions can be obtained from the influence coefficient matrix; Based on the conduction intensity values ​​at various locations within the high-intensity conduction region, areas with conduction intensity greater than a preset threshold are identified, thus obtaining high-risk conduction sub-regions. For high-risk transmission sub-regions, the K-means clustering method is used to spatially cluster the locations of the sub-regions to obtain quality deterioration clusters; The risk transmission intensity at each location within the quality deterioration cluster is ranked to determine the risk level of the cluster, resulting in high-risk and medium-risk clusters. Based on the spatial coordinates of high-risk and medium-risk clusters, a corresponding risk score grid is generated; By dividing the risk level ranges into intervals based on the scores of each grid point in the risk score grid, a risk distribution matrix is ​​obtained; By mapping the risk distribution matrix to the warehouse spatial layout coordinate system, a visual distribution of hotspots of quality degradation within the warehouse can be obtained.

6. The post-harvest quality dynamic monitoring and intelligent grading and pricing system and method for agricultural products according to claim 1, characterized in that, The aforementioned risk distribution mapping employs a classification algorithm to assess the current quality status of agricultural products and determine grading categories to distinguish batches of agricultural products with different levels of maturity and freshness, including: By retrieving relevant data from agricultural product batches from the storage system, a preliminary analysis of the quality status is initiated to obtain initial distribution data. Based on the initial distribution data, a classification algorithm is used to extract features of the maturity and freshness of agricultural product batches, and to determine the status assessment results of each batch. Based on the status assessment results, if the maturity level of a certain batch of agricultural products is lower than the preset threshold, it will be classified as a category to be processed, and detailed distribution data of the category to be processed will be obtained. By performing secondary analysis on the detailed distribution data, the differences in freshness among the batches to be processed are determined, and specific classification criteria are obtained. Based on the classification criteria, if the freshness of a batch to be processed meets the preset standard, it will be adjusted to the normal category, and the final batch category will be determined. After obtaining the final batch category, generate corresponding quality status records for agricultural product batches in the normal category and the pending category to complete the classification process; By organizing the quality status records, storing the classification results of each batch in a unified format, judging whether the storage is complete, and obtaining the final analysis archive.

7. The post-harvest quality dynamic monitoring and intelligent grading and pricing system and method for agricultural products according to claim 1, characterized in that, After obtaining the classification, the warehouse layout optimization model is updated in conjunction with the ethylene transmission path data to obtain an optimized coexistence configuration, which is used to adjust the shelf placement strategy to reduce ethylene cross-contamination, including: An initial coexistence constraint matrix is ​​constructed using hierarchical categories and ethylene transmission path data; Calculate the transmission strength sequence between each cargo pair based on the initial coexistence constraint matrix and path data; A genetic algorithm is used to iteratively optimize the conduction intensity sequence to obtain an optimized cargo coexistence configuration table; Extract the recommended set of goods to be placed in each shelf unit from the optimized goods coexistence configuration table; Generate a sequence of shelf adjustment instructions based on the recommended set of goods and the current shelf occupancy status; Update the actual occupancy record of the warehouse layout after executing the shelf adjustment instruction sequence; Based on the updated actual occupancy records, the classification and ethylene transmission path data are re-acquired to form a closed-loop update.

8. The post-harvest quality dynamic monitoring and intelligent grading and pricing system and method for agricultural products according to claim 1, characterized in that, Based on the optimized coexistence configuration, the propagation process of the ethylene transmission effect within the warehouse is simulated to determine potential cascading risks and obtain preventative adjustment plans. These plans are used to propose targeted storage environment intervention measures, including: By constructing a data model of the warehouse environment, the initial distribution state of ethylene conduction was obtained, and its concentration variation trend in different regions was determined. Based on the concentration change trend, a spatial grid division method is used to simulate the propagation path of ethylene in the warehouse environment and obtain the dynamic diffusion range of each region. Based on the dynamic diffusion range and the layout data of coexistence configuration, we analyze the trigger points of potential threats and chain risks, and determine the location distribution of high-risk areas. If the location distribution of high-risk areas exceeds the preset threshold, the risk assessment module will calculate the probability of chain risks in each area and determine the key areas for priority intervention. Obtain the storage environment parameters of key areas, combine them with the results of simulation analysis, generate targeted adjustment plans, and obtain the optimized layout configuration; By optimizing the layout and configuration, specific intervention measures are generated, environmental parameters are adjusted to target the weak links in ethylene transmission, and the final control effect is judged. Based on data feedback on the effectiveness of prevention and control, we update real-time monitoring data of the warehouse environment, continuously track changes in potential threats, and develop long-term prevention strategies.

9. The post-harvest quality dynamic monitoring and intelligent grading and pricing system and method for agricultural products according to claim 1, characterized in that, The process of iteratively updating the local environmental distribution map through the aforementioned prevention and adjustment scheme, quantifying the overall supply chain efficiency improvement effect, and determining the final grading and value are used to guide agricultural product warehousing and logistics decisions, including: Obtain the initial local environment distribution map; The local environmental distribution map is updated by one iteration of the prevention and adjustment plan to obtain the updated local environmental distribution map. Based on the updated local environmental distribution map, calculate the agricultural product storage suitability score for each region to obtain the storage suitability distribution; Based on the matching calculation of warehouse suitability distribution and logistics route distance data, a supply chain full-link efficiency score sequence is obtained; Random forest algorithm is used to perform regression analysis on the efficiency score sequence of the entire supply chain to obtain a quantitative value of efficiency improvement; If the efficiency improvement quantification value reaches the preset threshold, the current graded quantification value is retained; otherwise, the preventive adjustment plan is continued to update the local environmental distribution map for the next iteration. The final grade and value are determined and output to the agricultural product storage location allocation module and the logistics route planning module.

Images