A holographic whole-chain grain and oil product quality safety real-time intelligent discovery and prevention and early warning method and device

Abstract

The application discloses a kind of holographic full chain grain and oil product quality safety real-time intelligent discovery and prevention and control early warning method and device, it is related to food safety technical field, the method comprises: the value of main influencing factor at each preset position of target planting area is input into each pollutant content prediction model, the content of each pollutant in the crop planted at each preset position is obtained;The content of each pollutant in target grain and oil product production link, acquisition link, storage link, processing link and sales link is obtained;Based on the threshold range of each pollutant, when the content of each link pollutant and entity are constructed, the knowledge graph corresponding to each pollutant is constructed;According to knowledge graph and the content of each pollutant in production link in next cycle, the content of other link pollutant in next cycle is predicted and prevention and control early warning, the pollutant content of any one link in grain and oil product production, acquisition, storage, processing and sales link can be monitored and prevention and control early warning in real time.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of food safety, in particular to a holographic whole-chain grain and oil product quality safety real-time intelligent discovery, prevention and control and early warning method and device. BACKGROUND

[0002] In recent years, food safety crises have occurred frequently, seriously affecting people's health and attracting widespread attention worldwide. Therefore, it is very important to monitor and prevent and control the quality safety of any one of the production, procurement, storage, processing and sales links of grain and oil products in real time. SUMMARY

[0003] The purpose of the application is to provide a holographic whole-chain grain and oil product quality safety real-time intelligent discovery, prevention and control and early warning method and device, which can monitor and prevent and control the pollutant content of any one of the production, procurement, storage, processing and sales links of grain and oil products in real time.

[0004] To achieve the above purpose, the application provides the following solutions:

[0005] In a first aspect, the application provides a holographic whole-chain grain and oil product quality safety real-time intelligent discovery, prevention and control and early warning method, comprising:

[0006] For any one variety of crops planted in the target planting area, in the current period, the values of the main influencing factors at each preset position in the target planting area are input into each pollutant content prediction model corresponding to the variety of crops to obtain the content of each pollutant in the variety of crops planted at each preset position; the variety of crops is used to make a target grain and oil product; one period includes the production, procurement, storage, processing and sales links of the target grain and oil product; the pollutant content prediction model corresponding to the variety of crops is obtained by training a convolutional neural network model according to the main influencing factors at each target sampling point in the target planting area and the content of the pollutant in the variety of crops planted at each target sampling point in the current period; the main influencing factors are obtained by screening environmental variables and spatial structure variables;

[0007] The content of each pollutant in the variety of crops planted at each preset position and each target sampling point is determined as the content of each pollutant in the production link of the target grain and oil product;

[0008] The content of each pollutant in the variety of crops planted at each preset position and each target sampling point is determined as the content of each pollutant in the production link of the target grain and oil product;

[0009] The crops of the variety planted at each preset position and each target sampling point are stored into a warehouse, and the content of each pollutant at the time of warehousing is determined as the content of each pollutant in the storage link of the target grain and oil product, and the content of each pollutant at the time of warehousing is determined as the content of each pollutant in the processing link of the target grain and oil product;

[0010] The crops of the variety after warehousing are processed to obtain the target grain and oil product, and the content of each pollutant of the target grain and oil product to be sold is taken as the content of each pollutant in the sales link of the target grain and oil product; the content of each pollutant of the target grain and oil product to be sold is within the preset range;

[0011] Based on the threshold range of each pollutant, the content of each pollutant in the production link, the acquisition link, the storage link, the processing link and the sales link of the target grain and oil product in the current period, and the entity on the grain and oil whole-process holographic traceability chain, a knowledge graph corresponding to each pollutant is constructed.

[0012] According to the knowledge graph corresponding to each pollutant and the content of each pollutant in the production link of the target grain and oil product in the next period, the content of each pollutant in the acquisition link, the storage link, the processing link and the sales link of the target grain and oil product in the next period is predicted to obtain a prediction value, and the prediction value is used for prevention and early warning.

[0013] Optionally, the determination process of the target sampling point specifically includes:

[0014] For any one variety of crops planted in the target planting area, in the current period, remote sensing data of the target planting area is acquired;

[0015] The remote sensing data is divided into grids, and the crop planting density and area of each grid are determined; the crops of the variety are planted in the grid;

[0016] According to the crop planting density and area of each grid, the remote sensing data is divided according to the crop planting density grade, to obtain all grids corresponding to each crop planting density grade;

[0017] According to all grids corresponding to each crop planting density grade, a sampling point of the target planting area is determined as a target sampling point according to the stratified sampling principle.

[0018] Optionally, based on the threshold range of each pollutant, the content of each pollutant in the production link, the acquisition link, the storage link, the processing link and the sales link of the target grain and oil product in the current period, and the entity on the grain and oil whole-process holographic traceability chain, a knowledge graph corresponding to each pollutant is constructed, specifically including:

[0019] entity recognition and relationship recognition are performed on the entities and the relationships between the entities on the grain and oil whole-process holographic traceability chain based on the threshold range of each pollutant, the content of each pollutant in the production link, the acquisition link, the storage link, the processing link and the sales link of the target grain and oil product in the current period, to obtain a plurality of entities and the relationships between the entities;

[0020] A knowledge graph corresponding to each pollutant is constructed according to the entities and the relationships between the entities.

[0021] Optionally, before the knowledge graph corresponding to each pollutant is constructed based on the threshold range of each pollutant, the content of each pollutant in the production link, the acquisition link, the storage link, the processing link and the sales link of the target grain and oil product in the current period, and the entities on the grain and oil whole-process holographic traceability chain, the method further comprises:

[0022] The values of the target collection data, the monitoring data and the main influencing factors at each preset position in the target planting area in the current period and the values of the target collection data, the monitoring data and the main influencing factors at each target sampling point are stored on the grain and oil whole-process holographic traceability chain; the target collection data includes longitude and latitude, crop variety, plot, sampling time; the monitoring data includes crop production information, crop planting information, crop fertilization and irrigation information, crop pest control information, and the content of each pollutant in the production link, the acquisition link, the storage link, the processing link and the sales link of the target grain and oil product.

[0023] Optionally, according to the knowledge graph corresponding to each pollutant and the content of each pollutant in the production link of the target grain and oil product in the next period, the content of each pollutant in the acquisition link, the storage link, the processing link and the sales link of the target grain and oil product in the next period is predicted to obtain a prediction value, and a prevention and early warning is performed according to the prediction value, specifically including:

[0024] For any one of the acquisition link, the storage link, the processing link and the sales link of the target grain and oil product in the next period, the content of each pollutant in the link in the next period is predicted according to the knowledge graph corresponding to each pollutant and the content of each pollutant in the production link of the target grain and oil product in the next period to obtain a prediction value;

[0025] For any one pollutant, if the prediction value of the content of the pollutant in the link in the next period exceeds a set threshold, a prevention and early warning is performed on the pollutant in the link in the next period.

[0026] Optionally, the environmental variables include ecological elements and human activity elements, and the spatial structure variables include longitude and latitude, spatial autocorrelation variables and spatial regionalization variables.

[0027] Optionally, for any one variety of crop planted in the target planting area, the value of the spatial regionalization variable at the preset position in the target planting area is determined as follows:

[0028] For any pollutant, the pollutant content in the crops of the specified variety planted at each target sampling point in the target planting area is processed to obtain the treated pollutant content in the crops of the specified variety planted at each target sampling point in the target planting area; the treated pollutant content in the crops of the specified variety planted at each target sampling point in the target planting area conforms to a normal distribution.

[0029] Spatial statistics were performed on the content of the pollutants after treatment in the crops of the variety planted at each target sampling point in the target planting area to obtain the range of high clustering values ​​and the range of low clustering values ​​in the target planting area.

[0030] Based on the high and low clustering value ranges of the target planting area, the target planting area is divided into multiple sub-regions using the natural discontinuity method. The average value of the pollutants corresponding to each sub-region is determined as the value of the spatial regionalization variable. The average value of the pollutants corresponding to each sub-region is the average content of the pollutants in the crops used to produce grain and oil products planted at each target sampling point within the sub-region.

[0031] Optionally, spatial statistics are performed on the content of the treated pollutants in the crops of the aforementioned varieties planted at each target sampling point in the target planting area to obtain the high clustering value range and low clustering value range of the target planting area, specifically:

[0032] The high and low clustering value ranges of the target planting area were obtained by performing local Moran's I spatial statistics on the content of the pollutants after treatment in the crops of the variety planted at each target sampling point in the target planting area.

[0033] Optionally, the process for determining the value of the spatial autocorrelation variable at the preset location is as follows:

[0034] The range distance is obtained by fitting the semivariogram function;

[0035] For any preset location, the distances between each point in the nearest neighbor set corresponding to the preset location and the preset location are determined to obtain the distance set corresponding to the preset location; the nearest neighbor set corresponding to the preset location includes the k nearest neighbor target sampling points within the target planting area of ​​the preset location;

[0036] The range distance and the distance set corresponding to the preset position are determined as the values ​​of the spatial autocorrelation variables at the preset position.

[0037] Secondly, this application provides a holographic, full-chain grain and oil product quality and safety real-time intelligent detection and early warning device, comprising:

[0038] The pollutant content prediction module is used to, within the current cycle, input the values ​​of the main influencing factors at each preset location in the target planting area for any crop variety to the pollutant content prediction model corresponding to that crop variety, thereby obtaining the content of each pollutant in the crop variety planted at each preset location; the crop variety is used to produce target grain and oil products; one cycle includes the production, acquisition, storage, processing, and sales stages of the target grain and oil products; the pollutant content prediction model corresponding to the crop variety is obtained by training a convolutional neural network model based on the main influencing factors at each target sampling point in the target planting area within the current cycle and the pollutant content in the crop variety planted at each target sampling point; the main influencing factors are obtained by screening environmental variables and spatial structure variables.

[0039] The production process pollutant content determination module is used to determine the content of each pollutant in the crop of the variety planted at each preset location and each target sampling point as the content of each pollutant in the production process of the target grain and oil product.

[0040] The module for determining the pollutant content in the acquisition process is used to obtain the content of each pollutant in the crop of the variety planted at each preset location and each target sampling point after the new harvest, as the content of each pollutant in the acquisition process of the target grain and oil products.

[0041] The module for determining the content of pollutants in the storage and processing stages is used to store the crops of the aforementioned varieties planted at each preset location and each target sampling point after acquisition into a warehouse, determine the content of each pollutant at the time of entry into the warehouse as the content of each pollutant in the storage stage of the target grain and oil products, and determine the content of each pollutant at the time of exit from the warehouse as the content of each pollutant in the processing stage of the target grain and oil products.

[0042] The module for determining the contaminant content in the sales process is used to process the crops of the aforementioned varieties after they leave the warehouse to obtain the target grain and oil products. The content of each contaminant in the target grain and oil products to be sold is used as the content of each contaminant in the sales process of the target grain and oil products; the content of each contaminant in the target grain and oil products to be sold is within the preset range.

[0043] The knowledge graph construction module is used to construct the knowledge graph corresponding to each pollutant based on the threshold range of each pollutant, the content of each pollutant in the production, acquisition, storage, processing and sales stages of the target grain and oil products in the current period, and the entities on the whole-process holographic traceability chain of grain and oil.

[0044] The prevention and control early warning module is used to predict the content of each pollutant in the acquisition, storage, processing and sales stages of the target grain and oil products in the next cycle based on the knowledge graph corresponding to each pollutant and the content of each pollutant in the production stage of the target grain and oil products in the next cycle, and to carry out prevention and control early warning based on the predicted value.

[0045] According to the specific embodiments provided in this application, this application has the following technical effects:

[0046] This application provides a holographic, full-chain, real-time intelligent detection, prevention, and early warning method and device for grain and oil product quality and safety. By inputting the values ​​of the main influencing factors at preset locations into a pollutant content prediction model, the content of each pollutant in the crops planted at the preset locations is obtained; the content of each pollutant in the crops planted at the preset locations and target sampling points is determined as the content of each pollutant in the production stage; the content of each pollutant in newly harvested crops is obtained as the content of each pollutant in the acquisition stage; the acquired crops of the aforementioned varieties planted at each preset location and each target sampling point are stored in a warehouse, and the content of each pollutant upon entering the warehouse is determined as the content of each pollutant in the storage stage of the target grain and oil product; the content of each pollutant in the target grain and oil product to be sold is determined as the content of each pollutant in the sales stage of the target grain and oil product. This allows for real-time monitoring of the pollutant content at any stage of grain and oil product production, acquisition, storage, processing, and sales. Furthermore, based on the knowledge graph corresponding to each pollutant and the content of each pollutant in the production process of the target grain and oil products in the next cycle, the content of each pollutant in the acquisition, storage, processing and sales processes of the target grain and oil products in the next cycle is predicted to obtain the predicted value. Based on the predicted value, prevention and control warnings are carried out. This can achieve the prediction and prevention and control warning of pollutant content in each stage of acquisition, storage, processing and sales. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 A flowchart illustrating a holographic, full-chain, real-time intelligent detection, prevention, and early warning method for the quality and safety of grain and oil products, provided as an embodiment of this application;

[0049] Figure 2 A schematic diagram showing the division of rice planting areas into regions with high and low cadmium aggregation values;

[0050] Figure 3 This is a curve fitted to the semivariogram function;

[0051] Figure 4 A graph showing the linear fitting results between the model's predicted values ​​and the measured values;

[0052] Figure 5 This is a map showing the distribution of cadmium content. Detailed Implementation

[0053] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0054] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0055] In one exemplary embodiment, such as Figure 1 As shown, a holographic, full-chain, real-time intelligent detection, prevention, and early warning method for grain and oil product quality and safety is provided. The method includes the following steps, wherein:

[0056] Step 201: For any crop variety planted in the target planting area, within the current cycle, input the values ​​of the main influencing factors at each preset location in the target planting area into the pollutant content prediction model corresponding to the crop variety to obtain the content of each pollutant in the crop variety planted at each preset location. The crop variety is used to produce the target grain and oil products; one cycle includes the production, acquisition, storage, processing, and sales stages of the target grain and oil products; the pollutant content prediction model corresponding to the crop variety is obtained by training a convolutional neural network model based on the main influencing factors at each target sampling point in the target planting area within the current cycle and the pollutant content in the crop variety planted at each target sampling point; the main influencing factors are obtained by screening environmental variables and spatial structure variables. Pollutants can be mycotoxins and heavy metals.

[0057] Step 202: Determine the content of each pollutant in the crops of the variety planted at each preset location and each target sampling point as the content of each pollutant in the production process of the target grain and oil products.

[0058] Step 203: Obtain the content of each pollutant in the crops of the variety planted at each preset location and each target sampling point after the new harvest, as the content of each pollutant in the target grain and oil product acquisition process.

[0059] Step 204: Store the crops of the aforementioned varieties planted at each preset location and each target sampling point in a warehouse. Determine the content of each pollutant at the time of entry into the warehouse as the content of each pollutant in the storage stage of the target grain and oil products, and determine the content of each pollutant at the time of exit from the warehouse as the content of each pollutant in the processing stage of the target grain and oil products.

[0060] Step 205: Process the crop of the variety after it leaves the warehouse to obtain the target grain and oil products. The content of each pollutant in the target grain and oil products to be sold is taken as the content of each pollutant in the sales process of the target grain and oil products; the content of each pollutant in the target grain and oil products to be sold is within the preset range.

[0061] Step 206: Based on the threshold range of each pollutant, the content of each pollutant in the production, acquisition, storage, processing and sales stages of the target grain and oil products in the current period, and the entities on the whole-process holographic traceability chain of grain and oil, construct the knowledge graph corresponding to each pollutant.

[0062] Step 207: Based on the knowledge graph corresponding to each pollutant and the content of each pollutant in the production process of the target grain and oil products in the next cycle, predict the content of each pollutant in the acquisition, storage, processing and sales processes of the target grain and oil products in the next cycle, and carry out prevention and control early warning based on the predicted values.

[0063] By implementing steps 201 to 207 above, the pollutant content in any stage of the production, acquisition, storage, processing and sales of grain and oil products can be monitored and controlled in real time.

[0064] In another exemplary embodiment of this application, the process of determining the target sampling point specifically includes:

[0065] For any crop variety planted in the target planting area, acquire remote sensing data of the target planting area within the current period.

[0066] The remote sensing data is divided into grids, and the crop planting density and area of ​​each grid are determined; the crop of the specified variety is planted within each grid.

[0067] Based on the crop planting density and area of ​​each grid, the remote sensing data is divided according to the crop planting density level to obtain all grids corresponding to each crop planting density level.

[0068] Based on all grids corresponding to each crop planting density level, the sampling points of the target planting area are determined as target sampling points according to the principle of stratified sampling.

[0069] In another exemplary embodiment of this application, a knowledge graph corresponding to each pollutant is constructed based on the threshold range of each pollutant, the content of each pollutant in the production, acquisition, storage, processing and sales stages of the target grain and oil product in the current period, and the entities on the whole-process holographic traceability chain of grain and oil. Specifically, this includes:

[0070] Based on the threshold ranges of various pollutants and the content of each pollutant in the production, acquisition, storage, processing, and sales stages of the target grain and oil products within the current period, entity identification and relationship identification are performed on the grain and oil full-process holographic traceability chain to obtain multiple entities and the relationships between them. The grain and oil full-process holographic traceability chain is used to achieve traceability of grain and oil products.

[0071] Construct a knowledge graph for each pollutant based on the entities and the relationships between them.

[0072] In another exemplary embodiment of this application, before constructing a knowledge graph corresponding to each pollutant based on the threshold range of each pollutant, the content of each pollutant in the production, acquisition, storage, processing and sales stages of the target grain and oil product in the current period, and the entities on the grain and oil full-process holographic traceability chain, the following is also included:

[0073] The target data, monitoring data, and values ​​of major influencing factors at each preset location within the target planting area during the current period, as well as the target data, monitoring data, and values ​​of major influencing factors at each target sampling point, are stored on the whole-process holographic traceability chain for grain and oil. The target data includes: latitude and longitude, crop variety, location, and sampling time. The monitoring data includes: crop production information, crop planting information, crop fertilization and irrigation information, crop pest and disease control information, and the content of various pollutants in the production, acquisition, storage, processing, and sales stages of the target grain and oil products.

[0074] In another exemplary embodiment of this application, the content of each pollutant in the acquisition, storage, processing, and sales stages of the target grain and oil products in the next cycle is predicted based on the knowledge graph corresponding to each pollutant and the content of each pollutant in the production stage of the target grain and oil products in the next cycle. Predicted values ​​are then obtained, and prevention and control warnings are implemented based on these predicted values. Specifically, this includes:

[0075] For any one of the following stages in the next cycle: acquisition, storage, processing, and sales of the target grain and oil products, the predicted values ​​are obtained by predicting the content of each pollutant in the next cycle based on the knowledge graph corresponding to each pollutant and the content of each pollutant in the production stage of the target grain and oil products in the next cycle.

[0076] For any pollutant, if the predicted value of the pollutant content in the next cycle exceeds a set threshold, then a prevention and control warning will be issued for the pollutant in the next cycle.

[0077] In another exemplary embodiment of this application, the environmental variables include ecological elements and human activity elements, and the spatial structure variables include latitude and longitude, spatial autocorrelation variables, and spatial regionalization variables. The ecological elements include soil properties, topography, and climate. The human activity elements include road density and agricultural inputs.

[0078] In another exemplary embodiment of this application, the process for determining the value of the spatial regionalization variable at a preset location in the target planting area for any variety of crop planted in the target planting area is as follows:

[0079] For any pollutant, the pollutant content in the crops of the specified variety planted at each target sampling point in the target planting area is processed to obtain the processed pollutant content in the crops of the specified variety planted at each target sampling point in the target planting area; the processed pollutant content in the crops of the specified variety planted at each target sampling point in the target planting area conforms to a normal distribution.

[0080] Spatial statistics were performed on the content of the pollutants after treatment in the crops of the variety planted at each target sampling point in the target planting area to obtain the range of high clustering values ​​and the range of low clustering values ​​in the target planting area.

[0081] Based on the high and low clustering value ranges of the target planting area, the target planting area is divided into multiple sub-regions using the natural discontinuity method. The average value of the pollutants corresponding to each sub-region is determined as the value of the spatial regionalization variable. The average value of the pollutants corresponding to each sub-region is the average content of the pollutants in the crops used to produce grain and oil products planted at each target sampling point within the sub-region.

[0082] In another exemplary embodiment of this application, spatial statistics are performed on the content of the treated pollutants in the crops of the variety planted at each target sampling point in the target planting area to obtain the high clustering value range and low clustering value range of the target planting area, specifically:

[0083] The high and low clustering value ranges of the target planting area were obtained by performing local Moran's I spatial statistics on the content of the pollutants after treatment in the crops of the variety planted at each target sampling point in the target planting area.

[0084] In another exemplary embodiment of this application, the process for determining the value of the spatial autocorrelation variable at the preset location is as follows:

[0085] The variable range distance is obtained by fitting the semivariogram function.

[0086] For any preset location, the distance between each point in the nearest neighbor set corresponding to the preset location and the preset location is determined to obtain the distance set corresponding to the preset location; the nearest neighbor set corresponding to the preset location includes the k nearest neighbor target sampling points within the target planting area.

[0087] The range distance and the distance set corresponding to the preset position are determined as the values ​​of the spatial autocorrelation variables at the preset position.

[0088] This application can use a pollutant content prediction model to predict pollutants in the spatial dimension and a knowledge graph to predict pollutants in the temporal dimension. After obtaining the predicted values ​​for each stage using the knowledge graph, the process of exceeding the standard can be identified and an early warning can be issued. When product quality problems are discovered, the process of causing the problem can be accurately located. At the same time, it can also realize risk prediction and early warning in the quality control process. It can effectively assess the real-time risk prevention and control of the entire chain of grain and oil products from the place of origin to the product (production flow, logistics flow, and control flow), and provide important technical support for food quality and safety supervision and control.

[0089] This application provides a specific embodiment using the heavy metal cadmium as an example to illustrate the above method in detail:

[0090] Step 1: In the crop planting and production stage, based on the spectral, shape, texture and other characteristics of the crop, obtain satellite remote sensing data containing the corresponding crop planting range and planting density, with a resolution of 1km×1km.

[0091] Step 2: Based on the remote sensing data from Step 1, determine the sampling quantity and specific sampling locations for the target planting area according to the regional planting density distribution and planting area. Crops are planted within the target planting area.

[0092] Specifically, for any crop variety in the target planting area, the grids and the number of grids for different planting densities are counted, and the grids are divided into the corresponding crop planting density levels, so that the total planting area of ​​each crop planting density level is approximately equal to the proportion of the total planting area. Then, the sampling quantity and sampling points for each level are determined, and the latitude and longitude information of the sampling points is converted into specific sampling addresses using map software.

[0093] Taking rice as an example: satellite remote sensing data for rice is presented as a 1km × 1km grid with sequential coding numbers and latitude and longitude coordinates. Each grid corresponds to a rice planting density, and the density multiplied by the number of grids equals the corresponding planting area, measured in km². 2 That is, the planting area of ​​the i-th grid is its corresponding rice planting density. S is obtained by summing the planting areas of all grids.N Let L be the total number of crop planting density levels. L can be determined based on the total sampling volume M (the larger L is, the more fully it can reflect the distribution of different planting density levels). According to the formula... The average area is obtained. Based on the average area and the planting area for each rice planting density, the rice planting area for each crop planting density level is obtained, ensuring that the rice planting area for each crop planting density level is approximately equal and similar to the average area. According to the formula... Obtain all grids corresponding to the planting density level of crop l, S l The sum of the rice planting area at the l-th crop planting density level, s i Let be the planting area of ​​the i-th grid. If the grid density corresponding to the l-th level includes planting densities a, b, and c, then all grids corresponding to the l-th level include all grids corresponding to planting densities a, b, and c. According to the formula... The total number of samples m for each grade is obtained. Based on all the grids corresponding to each grade and the total number of samples for each grade, the sampling points (i.e., the grids to be extracted) for each grade are determined. To determine which grid to extract, a random sampling tool can be used to extract the number from the corresponding group interval code and obtain the corresponding latitude and longitude, i.e., the sampling point location. One group interval corresponds to one grade. The code in the group interval is the planting density value included in the corresponding grade. For specific sampling statistics, refer to Table 1.

[0094] Table 1. Example of Sampling Statistics

[0095]

[0096]

[0097] Step 3: In the production process, after determining the sampling points, the specific sampling points are communicated to the drone. The drone collects data on the crops and target data at the sampling points, and stores the target data in the grain and oil full-process holographic traceability chain for later traceability. The collected samples (crops at the sampling points) are automatically sent to unmanned testing equipment for testing of relevant quality and safety indicators, obtaining the content of various pollutants. In this example, the content is cadmium. The test data (cadmium content) is directly and automatically transmitted to the grain and oil full-process holographic traceability chain, forming monitoring data with other information from the production process, which is also used for traceability later. The monitoring data includes: production information, planting information, fertilization and irrigation information, pest and disease control information, and cadmium content, etc.

[0098] Step 4: Determine the input parameters of the neural network model and construct the neural network model.

[0099] Specifically, principal component analysis, correlation analysis, or variance inflation factor test are used to screen environmental and spatial structure variables to obtain the main influencing factors as input parameters. Environmental variables include ecological factors (soil properties, topography, climate, etc.) and human activity factors (road density, agricultural inputs, etc.), while spatial structure variables include geographical location (latitude and longitude), spatial autocorrelation variables, and spatial regionalization variables. This step fully considers both environmental and spatial structure variables, making the cadmium content predicted by the model more accurate in the later stages.

[0100] The steps for determining spatial regionalization variables are as follows:

[0101] The cadmium content collected in step 3 was processed to conform to a normal distribution. Local Moran's I statistical analysis was performed on the processed cadmium content to determine the ranges of high and low clustering values ​​for the target planting area, see [link to relevant documentation]. Figure 2 Based on the ranges of high and low clustering values, the target planting area was divided into different sub-regions using the natural discontinuity method. The average cadmium content at sampling points within each sub-region was taken as a spatial regionalization variable.

[0102] The steps for determining spatial autocorrelation variables are as follows:

[0103] For semivariance function (e.g.) Figure 3 The variable-range distance was obtained by fitting the data (as shown in the image), and the results are shown in Table 2. The distances between the preset location and the five nearest neighbor sampling points were calculated to obtain the distance set. The variable-range distance and the distances within the distance set were used as spatial autocorrelation variables.

[0104] Table 2. Parameters of the Semivariance Function

[0105]

[0106] In the target planting area, environmental variables and spatial structure variables were characterized as input data, and cadmium content in rice was used as output data to establish a convolutional neural network model. This convolutional neural network model is the Convolutional Neural Network-Rice Heavy Metal (CNN-RHM) model.

[0107] Specifically, the CNN-RHM model consists of a series of interconnected convolutional layers, activation functions, pooling layers, fully connected layers, and an output layer. In the CNN-RHM model, the main influencing factors are first convolved by convolutional layers to further extract features. There are 10 convolutional kernels; each kernel is 3×3 in size, generating 10 feature maps representing different features extracted from the original input data. The stride is set to 1 to ensure consistency between input and output dimensions. After the convolutional layers, an activation function supporting the convolutional layers is added; this model uses the common ReLU activation function. Next is the pooling layer, which uses max pooling to improve prediction accuracy. The pooling window size is 2×2, and the stride is 2. Following the pooling layers are fully connected layers arranged in a chain, with 100 neurons. Finally, the output layer is formed.

[0108] Step 5: Train and validate the neural network model constructed in Step 4 to obtain the cadmium content prediction model.

[0109] The specific process is as follows: the input data and output data are divided into training set and validation set according to a ratio of 70%:30% respectively. The training set is input into the model, the mean squared error loss function is selected, and the gradient descent and its variant Adam optimization algorithm are selected. The model is trained with a learning rate of 0.001. By continuously adjusting the setting parameters of the convolutional layer, pooling layer and fully connected layer, the optimal CNN-RHM model is obtained.

[0110] The validation set is input into the trained model, and the accuracy evaluation metric is the coefficient of determination (R²). 2 The model was validated using root mean square error (RMSE) and mean absolute error (MAE). The results show that the CNN-RHM model, which fully considers environmental variables, has a higher R-squared value. 2 With a value >0.85 and low MAE and RMSE, and the predicted value being closest to the test result, it demonstrates good performance in modeling and prediction. Figure 4 The blue line represents a linear fit between the predicted and measured values ​​at the sampling points. Since the predicted value should equal the measured value, a reference line (gray line) is provided to show how close the blue line is to the measured value. Figure 4 It can be seen that the CNN-RHM model, which fully considers environmental and spatial structure variables, performs very well in predicting cadmium content in rice in the study area. Figure 4 Part (a) is the ordinary kriging model that only considers spatial structure variables and not environmental variables. Figure 4 Part (b) is the CNN-RHM model that fully considers environmental variables and spatial structure variables.

[0111] Step 6: Based on the cadmium content prediction model obtained in Step 5, predict the cadmium content of rice at preset locations in the target planting area to obtain the predicted cadmium content and distribution map of rice in the entire target planting area.

[0112] The specific process is as follows: First, determine the preset location, then obtain the latitude and longitude of that location, extract the key influencing factors at that location, and substitute them into the cadmium content prediction model to predict the cadmium content at that location. Ensure sufficient sample points are available, then convert the data to a raster map to draw a cadmium content distribution map. Figure 5 As shown, Figure 5 Part A of the diagram shows the sampling point layout. Figure 5 Part B in the diagram represents the layout scheme for the predicted points.

[0113] Step 7: Embed the predicted cadmium content in rice obtained in Step 6 into the whole-process holographic traceability chain for grain and oil, and obtain the cadmium content values ​​of other links.

[0114] Specifically, irrigation, soil, atmosphere, and pesticide application can all cause changes in the cadmium content of grains during the production process. Therefore, cadmium detection and prediction are carried out during the grain planting stage, and the detected and predicted cadmium content is automatically assigned to the production stage of the traceability chain through the traceability system. That is, the predicted cadmium content obtained in step 6 above, together with the cadmium content obtained from sampling in step 3, is stored on the traceability chain as the cadmium content of the production stage.

[0115] During the purchase of newly harvested grain, cadmium testing is conducted to obtain the cadmium content in the grain, which is used as the cadmium content for the target grain and oil products at the purchase stage. During storage, since the grain is mostly in a static state, the impact of cadmium is almost negligible; therefore, the cadmium content at the time of storage is determined as the cadmium content for the target grain and oil products at the storage stage. After leaving storage and being transported to the processing stage, the cadmium content at the time of leaving storage is tested to obtain the cadmium content for the processing stage. The target grain and oil products are then processed, and their cadmium content is measured. This measured cadmium content is compared with the cadmium content at the production, purchase, and storage stages, and threshold judgments are made at each stage. Products that do not exceed the threshold continue to the sales stage. The cadmium content collected at each stage of production, purchase, storage, processing, and sales is embedded in the full-process holistic traceability chain of the grain and oil product prevention, early warning, and traceability system.

[0116] Step 8: Construct a knowledge graph and, in conjunction with the key heavy metal concerns of the HACCP quality system, use the established knowledge graph to predict and warn of cadmium levels along the vertical timeline of grain circulation, from production to purchase, storage, processing, and sales.

[0117] Specifically, based on the cadmium threshold range in the national standard, the initial raw data for cadmium early warning, namely the cadmium threshold range, needed to construct the knowledge graph was obtained.

[0118] Based on cadmium threshold ranges, cadmium content in the production, acquisition, storage, processing, and sales stages of target grain and oil products within the current period, combined with regional characteristics and climate change, natural language processing technology is used to identify entities stored in the full-process holographic traceability chain. This identifies entities in the grain and oil full-process holographic traceability chain, such as production enterprises, products, batches, and transport vehicles. Relationship extraction is performed on these entities to identify relationships between them, such as the relationship between production enterprises and products, products and batches, and batches and transport vehicles. Entities and relationships are represented as nodes and edges in a graph, allowing the extraction of a cadmium-related knowledge graph constructed using cadmium thresholds.

[0119] In the next cycle, the cadmium content in other stages is predicted based on the knowledge graph corresponding to cadmium and the cadmium content in the production stage of the next cycle, so as to achieve cadmium content prediction on the vertical time axis of "production, purchase, storage, processing and sales", and to provide early warning in combination with the cadmium content safety threshold specified in the HACCP quality system.

[0120] Inference algorithms are used to reason about the knowledge graph, uncovering hidden relationships and patterns between its nodes. The knowledge graph is then visualized, providing an intuitive graphical interface for easy viewing and analysis by users.

[0121] This application establishes a knowledge graph and a pollutant content prediction model to realize the functions of grain product prevention and control early warning and monitoring data, covering the entire process of grain production, purchase, storage, processing and sales.

[0122] Based on the same inventive concept, this application also provides a holographic full-chain grain and oil product quality and safety real-time intelligent detection and prevention and early warning device for implementing the aforementioned holographic full-chain grain and oil product quality and safety real-time intelligent detection and prevention and early warning method. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the holographic full-chain grain and oil product quality and safety real-time intelligent detection and prevention and early warning device provided below can be found in the limitations of the holographic full-chain grain and oil product quality and safety real-time intelligent detection and prevention and early warning method described above, and will not be repeated here.

[0123] In one exemplary embodiment, a holographic, full-chain grain and oil product quality and safety real-time intelligent detection and prevention early warning device is provided, comprising:

[0124] The pollutant content prediction module is used to, within the current cycle, input the values ​​of the main influencing factors at each preset location in the target planting area to the pollutant content prediction model corresponding to the crop variety, and obtain the content of each pollutant in the crop variety planted at each preset location. The crop variety is used to produce target grain and oil products. One cycle includes the production, acquisition, storage, processing, and sales stages of the target grain and oil products. The pollutant content prediction model corresponding to the crop variety is obtained by training a convolutional neural network model based on the main influencing factors at each target sampling point in the target planting area within the current cycle and the pollutant content in the crop variety planted at each target sampling point. The main influencing factors are obtained by screening environmental variables and spatial structure variables.

[0125] The production process pollutant content determination module is used to determine the content of each pollutant in the crop of the variety planted at each preset location and each target sampling point as the content of each pollutant in the production process of the target grain and oil product.

[0126] The module for determining pollutant content in the acquisition process is used to obtain the content of each pollutant in the crops of the variety planted at each preset location and each target sampling point after the new harvest, as the content of each pollutant in the acquisition process of the target grain and oil products.

[0127] The module for determining the content of pollutants in the storage and processing stages is used to store the crops of the aforementioned varieties planted at each preset location and each target sampling point after acquisition into a warehouse, determine the content of each pollutant at the time of entry into the warehouse as the content of each pollutant in the storage stage of the target grain and oil products, and determine the content of each pollutant at the time of exit from the warehouse as the content of each pollutant in the processing stage of the target grain and oil products.

[0128] The module for determining the contaminant content in the sales process is used to process the crops of the specified variety after they leave the warehouse to obtain the target grain and oil products. The content of each contaminant in the target grain and oil products to be sold is used as the content of each contaminant in the sales process of the target grain and oil products; the content of each contaminant in the target grain and oil products to be sold is within the preset range.

[0129] The knowledge graph construction module is used to construct a knowledge graph corresponding to each pollutant based on the threshold range of each pollutant, the content of each pollutant in the production, acquisition, storage, processing and sales stages of the target grain and oil products in the current period, and the entities on the whole-process holographic traceability chain of grain and oil.

[0130] The prevention and control early warning module is used to predict the content of each pollutant in the acquisition, storage, processing and sales stages of the target grain and oil products in the next cycle based on the knowledge graph corresponding to each pollutant and the content of each pollutant in the production stage of the target grain and oil products in the next cycle, and to carry out prevention and control early warning based on the predicted value.

[0131] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0132] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0133] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A holographic, full-chain, real-time intelligent detection, prevention, and early warning method for the quality and safety of grain and oil products, characterized in that, The holographic, full-chain grain and oil product quality and safety real-time intelligent detection, prevention, and early warning method includes: For any crop variety planted in the target planting area, within the current cycle, the values ​​of the main influencing factors at each preset location in the target planting area are input into the pollutant content prediction model corresponding to the crop variety to obtain the content of each pollutant in the crop variety planted at each preset location; the crop variety is used to produce target grain and oil products; one cycle includes the production, acquisition, storage, processing, and sales stages of the target grain and oil products; the pollutant content prediction model corresponding to the crop variety is obtained by training a convolutional neural network model based on the main influencing factors at each target sampling point in the target planting area within the current cycle and the pollutant content in the crop variety planted at each target sampling point; the main influencing factors are obtained by screening environmental variables and spatial structure variables; The content of each pollutant in the crops of the variety planted at each preset location and each target sampling point is determined as the content of each pollutant in the production process of the target grain and oil products. The content of each pollutant in the crops of the variety planted at each preset location and each target sampling point after the new harvest is obtained as the content of each pollutant in the target grain and oil product acquisition process. The crops of the aforementioned varieties planted at each preset location and each target sampling point are stored in a warehouse. The content of each pollutant at the time of entry into the warehouse is determined as the content of each pollutant in the storage stage of the target grain and oil products, and the content of each pollutant at the time of exit from the warehouse is determined as the content of each pollutant in the processing stage of the target grain and oil products. After the crops are removed from the warehouse, the target grain and oil products are processed. The content of each pollutant in the target grain and oil products to be sold is taken as the content of each pollutant in the sales process of the target grain and oil products. The content of each pollutant in the target grain and oil products to be sold is within the preset range. Based on the threshold range of each pollutant, the content of each pollutant in the production, acquisition, storage, processing and sales stages of the target grain and oil products in the current cycle, and the entities on the whole-process holographic traceability chain of grain and oil, a knowledge graph corresponding to each pollutant is constructed. Based on the knowledge graph corresponding to each pollutant and the content of each pollutant in the production process of the target grain and oil products in the next cycle, the content of each pollutant in the acquisition, storage, processing and sales processes of the target grain and oil products in the next cycle is predicted to obtain the predicted value, and prevention and control warnings are carried out based on the predicted value.

2. The holographic, full-chain grain and oil product quality and safety real-time intelligent detection, prevention, and early warning method according to claim 1, characterized in that, The process of determining the target sampling points specifically includes: For any crop variety planted in the target planting area, acquire remote sensing data of the target planting area within the current period; The remote sensing data is divided into grids, and the crop planting density and area of ​​each grid are determined; the crop of the specified variety is planted within each grid. Based on the crop planting density and area of ​​each grid, the remote sensing data is divided according to the crop planting density level to obtain all grids corresponding to each crop planting density level; Based on all grids corresponding to each crop planting density level, the sampling points of the target planting area are determined as target sampling points according to the principle of stratified sampling.

3. The holographic, full-chain, real-time intelligent detection and early warning method for grain and oil product quality and safety according to claim 1, characterized in that, Based on the threshold ranges of each pollutant, the content of each pollutant in the production, acquisition, storage, processing, and sales stages of the target grain and oil products within the current period, and the entities on the grain and oil full-process holographic traceability chain, a knowledge graph corresponding to each pollutant is constructed, specifically including: Based on the threshold range of each pollutant and the content of each pollutant in the production, acquisition, storage, processing and sales stages of the target grain and oil products in the current cycle, entity identification and relationship identification are performed on the entities in the whole-process holographic traceability chain of grain and oil to obtain multiple entities and the relationships between them. Construct a knowledge graph for each pollutant based on the entities and the relationships between them.

4. The holographic, full-chain grain and oil product quality and safety real-time intelligent detection, prevention, and early warning method according to claim 1, characterized in that, Before constructing a knowledge graph for each pollutant based on its threshold range, the content of each pollutant in the production, acquisition, storage, processing, and sales stages of the target grain and oil products within the current period, and the entities on the grain and oil full-process holographic traceability chain, the following steps are also included: The target data, monitoring data, and values ​​of major influencing factors at each preset location within the target planting area during the current period, as well as the target data, monitoring data, and values ​​of major influencing factors at each target sampling point, are stored on the whole-process holographic traceability chain for grain and oil. The target data includes: latitude and longitude, crop variety, location, and sampling time. The monitoring data includes: crop production information, crop planting information, crop fertilization and irrigation information, crop pest and disease control information, and the content of various pollutants in the production, acquisition, storage, processing, and sales stages of the target grain and oil products.

5. The holographic, full-chain grain and oil product quality and safety real-time intelligent detection, prevention, and early warning method according to claim 1, characterized in that, Based on the knowledge graph corresponding to each pollutant and the content of each pollutant in the production process of the target grain and oil products in the next cycle, the content of each pollutant in the acquisition, storage, processing and sales stages of the target grain and oil products in the next cycle is predicted to obtain the predicted values. Prevention and early warning are then carried out based on the predicted values, specifically including: For any one of the acquisition, storage, processing and sales stages of the target grain and oil products in the next cycle, the content of each pollutant in the next cycle is predicted based on the knowledge graph corresponding to each pollutant and the content of each pollutant in the production stage of the target grain and oil products in the next cycle. For any pollutant, if the predicted value of the pollutant content in the next cycle exceeds a set threshold, then a prevention and control warning will be issued for the pollutant in the next cycle.

6. The holographic, full-chain, real-time intelligent detection and early warning method for grain and oil product quality and safety according to claim 1, characterized in that, The environmental variables include ecological elements and human activity elements, and the spatial structure variables include latitude and longitude, spatial autocorrelation variables, and spatial regionalization variables.

7. The holographic, full-chain grain and oil product quality and safety real-time intelligent detection, prevention, and early warning method according to claim 6, characterized in that, For any crop variety planted in the target planting area, the process for determining the value of the spatial regionalization variable at the preset location of the target planting area is as follows: For any pollutant, the pollutant content in the crops of the specified variety planted at each target sampling point in the target planting area is processed to obtain the treated pollutant content in the crops of the specified variety planted at each target sampling point in the target planting area; the treated pollutant content in the crops of the specified variety planted at each target sampling point in the target planting area conforms to a normal distribution. Spatial statistics were performed on the content of the pollutants after treatment in the crops of the variety planted at each target sampling point in the target planting area to obtain the range of high clustering values ​​and the range of low clustering values ​​in the target planting area. Based on the high and low clustering value ranges of the target planting area, the target planting area is divided into multiple sub-regions using the natural discontinuity method. The average value of the pollutants corresponding to each sub-region is determined as the value of the spatial regionalization variable. The average value of the pollutants corresponding to each sub-region is the average content of the pollutants in the crops used to produce grain and oil products planted at each target sampling point within the sub-region.

8. The holographic, full-chain grain and oil product quality and safety real-time intelligent detection and early warning method according to claim 7, characterized in that, Spatial statistics were performed on the content of the pollutants after treatment in the crops of the aforementioned varieties planted at each target sampling point in the target planting area to obtain the high clustering value range and low clustering value range of the target planting area, specifically: The high and low clustering value ranges of the target planting area were obtained by performing local Moran's I spatial statistics on the content of the pollutants after treatment in the crops of the variety planted at each target sampling point in the target planting area.

9. The holographic, full-chain, real-time intelligent detection and early warning method for grain and oil product quality and safety according to claim 6, characterized in that, The process for determining the value of the spatial autocorrelation variable at the preset location is as follows: The range distance is obtained by fitting the semivariogram function; For any preset location, the distances between each point in the nearest neighbor set corresponding to the preset location and the preset location are determined to obtain the distance set corresponding to the preset location; The set of nearest neighbors corresponding to the preset location includes the k nearest neighbor target sampling points within the target planting area; The range distance and the distance set corresponding to the preset position are determined as the values ​​of the spatial autocorrelation variables at the preset position.

10. A holographic, full-chain grain and oil product quality and safety real-time intelligent detection, prevention, and early warning device, characterized in that, The holographic, full-chain grain and oil product quality and safety real-time intelligent detection, prevention, and early warning device includes: The pollutant content prediction module is used to, within the current cycle, input the values ​​of the main influencing factors at each preset location in the target planting area for any crop variety to the pollutant content prediction model corresponding to that crop variety, thereby obtaining the content of each pollutant in the crop variety planted at each preset location; the crop variety is used to produce target grain and oil products; one cycle includes the production, acquisition, storage, processing, and sales stages of the target grain and oil products; the pollutant content prediction model corresponding to the crop variety is obtained by training a convolutional neural network model based on the main influencing factors at each target sampling point in the target planting area within the current cycle and the pollutant content in the crop variety planted at each target sampling point; the main influencing factors are obtained by screening environmental variables and spatial structure variables. The production process pollutant content determination module is used to determine the content of each pollutant in the crop of the variety planted at each preset location and each target sampling point as the content of each pollutant in the production process of the target grain and oil product. The module for determining the pollutant content in the acquisition process is used to obtain the content of each pollutant in the crop of the variety planted at each preset location and each target sampling point after the new harvest, as the content of each pollutant in the acquisition process of the target grain and oil products. The module for determining the content of pollutants in the storage and processing stages is used to store the crops of the aforementioned varieties planted at each preset location and each target sampling point after acquisition into a warehouse, determine the content of each pollutant at the time of entry into the warehouse as the content of each pollutant in the storage stage of the target grain and oil products, and determine the content of each pollutant at the time of exit from the warehouse as the content of each pollutant in the processing stage of the target grain and oil products. The module for determining the contaminant content in the sales process is used to process the crops of the aforementioned varieties after they leave the warehouse to obtain the target grain and oil products. The content of each contaminant in the target grain and oil products to be sold is used as the content of each contaminant in the sales process of the target grain and oil products; the content of each contaminant in the target grain and oil products to be sold is within the preset range. The knowledge graph construction module is used to construct the knowledge graph corresponding to each pollutant based on the threshold range of each pollutant, the content of each pollutant in the production, acquisition, storage, processing and sales stages of the target grain and oil products in the current period, and the entities on the whole-process holographic traceability chain of grain and oil. The prevention and control early warning module is used to predict the content of each pollutant in the acquisition, storage, processing and sales stages of the target grain and oil products in the next cycle based on the knowledge graph corresponding to each pollutant and the content of each pollutant in the production stage of the target grain and oil products in the next cycle, and to carry out prevention and control early warning based on the predicted value.

Images