A method for monitoring, processing and analyzing agricultural industry chain information data

Abstract

The present invention relates to the field of agricultural information technology, specifically: S1. Crop demand analysis; S2. Meteorological data impact assessment; S3. Soil fertilization strategy; S4. Crop yield prediction; S5. Market prediction, including: the present invention uses image processing technology to determine the growth stage of crops in crop images of each sampling area of ​​farmland, matches to obtain specific nutrient requirements corresponding to the crop growth stage, evaluates the influence coefficient of meteorological data on crop nutrient requirements by acquiring farmland meteorological data, generates a soil specific nutrient spatial distribution map by spatial interpolation technology, generates a variable fertilization prescription map based on the influence coefficient of meteorological data on crop nutrient requirements and the specific nutrient requirements corresponding to the crop growth stage, constructs a time series model for the crop maturity time and yield of each historical crop cycle, obtains the maturity time and yield prediction result of the current crop and determines whether to issue an early warning, thereby improving agricultural production efficiency and enhancing agricultural risk management capabilities.

Description

Technical Field

[0001] The present invention relates to the field of agricultural information technology, and in particular to a method for monitoring, processing and analyzing agricultural industry chain information data. Background Art

[0002] With the continued growth of the global population and increasing demands for food safety and quality, traditional agricultural production methods face enormous challenges. The transformation of agriculture to modernization, technology, and intelligence has become an inevitable trend. Against this backdrop, monitoring and processing information data from the agricultural industry chain has become crucial. The agricultural industry chain encompasses multiple links, including planting, breeding, processing, and sales, each of which generates massive amounts of data. Effective monitoring and processing of this data can achieve precise management and optimized allocation of agricultural resources, improve agricultural production efficiency and quality, enhance agricultural risk resilience, and promote agricultural modernization.

[0003] In recent years, information technology has developed rapidly, and the application of technologies such as the Internet of Things, big data, and artificial intelligence in the agricultural field has gradually increased. However, the current monitoring and processing of information data in the agricultural industry chain is still in the development stage, and there is still a big gap from the goal of being comprehensive, efficient, and intelligent.

[0004] For example, the existing Chinese patent application number 202310744208.3 discloses a digital agricultural industry chain data processing and analysis system. This solution collects, stores, processes and applies data from different sources in agricultural data sets through the data management unit in the agricultural data warehouse, and builds an information system for "traceability and supervision" of agricultural product quality and safety through the industry end. It allows users to more intuitively understand the changes in each data and each turning point, which is conducive to agricultural decision-making and command.

[0005] However, the above patent has the following problems: First, although the solution monitors environmental parameters, disease conditions and other factors during crop growth, it does not associate them with historical related data, and cannot use historical data to mine patterns, accurately assess future crop growth conditions and provide early warning of potential risks.

[0006] 2. This plan only collects, categorizes, stores, and analyzes agricultural product market price data in real time in the price information unit, and does not involve price forecasting. When agricultural product market prices fluctuate greatly, price trends cannot be understood in advance, and it is difficult for all parties to formulate risk response strategies in advance. Summary of the Invention

[0007] In order to overcome the shortcomings of the background technology, an embodiment of the present invention provides an agricultural industry chain information data monitoring, processing and analysis method, which can effectively solve the problems involved in the above-mentioned background technology.

[0008] The purpose of the present invention can be achieved through the following technical solutions: A method for monitoring, processing and analyzing agricultural industry chain information data, the method comprising the following steps: S1. Crop demand analysis: obtaining crop images of each sampling area of ​​the farmland, judging the growth stage of the crop through image processing technology, and matching the specific nutrient requirements corresponding to the crop growth stage.

[0009] S2. Assessment of meteorological data impact: Obtain farmland meteorological data to assess the impact of meteorological data on crop nutrient requirements.

[0010] S3. Soil fertilization strategy: Analyze the specific nutrients in the sampled soil and use spatial interpolation technology to generate a spatial distribution map of the specific nutrients in the soil. Generate a variable fertilization prescription map based on the impact coefficient of meteorological data on crop nutrient demand and the specific nutrient demand corresponding to the crop growth stage.

[0011] S4. Crop yield prediction: Build a time series model based on the maturity time and yield of each historical crop cycle, obtain the maturity time and yield prediction results of the current crop, and determine whether to issue an early warning.

[0012] S5. Market Forecast: Generate a decision tree based on the price statistics of agricultural products in the market and output the price forecast of agricultural products.

[0013] Preferably, the specific analysis method of the crop demand analysis is: obtaining crop images of each sampling area of ​​the farmland through low-altitude remote sensing by an unmanned aerial vehicle, graying the images, using edge detection technology to extract edge contours of crops in the crop images of each sampling area of ​​the farmland, calculating the crop area of ​​the crop images of each sampling area of ​​the farmland, comparing the crop area with the crop image area of ​​each sampling area of ​​the farmland, obtaining the vegetation coverage of each sampling area of ​​the farmland, extracting a set amount of crops from the farmland, obtaining their images, extracting crop edge contours using edge detection technology, and comparing them with crop images of each growth stage stored in the management database to determine the growth stage of the crop, and matching the requirements for different nutrients at each growth stage of the crop stored in the management database to obtain the specific nutrient requirements corresponding to the crop growth stage, where the specific nutrient refers to a specific nutrient required for the crop of interest.

[0014] Preferably, the meteorological data includes rainfall, average temperature and air humidity.

[0015] Preferably, the specific analysis method for the meteorological data impact assessment is: setting several monitoring periods of equal length, recorded as each monitoring period, connecting to the meteorological data platform to obtain the rainfall, average temperature and air humidity in each monitoring period in the area where the farmland is located, comparing the meteorological data of each monitoring period in the area where the farmland is located with the set meteorological data threshold to obtain the comprehensive score of the meteorological data in each monitoring period in the area where the farmland is located, and accumulating the scores to obtain the impact coefficient of the meteorological data on the nutrient demand of the crop.

[0016] Preferably, the specific analysis method for performing specific nutrient analysis on the sampled soil is: setting each sampling node in each sampling area of ​​the farmland according to the principle of equal spacing, collecting a set amount of soil from each sampling node in each sampling area of ​​the farmland as a soil sample, obtaining soil samples from each sampling node in each sampling area of ​​the farmland, and obtaining the specific nutrient content of the soil samples from each sampling node in each sampling area of ​​the farmland by chemical analysis.

[0017] Preferably, the specific method for generating the soil specific nutrient spatial distribution map is as follows: taking each sampling area of ​​the farmland as each interpolation unit, selecting the center point of each interpolation unit as the interpolation point, respectively obtaining the distance between the interpolation point of each sampling area of ​​the farmland and each sampling node in the interpolation unit where it is located, assigning weights to the interpolation points of each sampling area of ​​the farmland and each sampling node in the interpolation unit where it is located according to the distance from near to far, accumulating the product of the weight of each sampling node in each sampling area of ​​the farmland and the specific nutrient content of the corresponding sampling node of the corresponding soil sample, obtaining the specific nutrient estimation value of the soil in each sampling area of ​​the farmland, obtaining the position coordinates of the interpolation point of each sampling area of ​​the farmland, and constructing an interpolation surface by combining the position coordinates of the interpolation point of each sampling area of ​​the farmland and the specific nutrient estimation value of the corresponding soil, and visually displaying the generated interpolation surface in a geographic information system, i.e., the soil specific nutrient spatial distribution map.

[0018] Preferably, the specific method for generating the variable fertilization prescription map is as follows: setting a stepped nutrient range for a specific nutrient, screening out farmland areas belonging to the same nutrient range according to the spatial distribution map of the specific soil nutrients, thereby dividing the farmland into fertilization areas, and numbering the fertilization areas of the farmland from large to small according to the specific nutrient content. , read the impact coefficient of meteorological data on crop nutrient demand, the specific nutrient demand corresponding to the crop growth stage, and calculate the required fertilizer amount for each fertilization area ,in Indicates the specific nutrient requirements corresponding to the crop growth stage. is the impact coefficient of meteorological data on crop nutrient demand, For the specific nutrient thresholds for each fertilization area, For the preset fertilizer utilization rate of farmland soil, a variable fertilizer prescription map is generated according to the fertilizer amount required in each fertilization area, and crops in each fertilization area are fertilized based on this map.

[0019] Preferably, the specific operation method of constructing the time series model is as follows: obtaining the historical crop growth data of each round and the meteorological data in the corresponding time period, as well as the crop maturity time and yield of each historical crop cycle, selecting a number of historical time points from the historical crop growth data of each round and the meteorological data in the corresponding time period according to the principle of equal time intervals, uniformly recording the corresponding crop growth data and the corresponding meteorological data as the planting data of each historical time point, sorting the historical time points in chronological order, generating time series data, and recording the planting data of each historical time point as , Indicates the The number of a historical time point, , select according to analysis requirements Moving average, for time series , calculate the moving average ,when hour, ,when Before calculation The average of the items is used as an approximate moving average, When Itself is used as a moving average, thus obtaining the The moving average is calculated based on the historical time points. The time series model is constructed based on the moving average of the items, the crop maturity time and yield of each historical crop cycle.

[0020] Preferably, the specific operation method of the crop yield prediction is: obtain the growth data and meteorological data of the current crop at each time point, calculate the The moving average of the items is input into the time series model to output the maturity time and yield prediction results of the current crop. The yield expectation threshold is set. If the yield prediction result of the current crop is less than the yield expectation threshold, an early warning is issued. If the yield prediction result of the current crop is greater than or equal to the yield expectation threshold, no early warning is required.

[0021] Preferably, the specific operation method of step S5 is: S51. Obtain statistical data on agricultural product prices in the market from the website of the Ministry of Agriculture and Rural Affairs, obtain the prices of specific agricultural products at each price time point for specific agricultural products, set a step price range, and when the price of a specific agricultural product at a certain price time point fluctuates to a new price range, mark the price time point, thereby obtaining each marked time point, and obtain the duration that each price range is maintained in the market by taking the difference between adjacent marked time points, which is recorded as the listing duration of each price range, and at the same time obtain the listing quantity of specific agricultural products within the listing duration of each price range.

[0022] S52. Define the duration of time for each price range to be on the market as feature 1, and the quantity of specific agricultural products on the market within each price range to be on the market as feature 2. Obtain the feature 1 data set and the feature 2 data set. According to the principle of equal spacing, divide the data corresponding to feature 1 and feature 2 into several intervals, and calculate the feature 1 and feature 2 intervals respectively. , The proportion of intervals in the feature data set , Representation feature 1 The number of the interval, , Representation feature 2 The number of the interval, , calculate the Gini coefficient of each interval of feature 1 and feature 2 according to the interval ratio 、 .

[0023] S53. Compare the Gini coefficients of each interval of Feature 1 and Feature 2, select the feature and partition point with the smallest Gini coefficient as the optimal partition feature and partition point of Feature 1 data set and Feature 2 data set, and divide Feature 1 data set and Feature 2 data set into two child nodes based on the selected optimal partition feature and partition point.

[0024] S54. For the newly generated child nodes, the number of data in each child node is less than the set threshold value, which is set as the stopping condition. If the number of data in a child node is greater than or equal to the set threshold value, repeat S52 and S53, continue to select the optimal partitioning features and partitioning points on the child nodes, perform node partitioning, and recursively construct the subtree. If the number of data in a child node is less than the set threshold value, mark it as a leaf node, calculate the average price of all data in the leaf node, and use the average price as the predicted value of the leaf node. This process continues until all nodes meet the stopping condition and the decision tree is constructed. Then, the characteristics of the agricultural products to be predicted are input into the decision tree, and the price prediction of the agricultural products is output.

[0025] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects: 1. The present invention generates a spatial distribution map of soil specific nutrients through spatial interpolation technology, which clearly shows the spatial variation of soil nutrients in farmland. Based on the influence coefficient of meteorological data on crop nutrient demand and the specific nutrient demand corresponding to the crop growth stage, a variable fertilizer prescription map is generated. Precise fertilization can be carried out according to the differences in soil nutrients in different areas of farmland, as well as the differences in meteorological conditions and crop growth stages in different areas.

[0026] 2. The present invention constructs a time series model for the maturity time and yield of each historical crop cycle to obtain the maturity time and yield prediction results of the current crop. The prediction results are compared with pre-set standards to determine whether the current crop growth is facing risks and whether an early warning is needed, so as to respond to potential risks and opportunities in advance and improve the stability and risk resistance of agricultural production.

[0027] 3. The present invention generates a decision tree based on the price statistics of agricultural products in the market and outputs price forecasts of agricultural products. In the case of large price fluctuations in the agricultural product market, understanding the price trends in advance will help all parties formulate risk response strategies in advance. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The present invention is further described with reference to the accompanying drawings. However, the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative effort.

[0029] Figure 1 Schematic diagram of the method of the present invention.

[0030] Figure 2 for Figure 1 Flow judgment block diagram of step S4.

[0031] Figure 3 for Figure 1 Schematic diagram of the method flow of step S5 in FIG. DETAILED DESCRIPTION

[0032] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0033] See also Figure 1As shown, the present invention provides a method for monitoring, processing and analyzing agricultural industry chain information data, which includes the following steps: S1. Crop demand analysis: obtaining crop images of each sampling area of ​​the farmland, judging the growth stage of the crop through image processing technology, and matching the specific nutrient demand corresponding to the crop growth stage.

[0034] The specific analysis method of the crop demand analysis is as follows: using low-altitude remote sensing of drones to obtain crop images of each sampling area of ​​the farmland, graying the images, using edge detection technology to extract edge contours of crops in the crop images of each sampling area of ​​the farmland, calculating the crop area of ​​the crop images of each sampling area of ​​the farmland, comparing the crop area with the area of ​​the crop images of each sampling area of ​​the farmland, obtaining the vegetation coverage of each sampling area of ​​the farmland, extracting a set amount of crops from the farmland, obtaining their images, extracting the edge contours of the crops using edge detection technology, and comparing them with the crop images of each growth stage stored in the management database to determine the growth stage of the crops, and matching the requirements for different nutrients of each growth stage of the crops stored in the management database to obtain the specific nutrient requirements corresponding to the crop growth stage; providing nutrients according to the actual needs of the crops helps to improve fertilizer utilization, reduce agricultural production costs, and improve the quality and quality of agricultural products. The specific nutrients refer to a specific nutrient component required for the crop of interest.

[0035] It should be noted that the specific analysis method of the edge detection technology is: grayscale processing is performed on the crop images of each sampling area of ​​the farmland, the grayscale change rate of the pixel points in the horizontal and vertical directions is calculated by the gradient operator to obtain the size and direction of the gradient, and two grayscale value thresholds, a high threshold and a low threshold, are set respectively. Pixel points with gradient values ​​greater than the high threshold are marked as strong edge points, pixel points with gradient values ​​between the low threshold and the high threshold are marked as weak edge points, and pixel points with gradient values ​​less than the low threshold are marked as non-edge points. Starting from the strong edge point, through an 8-neighborhood search, the weak edge points connected to the strong edge point are connected to form a complete edge contour, thereby obtaining the edge contour of the crop in the crop images of each sampling area of ​​the farmland.

[0036] S2. Meteorological data impact assessment: Obtain farmland meteorological data to assess the impact of meteorological data on crop nutrient requirements.

[0037] The meteorological data include rainfall, average temperature and air humidity.

[0038] The specific analysis method for the meteorological data impact assessment is as follows: setting a number of monitoring periods of equal length, recorded as each monitoring period, connecting to a meteorological data platform to obtain the rainfall, average temperature and air humidity in each monitoring period in the area where the farmland is located, comparing the meteorological data of each monitoring period in the area where the farmland is located with the set meteorological data threshold to obtain a comprehensive score of the meteorological data in each monitoring period in the area where the farmland is located, and accumulating the scores to obtain the impact coefficient of the meteorological data on the nutrient demand of the crop; this can more clearly observe the changes in meteorological data in different time periods, and ensure the timeliness and comprehensiveness of the data.

[0039] It should be noted that the specific analysis method for the comprehensive score of meteorological data in each monitoring period in the area where the farmland is located is as follows: set the meteorological data threshold, extract the maximum and minimum values ​​corresponding to rainfall, average temperature and air humidity, and record the maximum and minimum values ​​corresponding to rainfall as , the rainfall in each monitoring period in the area where the farmland is located is recorded as , Indicates the The number of the monitoring period, , through the formula Get the rainfall score for each monitoring period in the area where the farmland is located According to the method of analyzing the rainfall score in each monitoring period in the area where the farmland is located, the average temperature score and air humidity score in each monitoring period in the area where the farmland is located are obtained. , calculate the comprehensive score of meteorological data in each monitoring period in the area where the farmland is located : ,in Represent the weight factors of preset rainfall, average temperature and air humidity respectively.

[0040] It should be noted that the actual meteorological data is compared with the preset threshold value, and a score between 0 and 1 is given according to the comparison result, where 1 represents the most ideal meteorological conditions and 0 represents the least ideal conditions.

[0041] It should be noted that, in a specific embodiment, It can be set to 0.5, It can be set to 0.3, It can be set to 0.2. Precipitation is crucial to crop growth because it directly affects the water supply of crops. Excessive rainfall may lead to waterlogging, while insufficient rainfall may lead to drought. Temperature affects the physiological activities of crops, including photosynthesis and respiration. A suitable temperature range can promote crop growth, while too high or too low temperatures may inhibit growth or cause crop damage. Therefore, average temperature is also an important meteorological factor, but its weight may be slightly lower than rainfall because its impact may depend more on crop type and growth stage. Air humidity affects crop transpiration and the occurrence of pests and diseases. Suitable humidity can reduce water evaporation, keep the soil moist, and also help control pests and diseases. However, the impact of air humidity on crop growth may not be as direct as rainfall and temperature. Therefore, the weight corresponding to air humidity is the lowest, and the weight corresponding to rainfall is the highest.

[0042] S3. Soil fertilization strategy: Analyze the specific nutrients in the sampled soil and use spatial interpolation technology to generate a spatial distribution map of the specific nutrients in the soil. Generate a variable fertilization prescription map based on the impact coefficient of meteorological data on crop nutrient demand and the specific nutrient demand corresponding to the crop growth stage.

[0043] The specific analysis method for performing specific nutrient analysis on the sampled soil is as follows: sampling nodes are set in each sampling area of ​​the farmland according to the principle of equal spacing, a set amount of soil is collected from each sampling node in each sampling area of ​​the farmland as a soil sample, and soil samples are obtained from each sampling node in each sampling area of ​​the farmland. The specific nutrient content of the soil samples in each sampling node in each sampling area of ​​the farmland is obtained by chemical analysis; thus, accurate soil fertility information is provided for agricultural production, helping to formulate accurate fertilization plans based on soil nutrient conditions and realize fertilization on demand.

[0044] It should be noted that the specific division method of the equal spacing principle is: if the shape of the farmland is regular (such as a rectangle), it can be directly divided equally along the length and width of the farmland according to the set spacing to form a grid-like sampling node layout; if the shape of the farmland is irregular, it can be divided into several relatively regular sub-areas first, and then the sampling nodes are set for each sub-area according to the equal spacing principle.

[0045] It should be noted that the specific analysis method of the chemical analysis method is: for the pH of the soil, weigh a set amount of soil sample, put it into a beaker, add carbon dioxide-free water according to the set soil-water ratio, stir evenly to obtain a supernatant, and measure the pH value of the supernatant with a pH meter after standing.

[0046] For soil organic matter, a set amount of soil sample was weighed and placed in a hard test tube. A potassium dichromate-sulfuric acid solution of a set concentration and volume was added. The sample was heated and oxidized in an oil bath to allow the organic carbon in the soil to react with potassium dichromate. The remaining potassium dichromate was titrated with a standard ferrous sulfate solution, and the soil organic matter content was calculated based on the amount of ferrous sulfate consumed.

[0047] The specific method for generating the soil specific nutrient spatial distribution map is as follows: taking each farmland sampling area as each interpolation unit, selecting the center point of each interpolation unit as the interpolation point, respectively obtaining the distance between the interpolation point in each farmland sampling area and each sampling node in the interpolation unit, assigning weights to the interpolation points in each farmland sampling area and each sampling node in the interpolation unit according to the distance from near to far, accumulating the product of the weight of each sampling node in each farmland sampling area and the specific nutrient content of the corresponding sampling node of the corresponding soil sample to obtain the specific nutrient estimation value of the soil in each farmland sampling area, obtaining the position coordinates of the interpolation point in each farmland sampling area, combining the position coordinates of the interpolation point in each farmland sampling area with the specific nutrient estimation value of the corresponding soil, and visually displaying the generated interpolation surface in a geographic information system, i.e., the soil specific nutrient spatial distribution map. This makes the soil nutrient estimation value more reasonable and accurate, can better reflect the actual situation of soil nutrients around the interpolation point, avoids the estimation bias caused by simple averaging or ignoring the distance factor, and improves the accuracy and reliability of nutrient estimation.

[0048] It should be noted that the specific calculation method for allocating weights to the interpolation points of each sampling area of ​​farmland and each sampling node in the interpolation unit where they are located is as follows: , which is surrounded by Sampling nodes ( ), set the sampling node To interpolation point The distance is , use Euclidean distance to calculate interpolation points Weight .

[0049] It should be noted that the interpolation point The weight formula is based on the idea of ​​inverse distance weighted interpolation, that is, the closer the sampling node is to the interpolation point, the greater the influence on the interpolation point. In Euclidean space, the inverse square of the distance is widely used for weight distribution. A single sampling node Weight Reflects its relative influence on the interpolation of the interpolation point. According to the principle of inverse distance weight, the distance The reciprocal square of It can be used as a preliminary indicator to measure its influence. In order to make the sum of the weights of all sampling nodes equal to 1 (to ensure the normalization of weights), it is necessary to Normalize each Divide by the total number of sampling nodes sum .

[0050] The specific method for generating the variable fertilization prescription map is as follows: setting a ladder nutrient range for a specific nutrient, screening out farmland areas belonging to the same nutrient range according to the spatial distribution map of soil specific nutrients, dividing the farmland into fertilization areas, and numbering each fertilization area of ​​the farmland from high to low according to the specific nutrient content. , read the impact coefficient of meteorological data on crop nutrient demand, the specific nutrient demand corresponding to the crop growth stage, and calculate the required fertilizer amount for each fertilization area ,in Indicates the specific nutrient requirements corresponding to the crop growth stage. is the impact coefficient of meteorological data on crop nutrient demand, For the specific nutrient thresholds for each fertilization area, For the preset fertilizer utilization rate of farmland soil, a variable fertilizer prescription map is generated according to the fertilizer amount required in each fertilization area, and crops in each fertilization area are fertilized based on this map; this facilitates the formulation of fertilization plans and the arrangement of fertilization operations, and improves the operability and accuracy of fertilization.

[0051] It should be noted that among the fertilizer requirements in each fertilization area, Indicates the specific nutrient requirement corresponding to the crop growth stage. It is the amount of specific nutrients required for the crop to grow to the current stage under ideal conditions. However, in actual conditions, meteorological conditions will affect the nutrient demand of crops. Therefore, the influence coefficient of meteorological data on crop nutrient demand is introduced. , It indicates the total amount of specific nutrients that crops actually need to obtain from the soil after taking into account the influence of meteorological factors. For the The specific nutrient threshold of a fertilization area is the level of specific nutrient content in the soil of the area. The total amount of specific nutrients that crops actually need to obtain minus the existing specific nutrient content in the area is used to obtain the amount of specific nutrients that crops in the fertilization area need to supplement. The fertilizer utilization rate of the preset farmland soil indicates the proportion of fertilizer applied to the soil that can be absorbed and utilized by crops. Because the fertilizer utilization rate is not 100% during actual fertilization, the amount of nutrients that the crops need to supplement is divided by the fertilizer utilization rate to obtain the actual amount of fertilizer that needs to be applied to the fertilization area, that is, .

[0052] A feasible simulation process is to plant a wheat field and divide it into three fertilization areas ( ), it is known that wheat requires nitrogen (specific nutrients) during the jointing stage (growth stage) =15 kg / mu. Meteorological data show that the current meteorological conditions are conducive to the absorption of nitrogen by wheat. The coefficient of influence of meteorological data on crop nutrient demand is =0.2, the preset fertilizer utilization rate of farmland soil =0.4, the simulation results can be obtained. Please refer to Table 1 for details. Table 1 lists some representative data.

[0053] Table 1. Partially collected data and calculation results for each fertilization area

[0054] Fertilization area Specific nutrient threshold (kg / mu) The actual amount of fertilizer that needs to be applied to the fertilization area (kg / mu) 1 5 #timg# 2 8 #timg# 3 3 #timg#

[0055] S4. Crop yield prediction: Build a time series model based on the maturity time and yield of each historical crop cycle, obtain the maturity time and yield prediction results of the current crop, and determine whether to issue an early warning.

[0056] See also Figure 2 As shown, the specific operation method of constructing the time series model is as follows: obtain the historical crop growth data of each round and the meteorological data in the corresponding time period, as well as the crop maturity time and yield of each historical crop round, select several historical time points from the historical crop growth data of each round and the meteorological data in the corresponding time period according to the principle of equal time intervals, record the corresponding crop growth data and the corresponding meteorological data as the planting data of each historical time point, sort the historical time points in chronological order, generate time series data, and record the planting data of each historical time point as , Indicates the The number of a historical time point, , select according to analysis requirements Moving average, for time series , calculate the moving average ,when hour, ,when Before calculation The average of the items is used as an approximate moving average, When Itself is used as a moving average, thus obtaining the The moving average value is based on the historical time points A time series model is constructed by taking the moving average of the items and the crop maturity time and yield of each historical crop cycle as the basis. By learning and modeling historical data, the model can capture the potential relationship between crop growth and environmental factors and time, providing a scientific decision-making basis for agricultural production.

[0057] It should be noted that when When the amount of data is small, we can only use all the existing data to calculate an average value that can relatively reflect the overall trend. When, with The moving average value is calculated by taking the number of times forward from the current time point. The purpose of averaging the data points is to make the moving average reflect the average level of the data in a relatively stable period of time, eliminate the impact of accidental fluctuations of a single data point on the overall trend, and thus show the long-term trend of the data more clearly.

[0058] The specific operation method of the crop yield prediction is: obtain the growth data and meteorological data of the current crop at each time point, calculate the The moving average of the items is input into the time series model, and the maturity time and yield prediction results of the current crop are output. The yield expectation threshold is set. If the yield prediction result of the current crop is less than the yield expectation threshold, an early warning is issued. If the yield prediction result of the current crop is greater than or equal to the yield expectation threshold, no early warning is required. This provides a risk early warning mechanism for agricultural production. When the yield may be lower than expected, timely measures can be taken, such as increasing fertilization, strengthening pest and disease control, etc., to minimize losses. When the yield meets or exceeds expectations, production and sales can be reasonably arranged to improve economic benefits.

[0059] S5. Market Forecast: Generate a decision tree based on the price statistics of agricultural products in the market and output the price forecast of agricultural products.

[0060] See also Figure 3 As shown, the specific operation method of step S5 is: S51. Obtain price statistics of agricultural products in the market from the website of the Ministry of Agriculture and Rural Affairs, obtain the prices of specific agricultural products at each price time point for specific agricultural products, set a step price range, and when the price of a specific agricultural product at a certain price time point fluctuates to a new price range, mark the price time point, thereby obtaining each marked time point, and obtain the duration of each price range maintained in the market by subtracting adjacent marked time points, which is recorded as the listing duration of each price range, and at the same time obtain the listing volume of specific agricultural products within the listing duration of each price range; it provides a clear time mark for subsequent analysis of price change patterns, facilitates understanding of the fluctuation of specific agricultural product prices at different stages, helps to grasp market price trends, and provides important price information reference for producers, sellers and consumers.

[0061] S52. Define the duration of time for each price range to be on the market as feature 1, and the quantity of specific agricultural products on the market within each price range to be on the market as feature 2. Obtain the feature 1 data set and the feature 2 data set. According to the principle of equal spacing, divide the data corresponding to feature 1 and feature 2 into several intervals, and calculate the feature 1 and feature 2 intervals respectively. , The proportion of intervals in the feature data set , Representation feature 1 The number of the interval, , Representation feature 2 The number of the interval, , calculate the Gini coefficient of each interval of feature 1 and feature 2 according to the interval ratio 、 The Gini coefficient can be used to quantify the distribution of feature data and understand the degree of concentration and difference of the data. A smaller Gini coefficient indicates that the data distribution is more uniform, and a larger Gini coefficient indicates that the data distribution is more concentrated.

[0062] It should be noted that, taking the Gini coefficient of each interval of feature 1 as an example, Representation feature 1 The proportion of intervals in the node, that is, the data of feature 1 is divided into intervals, and the proportion of each interval in the total is , assuming that the proportions of all intervals are equal, that is , the distribution is most uniform at this time, ,hesitate is a constant, and according to the constant summation formula we can get When the distribution is completely uniform, the Gini coefficient is defined as 0. As the degree of uneven distribution increases, the Gini coefficient gradually increases to 1. In order to make the Gini coefficient 0 when the distribution is completely uniform, subtract 1 from ,get , when all data are concentrated in one interval, e.g. ,the remaining ,at this time , then the Gini coefficient , indicating the most uniform distribution.

[0063] It should be noted that the calculation of feature 1 and feature 2 The specific analysis method for the proportion of each interval in the node is as follows: filter out the maximum and minimum values ​​of feature 1 (the duration of listing in each price range) and feature 2 (the market volume of specific agricultural products within each price range), obtain the data range of feature 1 and feature 2, divide the data range of feature 1 and feature 2 into several intervals according to the principle of equal spacing, traverse all data of feature 1 and feature 2, count the number of data points in each interval, divide the number of data points in each interval by the total number of feature data, and obtain the proportion of each interval in the node.

[0064] S53. Compare the Gini coefficients of each interval of Feature 1 and Feature 2, select the feature and partition point with the smallest Gini coefficient as the optimal partition feature and partition point of Feature 1 data set and Feature 2 data set, and divide Feature 1 data set and Feature 2 data set into two sub-nodes based on the selected optimal partition feature and partition point; by continuously dividing nodes, the decision tree can gradually build a complex classification structure and accurately classify and predict data under different feature combinations. Each sub-node represents a more specific data subset, which helps to analyze the relationship between data more deeply and improve the model's understanding and prediction capabilities of market price changes.

[0065] It should be noted that, in a specific embodiment, the duration of the price range is used as the division feature, and the optimal division point found is 10 days. Then, samples with a price range duration of less than 10 days are divided into one sub-node, and samples with a price range duration of greater than or equal to 10 days are divided into another sub-node.

[0066] S54. For the newly generated child nodes, the number of data in each child node is less than the set threshold value, which is set as the stopping condition. If the number of data in a child node is greater than or equal to the set threshold value, repeat S52 and S53, continue to select the optimal partitioning features and partitioning points on the child nodes, perform node partitioning, and recursively construct the subtree. If the number of data in a child node is less than the set threshold value, mark it as a leaf node, calculate the average price of all data in the leaf node, and use the average price as the predicted value of the leaf node. This process continues until all nodes meet the stopping condition and the decision tree is constructed. Then, the agricultural product features to be predicted are input into the decision tree, and the price prediction of the agricultural product is output. The predicted value of the leaf node provides a specific numerical reference for the price prediction of the agricultural product. As an intuitive model structure, the decision tree can quickly and accurately output the price prediction results based on the input agricultural product features.

[0067] It should be noted that in a specific embodiment, there is price statistics of apples in the past 365 days obtained from the website of the Ministry of Agriculture and Rural Affairs. For apples, the price of each day (price time point) is recorded, and the step price range is set as: low price range (less than 8 yuan / jin), medium price range (8-12 yuan / jin), and high price range (more than 12 yuan / jin).

[0068] During these 365 days, on the 30th day, the price of apples rose from 7.5 yuan / jin to 8.2 yuan / jin, fluctuating into the middle price range, marked as day 30; on the 120th day, the price rose from 11.8 yuan / jin to 12.5 yuan / jin, fluctuating into the high price range, marked as day 120; on the 200th day, the price fell from 13 yuan / jin to 11.5 yuan / jin, returning to the middle price range, marked as day 200.

[0069] The duration of listing in the low price range is 30 days, the duration of listing in the medium price range after the first fluctuation is 120-30=90 days, the duration of listing in the high price range is 200-120=80 days, and the duration of listing in the medium price range after the second fluctuation is 365-200=165 days. At the same time, the market volume of apples within the low price range is 50 tons, the market volume within the first medium price range is 120 tons, the market volume within the high price range is 90 tons, and the market volume within the second medium price range is 180 tons.

[0070] The duration of time the apples are put on the market in each price range (feature 1) is defined as: [30, 90, 80, 165], and the quantity of apples put on the market in each price range (feature 2) is defined as: [50, 120, 90, 180]. The feature 1 data set and feature 2 data set are obtained. According to the principle of equal interval, the data range of feature 1 [30, 165] is divided into three intervals: [30-70], [71-110], and [111-165]. The data range of feature 2 [50, 180] is also divided into three intervals: [50-90], [91-130], and [131-180].

[0071] Statistics show that the number of data points of feature 1 in the three intervals are 1, 1, and 2 respectively. Then the proportion of each interval in the feature 1 data set is , the number of data points of feature 2 in the three intervals are 1, 1, and 2 respectively, and the proportion of each interval in the feature 2 data set is , calculate the Gini coefficient of feature 1 and feature 2 according to the formula .

[0072] Since the Gini coefficients of feature 1 and feature 2 are equal, feature 1 is randomly selected as the optimal partitioning feature. The data set [30, 90, 80, 165] of feature 1 is analyzed, and the optimal partitioning point is determined to be 80 days. At this time, the samples in the feature 1 data set with a price range duration of less than 80 days (i.e., samples corresponding to 30 days) are divided into one child node, and the samples with a price range duration of greater than or equal to 80 days (i.e., samples corresponding to 90 days, 80 days, and 165 days) are divided into another child node. At the same time, the feature 2 data set [50, 120, 90, 180] is divided accordingly according to the partitioning method of feature 1, that is, the samples corresponding to less than 80 days (here, the listed volume of 50 tons corresponding to the duration of 30 days) are divided into one child node, and the samples corresponding to greater than or equal to 80 days (i.e., the listed volumes of 120 tons, 90 tons, and 180 tons corresponding to the durations of 90 days, 80 days, and 165 days) are divided into another child node.

[0073] Set the stopping threshold for the number of child node data to 2. For the two newly generated child nodes: In the first child node, the number of samples in the feature 1 data set with a price range duration of less than 80 days is 1, and the number of samples in the feature 2 data set with the corresponding supply volume data is also 1, which is less than the set threshold of 2. Therefore, mark this child node as a leaf node, calculate the average price of the data in this leaf node (the data combination of apples with a low price range for 30 days and a supply volume of 50 tons), which is 7.5 yuan / jin, and use 7.5 yuan / jin as the predicted value for this leaf node.

[0074] In the second child node, the number of samples in the feature 1 data set whose price range lasts for more than or equal to 80 days is 3, and the number of corresponding listing volume data samples in the feature 2 data set is also 3, which is greater than the set threshold of 2. Therefore, it is necessary to repeat steps S52 and S53 until the number of data in all child nodes is less than the set threshold. These child nodes are marked as leaf nodes, and the average price of all data in each leaf node is calculated as the predicted value.

[0075] The apple-related feature data that needs to be predicted (such as the price range, duration of market launch, and market volume within a certain period of time) is input into the constructed decision tree. The decision tree quickly and accurately outputs the apple price prediction results based on its internal structure and rules.

[0076] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention, which are still covered by the scope of protection of the present invention.

Claims

1. A method for monitoring, processing and analyzing agricultural industry chain information data, characterized in that: The steps include: S1. Crop Requirements Analysis: Obtain crop images from each sampling area of ​​the farmland, use image processing technology to determine the crop growth stage, and match the specific nutrient requirements corresponding to the crop growth stage; S2. Meteorological data impact assessment: Obtain farmland meteorological data to assess the impact of meteorological data on crop nutrient requirements; S3. Soil Fertilization Strategy: Analyze specific nutrients in sampled soils and generate spatial distribution maps of specific nutrients using spatial interpolation techniques. Generate variable-rate fertilization prescription maps based on the impact coefficients of meteorological data on crop nutrient requirements and the specific nutrient requirements corresponding to crop growth stages. S4. Crop Yield Forecasting: Build a time series model based on the maturity and yield of each historical crop cycle, obtain the maturity and yield forecast results of the current crop, and determine whether to issue an early warning; S5. Market Forecast: Generate a decision tree based on the market price statistics of agricultural products and output the price forecast of agricultural products; The specific method for generating the variable rate fertilization prescription map is as follows: Set a ladder nutrient range for specific nutrients, and screen out farmland areas belonging to the same nutrient range based on the spatial distribution map of soil specific nutrients. In this way, the farmland is divided into fertilization areas, and the fertilization areas of the farmland are numbered from high to low according to the specific nutrient content. , read the impact coefficient of meteorological data on crop nutrient demand, the specific nutrient demand corresponding to the crop growth stage, and calculate the required fertilizer amount for each fertilization area ,in Indicates the specific nutrient requirements corresponding to the crop growth stage. is the impact coefficient of meteorological data on crop nutrient demand, For the specific nutrient thresholds for each fertilization area, For the preset fertilizer utilization rate of farmland soil, a variable fertilizer prescription map is generated according to the fertilizer amount required in each fertilization area, and crops in each fertilization area are fertilized based on this map.

2. The method for monitoring, processing and analyzing agricultural industry chain information data according to claim 1, characterized in that: The specific analysis method of the crop demand analysis is as follows: Crop images of each sampling area of ​​the farmland are obtained through low-altitude remote sensing using unmanned aerial vehicles, and grayscale processing is performed on them. Edge detection technology is used to extract the edge contours of the crops in the crop images of each sampling area of ​​the farmland, and the crop area of ​​the crop images of each sampling area of ​​the farmland is calculated. The crop area is compared with the crop image area of ​​each sampling area of ​​the farmland to obtain the vegetation coverage of each sampling area of ​​the farmland. A set amount of crops are extracted from the farmland and their images are obtained. The edge contours of the crops are extracted using edge detection technology, and the crop edge contours are compared with the crop images of each growth stage stored in the management database to determine the growth stage of the crops. By matching the requirements for different nutrients of each growth stage of the crops stored in the management database, the specific nutrient requirements corresponding to the crop growth stage are obtained. The specific nutrient refers to a specific nutrient required for the crop of interest.

3. The method for monitoring, processing and analyzing agricultural industry chain information data according to claim 1, characterized in that: The meteorological data include rainfall, average temperature and air humidity.

4. The method for monitoring, processing and analyzing agricultural industry chain information data according to claim 3, characterized in that: The specific analysis method for the meteorological data impact assessment is as follows: Set several monitoring periods of equal length, recorded as each monitoring period, and connect to the meteorological data platform to obtain the rainfall, average temperature and air humidity in each monitoring period in the area where the farmland is located. Compare the meteorological data of each monitoring period in the area where the farmland is located with the set meteorological data threshold to obtain the comprehensive score of the meteorological data in each monitoring period in the area where the farmland is located, and accumulate them to obtain the impact coefficient of meteorological data on crop nutrient demand.

5. The method for monitoring, processing and analyzing agricultural industry chain information data according to claim 1, characterized in that: The specific analysis method for performing specific nutrient analysis on the sampled soil is: According to the principle of equal spacing, sampling nodes are set in each sampling area of ​​the farmland. A set amount of soil is collected from each sampling node in each sampling area of ​​the farmland as soil samples to obtain soil samples from each sampling node in each sampling area of ​​the farmland. The specific nutrient content of the soil samples from each sampling node in each sampling area of ​​the farmland is obtained by chemical analysis.

6. The method for monitoring, processing and analyzing agricultural industry chain information data according to claim 5, characterized in that: The specific method for generating the soil specific nutrient spatial distribution map is as follows: Each sampling area of ​​the farmland is taken as each interpolation unit, and the center point of each interpolation unit is selected as the interpolation point. The distance between the interpolation point of each sampling area of ​​the farmland and each sampling node in the interpolation unit is obtained respectively. The weights are assigned to the interpolation points of each sampling area of ​​the farmland and each sampling node in the interpolation unit according to the distance from near to far. The specific nutrient estimation value of the soil in each sampling area of ​​the farmland is obtained by accumulating the product of the weight of each sampling node in each sampling area of ​​the farmland and the specific nutrient content of the corresponding sampling node of the corresponding soil sample. The position coordinates of the interpolation point of each sampling area of ​​the farmland are obtained. The position coordinates of the interpolation point of each sampling area of ​​the farmland and the specific nutrient estimation value of the corresponding soil are combined to construct an interpolation surface. The generated interpolation surface is visualized in the geographic information system, that is, the spatial distribution map of soil specific nutrients.

7. The method for monitoring, processing and analyzing agricultural industry chain information data according to claim 1, characterized in that: The specific operation method of constructing the time series model is: Obtain the historical crop growth data of each round and the meteorological data in the corresponding time period, as well as the crop maturity time and yield of each historical crop round. According to the principle of equal time interval, select several historical time points from the historical crop growth data of each round and the meteorological data in the corresponding time period. Record the corresponding crop growth data and corresponding meteorological data as the planting data of each historical time point. Sort the historical time points in chronological order to generate time series data. Record the planting data of each historical time point as , Indicates the The number of a historical time point, , select according to analysis requirements Moving average, for time series , calculate the moving average ,when hour, ,when Before calculation The average of the items is used as an approximate moving average, When Itself is used as a moving average, thus obtaining the The moving average value is based on the historical time points The time series model is constructed based on the moving average of the items, the crop maturity time and yield of each historical crop cycle.

8. The method for monitoring, processing and analyzing agricultural industry chain information data according to claim 7, characterized in that: The specific operation method of the crop yield prediction is as follows: Obtain the current crop growth data and meteorological data at each time point, and calculate the The moving average of the items is input into the time series model to output the maturity time and yield prediction results of the current crop. The yield expectation threshold is set. If the yield prediction result of the current crop is less than the yield expectation threshold, an early warning is issued. If the yield prediction result of the current crop is greater than or equal to the yield expectation threshold, no early warning is required.

9. The method for monitoring, processing and analyzing agricultural industry chain information data according to claim 1, characterized in that: The specific operation method of step S5 is: S51. Obtain market agricultural product price statistics from the Ministry of Agriculture and Rural Affairs website. For specific agricultural products, obtain the price of the specific agricultural product at each price point in time. Set a stepped price range. When the price of the specific agricultural product at a certain price point in time fluctuates into a new price range, mark the price point in time. Thus, obtain each marked time point. By subtracting adjacent marked time points, calculate the duration that each price range persists in the market. This is recorded as the market duration of each price range. Simultaneously, obtain the market volume of the specific agricultural product within each price range in terms of market duration. S52. Define the duration of time for each price range to be on the market as feature 1, and the quantity of specific agricultural products on the market within each price range to be on the market as feature 2. Obtain the feature 1 data set and the feature 2 data set. According to the principle of equal spacing, divide the data corresponding to feature 1 and feature 2 into several intervals, and calculate the feature 1 and feature 2 intervals respectively. , The proportion of intervals in the feature data set , Representation feature 1 The number of the interval, , Representation feature 2 The number of the interval, , calculate the Gini coefficient of each interval of feature 1 and feature 2 according to the interval ratio 、 ; S53. Compare the Gini coefficients of each interval of Feature 1 and Feature 2, select the feature and partition point with the smallest Gini coefficient as the optimal partition feature and partition point for the Feature 1 data set and the Feature 2 data set, and partition the Feature 1 data set and the Feature 2 data set into two child nodes based on the selected optimal partition feature and partition point; S54. For the newly generated child nodes, the number of data in each child node is less than the set threshold value, which is set as the stopping condition. If the number of data in a child node is greater than or equal to the set threshold value, repeat S52 and S53, continue to select the optimal partitioning features and partitioning points on the child nodes, perform node partitioning, and recursively construct the subtree. If the number of data in a child node is less than the set threshold value, mark it as a leaf node, calculate the average price of all data in the leaf node, and use the average price as the predicted value of the leaf node. This process continues until all nodes meet the stopping condition and the decision tree is constructed. Then, the characteristics of the agricultural products to be predicted are input into the decision tree, and the price prediction of the agricultural products is output.

Images